THESIS TOPICS m

2 INTRODUCING MORPHOLOGY

1. In this text I presuppose that you have already learned at least that part of the International Phonetic Al-

phabet (IPA) that is commonly used for transcribing English. You’ll fi nd an IPA chart at the beginning of this

book, if you need to refresh your memory.

1.1 Introduction

The short answer to the question with which we begin this text is that

morphology is the study of word formation, including the ways new

words are coined in the languages of the world, and the way forms of

words are varied depending on how they’re used in sentences. As a native

speaker of your language you have intuitive knowledge of how to form

new words, and every day you recognize and understand new words that

you’ve never heard before.

Stop and think a minute:

• Suppose that splinch is a verb that means ‘step on broken glass’; what is its past tense?

• Speakers of English use the suffixes -ize ( crystallize ) and -ify ( codify ) to form verbs from nouns. If you had to form a verb that means ‘do

something the way ex-Prime Minister Tony Blair does it’, which suffix

would you use? How about a verb meaning ‘do something the way

ex-President Bill Clinton does it’?

• It’s possible to rewash or reheat something. Is it possible to relove, reexplode , or rewiggle something?

Chances are that you answered the first question with the past tense

splinched (pronounced [splintʃt]), 1 the second with the verbs Blairify and

Clintonize , and that you’re pretty sure that relove, reexplode, and rewiggle

sound rather weird, if not downright unacceptable. Your ability to make

up these new words, and to make judgments about words that you think

could never exist, suggests that you have intuitive knowledge of the prin-

ciples of word formation in your language, even if you can’t articulate

what they are. Native speakers of other languages have similar knowledge

of their languages. This book is about that knowledge, and about how we

as linguists can find out what it is. Throughout this book, you will be look-

ing into how you form and understand new words, and how speakers of

other languages do the same. Many of our examples will come from

English – since you’re reading this book, I assume we have that language

in common – but we’ll also look beyond English to how words are formed

in languages with which you might be familiar, and languages which you

might never have encountered before. You’ll learn not only the nuts and

bolts of word formation – how things are put together in various lan-

guages and what to call those nuts and bolts – but also what this knowl-

edge says about how the human mind is organized.

The beauty of studying morphology is that even as a beginning student

you can look around you and bring new facts to bear on our study. At this

point, you should start keeping track of interesting cases of new words

that you encounter in your life outside this class. Look at the first

Challenge box.

What is morphology? 3

Challenge: your word log

Keep track of every word you hear or see (or produce yourself) that

you think you’ve never heard before. You might encounter words

while listening to the radio, watching TV, surfing the internet, or

reading, or someone you’re talking to might slip one in. Write those

new words down, take note of where and when you heard/read/

produced them, and jot down what you think they mean. What you

write down may or may not be absolutely fresh new words – they just

have to be new to you. We’ll be coming back to these as the course

progresses and putting them under the microscope.

Of course, if the answer to our initial question were as simple as the

task in the box, you might expect this book to end right here. But there is

of course much more to say about what makes up the study of morphol-

ogy. Simple answers frequently lead to further questions, and here’s one

that we need to settle before we go on.

1.2 What’s a word?

Ask anyone what a word is and . . . they’ll look puzzled. In some sense, we

all know what words are – we can list words of various sorts at the drop of

a hat. But ask us to define explicitly what a word is, and the average non-

linguist is flummoxed. One person might say that a word is a stretch of

letters that occurs between blank spaces. But another is bound to point

out that words don’t have to be written for us to know that they’re words.

And in spoken (or signed) language, there are no spaces or pauses to delin-

eate words. Yet we know what they are. Still another person might at this

point try an answer like this: “A word is a small piece of language that

means something,” to which a devil’s advocate might respond, “But what

do you mean by ‘a small piece of language’?” This is the point at which it

becomes necessary to define a few specialized linguistic terms.

Linguists define a morpheme as the smallest unit of language that has

its own meaning. Simple words like giraffe, wiggle , or yellow are mor-

phemes, but so are prefixes like re- and pre- and suffixes like -ize and -er . 2

There’s far more to be said about morphemes – as you’ll see in later chap-

ters of this book – but for now we can use the term morpheme to help us

come up with a more precise and coherent definition of word . Let us now

define a word as one or more morphemes that can stand alone in a lan-

guage. Words that consist of only one morpheme, like the words in (1) , can

be termed simple or simplex words. Words that are made up of more than

one morpheme, like the ones in (2) , are called complex :

2. In Chapter 2 we will give a more formal defi nition of prefi x and suffi x . For now it is enough to know that

they are morphemes that cannot stand on their own, and that prefi xes come before, and suffi xes after, the

root or main part of the word.

4 INTRODUCING MORPHOLOGY

(1) Simplex words

giraffe

fraud

murmur

oops

just

pistachio

(2) Complex words

opposition

intellectual

crystallize

prewash

repressive

blackboard

We now have a first pass at a definition of what a word is, but as we’ll see,

we can be far more precise.

1.3 Words and lexemes, types and tokens

How many words occur in the following sentence?

My friend and I walk to class together, because our classes are in the

same building and we dislike walking alone.

You might have thought of at least two ways of answering this question,

and maybe more. On the one hand, you might have counted every item

individually, in which case your answer would have been 21. On the other

hand, you might have thought about whether you should count the two

instances of and in the sentence as a single word and not as separate

words. You might even have thought about whether to count walk and

walking or class and classes as different words: after all, if you were not a

native speaker of English and you needed to look up what they meant in

the dictionary, you’d just find one entry for each pair of words. So when

you count words, you may count them in a number of ways.

Again, it’s useful to have some special terms for how we count words.

Let’s say that if we are counting every instance in which a word occurs in

a sentence, regardless of whether that word has occurred before or not, we

are counting word tokens . If we count word tokens in the sentence above,

we count 21. If, however, we are counting a word once, no matter how

many times it occurs in a sentence, we are counting word types .

Counting this way, we count 20 types in the sentence above: the two

tokens of the word and count as one type. A still different way of counting

words would be to count what are called lexemes . Lexemes can be

thought of as families of words that differ only in their grammatical end-

ings or grammatical forms; singular and plural forms of a noun ( class,

classes ), present, past, and participle forms of verbs ( walk, walks, walked,

walking ), different forms of a pronoun ( I, me, my, mine ) each represent a

single lexeme. One way of thinking about lexemes is that they are the

What is morphology? 5

basis of dictionary entries; dictionaries typically have a single entry for

each lexeme. So if we are counting lexemes in the sentence above, we

would count class and classes , walk and walking , I and my , and our and we as

single lexemes; the sentence then has 16 lexemes.

1.4 But is it really a word?

In some sense we now know what words are – or at least what word types,

word tokens, and lexemes are. But there’s another way we can ask the

question “What’s a word?” Consider the sort of question you might ask

when playing Scrabble: “Is aalii a word?” Or when you encounter an unfa-

miliar word: “Is bouncebackability a word?” What you’re asking when you

answer questions like these, is really the question “Is xyz a REAL word?”

Our first impulse in answering those questions is to run for our favorite

dictionary; if it’s a real word it ought to be in the dictionary.

But think about this answer for just a bit, and you’ll begin to wonder if

it makes sense. Who determines what goes in the dictionary in the first

place? What if dictionaries differ in whether they list a particular word?

For example, the Official Scrabble Player’s Dictionary lists aalii but not bounce-

backability . The Oxford English Dictionary Online doesn’t list aalii , but it does

list bouncebackability . So which one is right? Further, what about words like

paralpinism , eruptionist , or schlumpadinka that don’t occur in any published

dictionary yet, but can be encountered in the media? According to Word

Spy ( www.wordspy.com ) paralpinism is a sport involving first climbing a

mountain and then using a paraglider to get down; an eruptionist is a per-

son who believes that the world will end in a huge volcanic explosion; and

schlumpadinka is an adjective meaning ‘unkempt’. And what about the

word schlumpadinkahood , which I just made up? Once you know what

schlumpadinka means, you have no trouble understanding my new word. If

it consists of morphemes, has a meaning, and can stand alone, doesn’t it

qualify as a word according to our definition even if it doesn’t appear in

the dictionary?

What all these questions suggest is that we each have a mental lexicon ,

a sort of internalized dictionary that contains an enormous number of

words that we can produce, or at least understand when we hear them.

But we also have a set of word formation rules which allows us to create

new words and understand new words when we encounter them. In the

chapters to follow, we will explore the nature of our mental lexicon in

detail, and think further about the “Is it really a word?” question. In

answering this question we’ll be led to a detailed exploration of the

nature of our mental lexicon and our word formation rules.

1.5 Why do languages have morphology?

As native speakers of a language we use morphology for different reasons.

We will go into both the functions of morphology and means of forming

new words in great depth in the following chapters, but here, we’ll just

give you a taste of what’s to come.

6 INTRODUCING MORPHOLOGY

One reason for having morphology is to form new lexemes from old

ones. We will refer to this as lexeme formation . (Many linguists use the

term word formation in this specific sense, but this usage can be confus-

ing, as all of morphology is sometimes referred to in a larger sense as

‘word formation’.) Lexeme formation can do one of three things. It can

change the part of speech (or category) of a word, for example, turning

verbs into nouns or adjectives, or nouns into adjectives, as you can see in

the examples in (3) :

(3) Category-changing lexeme formation 3

V → N: amuse → amusement

V → A: impress → impressive

N → A: monster → monstrous

Some rules of lexeme formation do not change category, but they do add

substantial new meaning:

(4) Meaning-changing lexeme formation

A → A ‘negative A’ happy → unhappy

N → N ‘place where N lives’ orphan → orphanage

V → V ‘repeat action’ wash → rewash

And some rules of lexeme formation both change category and add sub-

stantial new meaning:

(5) Both category and meaning-changing lexeme formation

V → A ‘able to be Ved’ wash → washable

N → V ‘remove N from’ louse → delouse

Why have rules of lexeme formation? Imagine what it would be like to

have to invent a wholly new word to express every single new concept. For

example, if you wanted to talk about the process or result of amusing

someone, you couldn’t use amusement , but would have to have a term like

zorch instead. And if you wanted to talk about the process or result of

resenting someone, you couldn’t use resentment , but would have to have

something like plitz instead. And so on. As you can see, rules of lexeme

formation allow for a measure of economy in our mental lexicons: we can

recycle parts to come up with new words. It is probably safe to say that all

languages have some ways of forming new lexemes, although, as we’ll see

as this book progresses, those ways might be quite different from the

means we use in English.

On the other hand, we sometimes use morphology even when we don’t

need new lexemes. For example, we saw that each lexeme can have a num-

ber of word forms . The lexeme WALK has forms like walk, walks, walked,

walking that can be used in different grammatical contexts. When we

change the form of a word so that it fits in a particular grammatical con-

text, we are concerned with what linguists call inflection . Inflectional

word formation is word formation that expresses grammatical distinc-

tions like number (singular vs . plural); tense (present vs . past); person

3. The notation V → N means ‘changes a verb to a noun.’

What is morphology? 7

(first, second, or third); and case (subject, object, possessive), among oth-

ers. It does not result in the creation of new lexemes, but merely changes

the grammatical form of lexemes to fit into different grammatical con-

texts (we will look at this in detail in Chapter 6 ).

Interestingly, languages have wildly differing amounts of inflection .

English has relatively little inflection. We create different forms of nouns

according to number ( wombat, wombats ); we mark the possessive form of a

noun with -’s or -s’ ( the wombat’s eyes ). We have different forms of verbs for

present and past and for present and past participles ( sing, sang, singing,

sung ), and we use a suffix - s to mark the third person singular of a verb ( she

sings ).

However, if you’ve studied Latin , Russian , ancient Greek , or even Old

English , you’ll know that these languages have quite a bit more inflec-

tional morphology than English does. Even languages like French and

Spanish have more inflectional forms of verbs than English does.

But some languages have much less inflection than English does.

Mandarin Chinese , for example, has almost none. Rather than marking

plurals by suffixes as English does, or by prefixes as the Bantu language

Swahili does, Chinese does not mark plurals or past tenses with morphol-

ogy at all. This is not to say that a speaker of Mandarin cannot express

whether it is one giraffe, two giraffes, or many giraffes that are under

discussion, or whether the sighting was yesterday or today. It simply

means that to do so, a speaker of Mandarin must use a separate word like

one, two or many or a separate word for past to make the distinction.

(6) Wo jian guo yi zhi chang jing lu.

I see past one CLASSIFIER giraffe 4

(7) Wo jian guo liang zhi chang jing lu .

I see past two CLASSIFIER giraffe

The word chang jing lu ‘giraffe’ has the same form regardless of how many

long-necked beasts are of interest. And the verb ‘to see’ does not change its

form for the past tense; instead, the separate word guo is added to express

this concept. In other words, some concepts that are expressed via inflec-

tion in some languages are expressed by other means (word order, sepa-

rate words) in other languages.

1.6 The organization of this book

In what follows, we’ll return to all the questions we’ve raised here. In

Chapter 2 , we’ll revisit the question of what a word is, by further probing

the differences between our mental lexicon and the dictionary, and look

further into questions of what constitutes a “real” word. We’ll look at the

ways in which word formation goes on around us all the time, and con-

sider how children (and adults) acquire words, and how our mental lexi-

cons are organized so that we can access the words we know and make up

4. We will explain in Chapter 6 what we mean by classifi er . For now it is enough to know that classifi ers are

words that must be used together with numbers in Mandarin.

8 INTRODUCING MORPHOLOGY

new ones. In Chapter 3 , we’ll get down to the work of looking at some of

the most common ways that new lexemes are formed: by adding prefixes

and suffixes, by making up compound words, and by changing the catego-

ry of words without changing the words themselves. In this chapter we’ll

concentrate on how words are structured in terms of both their forms and

their meanings. Many of our examples will be taken from English, but

we’ll also look at how these kinds of word formation work in other lan-

guages. Chapter 4 takes up a related topic, productivity: some processes of

word formation allow us to form many new words freely, but others are

more restricted. In this chapter we’ll look at some of the determinants of

productivity, and how productivity can be measured. Chapter 5 will also

be concerned with lexeme formation, but with kinds of lexeme formation

that are less familiar to speakers of English. We’ll look at forms of affixa-

tion that English does not have (infixation, circumfixation), processes like

reduplication, and templatic morphology. Our focus will be on learning to

analyze data that might on the surface seem to be quite unfamiliar. In

Chapter 6 we will turn to inflection, looking not only at the sorts of inflec-

tion we find in English and other familiar languages, but also at inflec-

tional systems based on different grammatical distinctions than we find

in English, and systems that are far more complex and intricate. Chapter 7

will be devoted to the subject of typology, different ways in which the

morphological systems of the languages of the world can be classified and

compared to one another. We’ll look at some traditional systems of clas-

sification, as well as some that have been proposed more recently, and

assess their pros and cons. Chapters 8 and 9 will explore the relationship

between the field of morphology and the fields of syntax on the one hand

and phonology on the other. Our final chapter will introduce you to some

of the interesting theoretical debates that have arisen in the field of mor-

phology over the last two decades and prepare you to do more advanced

work in morphology.

Exercises 1. Are the following words simple or complex?

a. members b. prioritize c. handsome d. fizzy e. dizzy

Summary Morphology is the study of words and word formation. In this chapter we have considered what a word is and looked at the distinction

between word tokens, word types, and lexemes. We have divided word

formation into derivation – the formation of new lexemes – and

inf lection, the different grammatical word forms that make up

lexemes.

What is morphology? 9

f. grammar g. writer h. rewind i. reject j. alligator If you have difficulty deciding whether particular words are simple or complex, explain why you find them problematic.

2. Do the words in the following pairs belong to the same lexeme or to different lexemes? a. revolve revolution b. revolution revolutions c. revolve dissolve d. go went e. wash rewash

3. In the following sentences, count word tokens, types, and lexemes: a. I say now, just as I said yesterday, that the price of a wombat is high

but the price of a platypus is higher.

tokens ____________ types ____________ lexemes ____________

a. I’ve just replaced my printer with a new one that prints much faster.

tokens ____________ types ____________ lexemes ____________

4. In sentence (3b), what sorts of problems does the word I’ve pose for our definition of ‘word’?

5. What words belong to the same word family or lexeme as sing ?

KEY TERMS word

mental lexicon

lexicography

the Gavagai problem

fast mapping

aphasia

CHAPTER OUTLINE

In this chapter you will learn why we make a basic

distinction between the dictionary and the mental lexicon.

♦ We will look at how linguists study the mental lexicon

and how children acquire words.

♦ We will consider whether complex words are stored in

the mental lexicon, or derived by rules, or both.

♦ And we will look further at how dictionaries have

evolved and how they differ from one another and from

the mental lexicon.

2 C H A P T E R

Words, dictionaries, and the mental lexicon

12 INTRODUCING MORPHOLOGY

2.1 Introduction

In the last chapter, we raised the question “What’s a word?” And we saw

in section 1.2 that this question actually subsumes two more specific ques-

tions. In this chapter we will look more closely at those questions.

On the one hand, when we ask “What’s a word?,” we may be asking

about the fundamental nature of wordhood – as we saw, a far thornier

philosophical question than it would seem at first blush. Native speakers

of a language seem to know intuitively what a ‘word’ is in their language,

even if they have trouble coming up with a definition of ‘word’.

Interestingly, the Oxford American Dictionary seems to bank on this intui-

tive knowledge when it defines a word as “a single distinct meaningful

element of speech or writing, used with others (or sometimes alone) to

form a sentence and typically shown with a space on either side when

written or printed.” We’ve already debunked part of the OAD definition:

languages need not be written, but they still have words, and words don’t

have blank space between them in spoken language. Nevertheless, the

OAD ’s definition works for most people: most dictionary users probably

do not know the word morpheme , which we used in our definition of

word in the last chapter, but the OAD relies on the likelihood that they

will not first think of something like the prefix re- as a single meaningful

element, or something like irniarualiunga which means ‘I am making a

doll’ in Central Alaskan Yup’ik (Mithun 1999: 203), and constitutes not

only a word, but also a whole sentence. In other words, the OAD ’s defini-

tion works because dictionary users already have an intuitive idea of

what a word is!

Morphologists, however, have the luxury of being more precise: we can

define a word as a sequence of one or more morphemes that can stand

alone in a language. But in doing so, we have not exhausted what’s inter-

esting about our question.

Indeed, in Chapter 1 we saw that there is a second way of interpreting

it, one that seems far more concrete at first: we can interpret our question

as meaning “Is xyz a word?” where xyz is a specific morpheme or sequence

of morphemes. Taken this way, our question asks what it means to say

that xyz is a word of English, or Central Alaskan Yup’ik, or some other

language. On the one hand, we are always making up new words, and

when we say them, others understand what we mean. In the last chapter,

I mentioned the words eruptionist and schlumpadinka , neither of which is

in a (conventional) dictionary, at least as of the writing of this chapter, but

both of which have been used (at least by me!). Does this qualify them as

words? And two paragraphs up, I used the word wordhood , which you may

or may not like, but which you certainly understood. This is the version of

the “What’s a word” question that we’ll concentrate on in this chapter. In

doing so we’ll begin to explore the nature of dictionaries, and more

importantly of our native speaker knowledge of words, which we might

term our mental lexicon .

Words, dictionaries, and the mental lexicon 13

2.2 Why not check the dictionary?

When the question “Is xyz really a word?” comes up – whether in casual

conversation, in reading an article in the newspaper, or in playing

Scrabble – people will often look to the dictionary for an answer. Which

dictionary, of course, depends on what’s lying around the house or the

office, or these days, what’s available online. But is this the right way to

answer our question? As morphologists, we need to think about how dic-

tionaries come to be, and how much we credit them with the authority to

decide what’s a word.

There’s a lot to be said about how dictionaries have evolved and how

they are produced today. For a short history of English dictionaries, you

can read section 2.4 of this chapter. But for our immediate purposes, we

can identify a number of reasons why we wouldn’t always want to base

the answer to our question on what we find (or don’t) in a dictionary. Here

are a few such reasons.

2.2.1 Which dictionary? Dictionaries come in all shapes and sizes, for all sorts of intended audi-

ences. Size and audience are determined by individual publishers, and

indeed the finished product is shaped by all sorts of market forces. And

makers of dictionaries – lexicographers – are of course human; what gets

into dictionaries has historically been subject to the individual foibles of

lexicographers, not to mention the mores of society. If you grew up when

I did, it was typical for dictionaries not to have taboo words like fuck , much

less its derivatives fucking, fuck up, fuckable, fuck all, and fucker , all of which

can be found today in the Concise Oxford English Dictonary ; but until the

1970s, dictionaries avoided words that might offend. It is perhaps safe to

say that individual or societal foibles play less of a role in dictionary-

making today, but it’s still a good idea to keep in mind that neither lexi-

cographers nor the dictionaries they create are infallible.

Our first problem with giving final authority for wordhood to the dic-

tionary, then, follows from the very concrete and temporal nature of dic-

tionaries: if you look up a word in a pocket dictionary, or even a standard

college desk dictionary, and it isn’t listed, you might still have the nagging

suspicion that a bigger dictionary or a more specialized dictionary might

list the word. But even if you check the largest available dictionary – say,

for English the Oxford English Dictionary Online – or the most complete tech-

nical dictionary in a particular field, can you be sure that a word that’s

not listed isn’t a word? Maybe it’s too new a word to have gotten into the

dictionary yet.

2.2.2 Nonces, mistakes, and mountweazels Further, sometimes we find items in dictionaries that we might hesitate

to call words – even if they do occur in the dictionary. Among these items

are words that are labeled as ‘ nonce ’, meaning that they’ve been found

14 INTRODUCING MORPHOLOGY

just once, often in the writing of someone important, but that neverthe-

less don’t seem to occur anywhere else. The OED Online , for example, lists

as a nonce the word agreemony , which they define as ‘agreeableness’, and

illustrate with a single quotation from the seventeenth-century writer

Aphra Behn. Was this ever really a word? Indeed, the OED even lists some

words that occur only once, and further, in contexts which don’t illumi-

nate their meaning; for example, we can find the word umbershoot used by

James Joyce in Ulysses , about which the OED maddeningly says only “mean-

ing obscure”! Words or not?

Very extensive dictionaries like the OED sometimes also contain words

that they identify as mistakes. For example, we can find an entry for the

word ambassady , which occurs in a single quotation from 1693 and is,

according to the OED , perhaps a mistake, where the author might have

meant the word ambassade “the mission or function of an ambassador.” It

occurs in the dictionary, but is it really a word?

And finally, there are what have come to be called ‘mountweazels’. A

mountweazel is a phony word that is inserted into a dictionary so that

its makers can identify lexicographic piracy. You can find a fuller

explanation of this tradition in section 2.4 , but the short version is

this: lexicographers sometimes make up an entry and include it so that

they can tell if another lexicographer is using their dictionary as a

source without attribution (which is plagiarism, of course). Surely we

wouldn’t want to count such impostors as real words, but they’re in

the dictionary!

2.2.3 And the problem of complex words We will learn much more about this in the chapters to come, but perhaps

the worst problem for us with the idea of giving the dictionary the

authority to determine whether xyz is a word is that dictionaries don’t

need to include every word. Every language has ways of forming new

words that are so active and transparent that putting all the words

formed that way into the dictionary would be a waste of space. For exam-

ple, speakers of English know that any verb at all can have a present

progressive form made with the suffix -ing . As soon as I make up a new

verb, say zax , we know that the present progressive verb form is zaxing . So

although a dictionary might eventually have to include the verb zax , it

might never list zaxing as a word. But of course zaxing should be consid-

ered a word. Similarly, just about any adjective in English can be made

into a noun by adding the suffix -ness . For example, the Concise Oxford

English Dictionary contains the adjective bovine , but not the noun bovine-

ness . Nevertheless, I’d have no problem if I saw the word bovineness written

somewhere, and would never think to look it up in the dictionary. The

dictionary doesn’t have the word precisely because we’d never need to

look it up.

The conclusion that we are inexorably led to is that we cannot rely

on dictionaries to answer the question “Is xyz a word?” On the one

hand, dictionaries don’t list all the words of any language. They can’t

list all derivatives with living prefixes and suffixes, or all technical,

Words, dictionaries, and the mental lexicon 15

1. A transitive verb is one that has both a subject and an object. An intransitive verb has only a subject.

scientific, regional, or slang words. And on the other hand, they some-

times include words used only once whose meanings are completely

unknown. They occasionally even include purposely made-up words to

guard their own copyrights. For the most part, dictionaries do not fix

or codify the words of a language, but rather ref lect the words that

native speakers use. Those words are encoded in what we will call the

mental lexicon , the sum total of word knowledge that native speakers

carry around in their heads. So to answer our question, we must look

more closely at what is in that mental lexicon.

2.3 The mental lexicon

By the mental lexicon I mean the sum total of everything an individual

speaker knows about the words of her language. This knowledge includes

information about pronunciation, category (part of speech), and mean-

ing, of course, but also information about syntactic properties (e.g.,

whether a verb is transitive or intransitive), 1 level of formality, and what

lexicographers call ‘range of application’, that is, the specific conditions

under which we might use the word. For example, I know that the word

verandah is a noun, pronounced (in my American English) [və.�ɹæn.də],

that it refers to a type of porch, and that I’d only use it in reference to the

sort of porch one finds in the southern part of the US or perhaps in some

exotic tropical country. Unless I was being ironic, I probably would not

call my own back porch ‘the verandah’. I also know that barf is a verb

that’s pronounced [bɑɹf], that it means ‘vomit’, that it is intransitive

(unless used with a particle like up ) and that it is used only colloquially

(I wouldn’t use it if I were describing the symptoms of a stomach flu to the

doctor).

It is quite likely that in our mental lexicons we have entries that are

only partial. We may know the pronunciation of a word, but not its mean-

ing (e.g., I know how to pronounce amortize , but I’m not sure what it

means). Or the opposite: for example, I know what the word hegemony

means, but I don’t know if it’s pronounced with the stress on the first or

second syllable. We may also have only partial knowledge of the meaning

of a word. I know, for example, that a distributor is part of a car and that if

you have to replace it, it’s a relatively expensive job, but I don’t know what

a distributor looks like or what it does.

Each person’s mental lexicon is sure to contain things that are different

from other people’s mental lexicons. One person may know lots of words

for types of birds or flowers, another might know all the specialized

vocabulary of sailing, and so on. Auto mechanics surely know more details

of the meaning of the word distributor than I do. But our individual mental

lexicons overlap enough that we speak the same language. In this section

we will look in more detail at the contents of our mental lexicons, both

what is stored and what is created by rules of word formation, and how

our mental lexicons are organized.

16 INTRODUCING MORPHOLOGY

2.3.1 How many words? Psycholinguists estimate that the average English-speaking six-year-old

knows 10,000 words, and the average high-school graduate around 60,000

words. Paul Bloom describes how this estimate can be made ( 2000 : 5):

Words are taken from a large unabridged dictionary, including only

those words whose meanings cannot be guessed using principles of

morphology or analogy . . . Since it would take too long to test people on

hundreds of thousands of words, a random sample is taken. The

proportion of the sample that people know is used to generate an

estimate of their overall vocabulary size, under the assumption that the

size of the dictionary is a reasonable estimate of the size of the language

as a whole. For example, if you use a dictionary with 500,000 words, and

test people on a 500-word sample, you would determine the number of

English words they know by taking the number that they got correct

from this sample and multiplying by 1000.

Children generally begin to produce their first words around the age of

one. Bloom calculates that between the ages of one and eighteen we

would have to learn approximately ten words every day to have a vocabu-

lary of 60,000 words. It’s worth pointing out, I think, that this figure just

takes into account the words that we have stored (fully or partially) in our

mental lexicon, and not the words – perhaps an infinite number of them –

that we can create by using rules of word formation. We will return

shortly to our knowledge of word formation rules and its relation to our

mental lexicon. First, however, we will look more closely at how we

acquire our mental lexicon.

2.3.2 The acquisition of lexical knowledge Psycholinguists have devised experiments to try to learn how children and

adults are able to acquire words so easily. You might think that the learn-

ing of new words is a simple matter of association: someone points at

something and says “flurge” and you learn that that something is called

a f lurge . This may be the way that we learn some words, but surely not the

way we learn the majority of words in our mental lexicons. For one thing,

not everything for which we have a word can be pointed at.

And even if someone points and says a word, it is often not clear from

the context what exactly is being pointed out. Psycholinguists sometimes

call this the Gavagai problem , following a scenario first discussed by the

philosopher W.O. Quine . To summarize:

Picture yourself on a safari with a guide who does not speak English.

All of a sudden, a large brown rabbit runs across a field some distance

from you. The guide points and says “gavagai!” What does he mean?

One possibility is, of course, that he’s giving you his word for ‘rabbit’. But

why couldn’t he be saying something like “There goes a rabbit running

across the field”? or perhaps “a brown one,” or “Watch out!,” or even

“Those are really tasty!”? How do you know?

In other words, there may be so much going on in our immediate envi-

ronment that an act of pointing while saying a word, phrase, or sentence

Words, dictionaries, and the mental lexicon 17

will not determine clearly what the speaker intends his utterance to

refer to.

Besides, we are rarely in a situation in which someone is actively

instructing us about the meanings of words; although parents may point

to things in a picture book and name them for a child, or school children

may be asked to memorize a list of vocabulary words, we learn most words

without explicit instruction and seemingly with very little exposure.

Although we do not know nearly enough about this subject, there are

several things that we do know about how word learning occurs.

First, it is believed that both children and adults are able to do what the

psycholinguist Susan Carey has called fast mapping (Carey 1978 ). Fast

mapping is the ability to pick up new words on the basis of a few random

exposures to them. In one experiment, Carey showed that children who

were casually exposed to a new color name chromium during an unrelated

activity (following instructions to pick up trays of various colors) were able

to absorb the word and recall it even six weeks later. Experiments have

shown that adults exhibit this fast mapping ability as well; while the abil-

ity to learn linguistic rules (say, of syntax or phonology) is thought to

decline after puberty, the ability to learn new words remains robust.

Challenge

Here’s an experiment you can try. Collect five or six objects. All but

one of your objects should be familiar items (a bunch of keys, a mug,

a pencil, etc.). One object, however, should be something odd and not

familiar to many people. Put all your objects on a tray, and ask your

subject (anyone outside your class will do) to point out the zorch .

Observe what you subject does. Now take away the unfamiliar object,

leaving only the familiar objects, and ask a different subject to point

out the plitz . Again, observe closely what the subject does.

Psycholinguists have proposed a number of other strategies that both

children and adults seem to use in learning new words. 2 One might be

called the Lexical Contrast Principle . For example, in an experiment

similar to yours, children were asked to point to the zorch (or some other

made-up word), and what they invariably did was to point out the unfa-

miliar object. According to the Lexical Contrast Principle, the language

learner will always assume that a new word refers to something that does

not already have a name.

A second word learning strategy might be called the Whole Object

Principle . In the experimental condition described above, when subjects

are presented with the word zorch and an unnamed object, they will

assume the whole unnamed object to be a zorch . They will not assume that

zorch refers to a part of the object, to its color or shape, or to a superordi-

nate category of objects to which it might belong.

2. See Bloom ( 2000 ) for an extensive discussion of this subject.

18 INTRODUCING MORPHOLOGY

A related strategy might be dubbed the Mutual Exclusivity Principle . In

the second experiment above, there are only familiar objects for which

subjects already have names. When asked to point out the plitz, experi-

mental subjects typically do one of two things: they might first look

around the room for something else that might be called a plitz , or they

might assume that the word plitz refers to a part of one of the familiar

objects or a special type of one of them. Subjects, in other words, will

assume that if an object already has a word for it, the word plitz cannot be

synonymous with those words.

These experiments are of course not just hypothetical. Paul Bloom ,

Susan Carey , and many other psycholinguists have conducted them

both with children of various ages and with adults, and have obtained

the results described above. What is perhaps most astonishing about

their results is that their experimental subjects often remember the

words they’ve been exposed to when they are retested weeks after the

original experiment. But maybe we should not be surprised by this:

how otherwise could we have learned 60,000 words by the time we’re

eighteen?

Children not only learn simple words, but – as we’ll see in the chapters

to come – they learn the rules that allow us to create and understand new

complex words. English-speaking children acquire some rules of word

formation quite early on; for example, Berman ( 2009 ) notes that English-

speaking children are known to produce compound words (roughly, words

made up of two or more independent words) as early as eighteen to

twenty-four months, and three-year-olds acquiring English can come up

with novel compound words like wagon boy when asked what they would

call a boy who pulls wagons. The same appears to be true of children

acquiring other Germanic languages like German , Dutch , and Swedish ,

but less so for languages like Hebrew and French , where compounding is

not used nearly as much in adult language as in the language of adults

speaking Germanic languages.

Children also use prefixed and suffixed words early on, but as Clark

( 2014 ) points out, we only know when they have learned the rules for

forming prefixed and suffixed words when we see that they can

produce new words or say what new complex words mean; in other

words, a child may know the word baker as a whole chunk without

knowing that it’s made up of two pieces, but when a child can tell us

that someone who zaxes is a zaxer they show us that they know what

the suffix - er means and how it can be put together with a verb to form

a new noun. Clark ( 2014 ) indicates that some suffixes like - er and pre-

fixes like un - are acquired very early by English-speaking children,

emerging at around three years of age but becoming better established

by five. According to Clark, children acquiring French start using pre-

fixes and suffixes as early as two years old. It seems that because

French relies on prefixes and suffixes more than compounding, these

elements of word formation begin to emerge earlier in French-speaking

children .

Words, dictionaries, and the mental lexicon 19

Novel uses of complex words in child language

(Clark 2014 : 431–2)

D: (2:5.26, as he reached across the kitchen counter): I’m a big reacher.

D: (3:5.27): did you know Gabriel is a stealer? sometimes he steals Judy’s

things.

D: (3:6.25 . . .): These are my pantsies. D’you know what I call a cat? . . . A

cattie. D’you know what I call a bear? A bearie! D’you know what I call a dog?

A doggie!

2.3.3 The organization of the mental lexicon: storage versus rules Although linguists like to describe our knowledge of words as a mental

lexicon, we know that the mental lexicon is not organized alphabetically

like a dictionary. Rather, it is a complex web composed of stored items

(morphemes, words, idiomatic phrases) that may be related to each other

by the sounds that form them and by their meanings. Along with these

stored items we also have rules that allow us to combine morphemes in

different ways, a subject into which we will delve in the chapters to come.

The question that we must first ask is what exactly our mental lexicons

contain and how that content is organized. Certainly we must have sim-

plex words stored in our lexicons, and our ability to form new complex

words shows that we have rules stored, but what about complex words

that we have already encountered? Do we store them as unanalyzed

wholes? Or do we use our word formation rules to recreate a familiar

inflected or derived word every time we use it or break it down every time

we hear it? How could we even begin to figure this out?

This is the task of psycholinguists who study morphology, and there are

many ways of trying to tease out how we store, form, access, and process

both simple and complex words. Some of these techniques involve what

are called lexical decision experiments in which a subject has to decide

whether a word they are exposed to is an existing word (like packer ), a

potential word (like zaxer ), or a nonsense word (like plurgle ). The time that

it takes the subject to decide, called the reaction time or response latency ,

gives us a clue as to whether the subject treats a complex word as a whole

or breaks it down into its component parts. Psycholinguists also use a tech-

nique called eye tracking to study the way subjects process complex words.

In an eye-tracking study, experimenters record where subjects’ eyes land

when they look at a complex word and whether or not their eyes backtrack

once they have moved on to a new word. These eye motions also give us a

clue as to how subjects process and access complex words. Finally, in recent

years psycholinguists have begun to study the organization of the mental

lexicon with the use of different kinds of imaging techniques like

electroencephalography (EEG) , which measures electrical activity on the

scalp, magnetoencephalography (MEG), which measures magnetic activity

20 INTRODUCING MORPHOLOGY

in the brain, and fMRI, which stands for functional magnetic resonance

imaging , a way of mapping activity in the brain over the course of time

(Baayen 2014 ).

What psycholinguists have begun to show using all these techniques is

that the picture is complicated and experimental results frequently do

not agree. But a consensus is building that the organization of the mental

lexicon involves an intricate combination of storage of both simple and

complex words and creation of complex words via rule. It appears that

storage versus rules is not necessarily an either/or matter. Apparently

whether we access a complex word like packer whole or create it anew

using our rules depends on many factors including our familiarity with

the complex word (i.e., the frequency with which we have encountered it

before), the frequency of the root or base on which the complex word is

formed (in this case, pack ), the number and frequency of the family of

words containing the same root or the same prefix or suffix (e.g., packing ,

package , pack rat , and so on), and a variety of other factors. Psycholinguists

have by no means come to an agreement on how our mental lexicons are

arranged, but this continues to be an active area of research in morphol-

ogy. We will illustrate some of the complexities of the debate by looking

at the past tense of verbs in English , a subject on which there has been

ongoing debate. Let’s first take a quick look at how we form past tense

verbs in English.

At this point, if I asked you how to form the past tense of a verb in

English, you would probably say that you usually add an -ed . And then you

might point out that there are a number of verbs that have irregular past

tenses like sing�sang, tell�told, win�won , f ly�f lew, and the like. While it is

true that in writing we add an -ed to form the past tense of a verb, in terms

of spoken speech, the situation is a bit more complicated. Consider the

next Challenge.

Challenge

Consider how you pronounce the past tenses of these verbs:

1. rap, tack, laugh, sheath, pass, lurch

2. pat, prod

3. rob, rove, bathe, buzz, rouge, judge, warm, warn, bang, roar, rule,

tango

Transcribe the past tenses of these words in the International

Phonetic Alphabet and observe how they differ.

You pronounce the past tenses of the first set of words in the Challenge

box with a [t] sound, in the second with a sound like [əd], and the third

with a [d] sound.

We do not choose the pronunciation of the past tense at random.

Rather, the choice of which of the three endings to use depends on the

Words, dictionaries, and the mental lexicon 21

final sound of the verb. Those words that are pronounced with final [t] or

[d] sounds – those in the second list – get the [əd] pronunciation. The

words that end in voiceless (with the exception of [t]) sounds get the [t]

pronunciation. And all the rest get the [d] pronunciation. As for irregular

forms like sang and f lew , we can assume for now that English speakers

simply learn them as exceptions.

We know that speakers of English have an unconscious knowledge of

the regular past tense rule because we can automatically create the past

tense of novel verbs. For example, if I coin a verb blick , you know that the

past tense morpheme is pronounced [t]. Similarly, the novel verb f lurd will

have the past tense [əd], and the verb zove will be made past tense with [d].

We can even form the past tense of verbs that contain final sounds that do

not occur at all in English, and when we do, we still follow the rule. For

example, if we imagine that there are many composers imitating the style

of Johann Sebastian Bach, and we coin the verb to bach to denote the

action of imitating Bach, we will automatically form the past tense with

the past tense variant pronounced [t], because the final sound of Bach is

[x], a voiceless velar fricative. The important point here is that when we

hear this sound at the end of a verb we know (unconsciously) that it’s

voiceless, and apply the past tense rule to it in the usual way.

Now that we know something about the English past tense, we can

return to the question of how the mental lexicon is organized. One very

influential hypothesis that has been proposed by psycholinguist Steven

Pinker ( 1999 ) is the ‘words and rules’ hypothesis , which says that speakers

of English use the past tense rule every time they create the past tense of

a regular verb, but simply store the past tense forms of irregular verbs

whole and unanalyzed in their mental lexicons. There is some evidence

from studies of people with aphasia – that is, people whose language fac-

ulty has been impaired due to stroke or other brain trauma – that sup-

ports the ‘words and rules’ hypothesis, and also evidence from studies of

people with various genetic disorders that affect language. But there is

also other evidence from the sorts of experiments and imaging that calls

this simple picture into question.

2.3.4 Evidence from aphasia We will look first at the evidence from the study of aphasics. Studies by

Badecker and Caramazza ( 1999 ) and Ullman et al. ( 2005 ) with different

kinds of aphasic subjects offer some support for the words and rules

hypothesis; these studies suggest that irregular past tenses are stored in

one area of the brain and rules creating regular past tenses are located

elsewhere.

Some aphasics display a form of aphasia called agrammatism. Also

called non-fluent aphasics, these individuals speak haltingly and with

effort and have difficulty in producing or processing function words in

sentences. However, they can still produce and understand content words.

Experiments seem to show that when asked to read or produce the past

tenses of different kinds of verbs, non-fluent aphasics have more trouble

with regular past tenses than with irregular ones. These subjects tend to

22 INTRODUCING MORPHOLOGY

have damage to frontal areas of the left hemispheres of the brain that

have been associated with rule-based linguistic activity. In contrast, so-

called fluent aphasics produce syntactically complex sentences but have

trouble accessing content words. When asked to read or produce the past

tenses of regular and irregular verbs, they perform more accurately for

regulars than for irregulars. And their linguistic deficits seem to be associ-

ated with damage to the posterior part of the left hemisphere. Linguists

like Ullman et al. ( 2005 ) have suggested that this evidence supports the

idea that morphological rules are associated with a different part of the

brain than items that are stored. For non-fluent aphasics, the rule is

unavailable, presumably because the part of the brain has been damaged

that apparently allows us to apply morphological rules, but the irregular

forms are still accessible from an undamaged part of the brain. For fluent

aphasics, the irregular forms have been lost because the part of the brain

that apparently allows access to stored forms has been damaged, but the

regular rule is still intact. This finding is suggestive, but as McWhinney

( 2005 ) rightly points out, it must be taken as tentative for two reasons:

first, because the number of subjects studied is very small; and second,

because currently available studies do not absolutely rule out the possibil-

ity that frequent regular past tenses can be stored for some non-fluent

aphasics.

2.3.5 Evidence from genetic disorders Studies of subjects with genetic disorders yield a similar picture.

Psycholinguists have studied two disorders in particular – Specific

Language Impairment (SLI) and Williams Syndrome – that affect lan-

guage in different ways. Individuals with SLI are generally of normal intel-

ligence and have no hearing impairment. But they are slow to develop and

understand language, and their speech is characterized by the omission

of various inflectional morphemes. Individuals with Williams Syndrome

have a genetic disorder linked to various heart problems, elevated levels

of calcium in their blood, and a characteristic appearance (short stature,

an upturned nose, a long neck, among other things). Their language and

social skills are in the normal range, but in other respects such as motor

control and spatial perception they display mild or moderate developmen-

tal delay.

What is significant for our purposes is that these disorders are also

thought to provide evidence for the organization of our mental lexicon.

Individuals with SLI find it difficult to create the past tenses of novel

verbs, and often fail to inflect unfamiliar regular verbs correctly; they

have less difficulty with irregular verbs, though. In spontaneous speech,

they may leave the regular past tense off verbs (Redmond and Rice 2001 ;

Marshall and van der Lely 2012 ). In contrast, individuals with Williams

Syndrome speak fluently and produce sentences with correct regular past

tenses, but have more trouble with irregular ones; indeed they seem to

use regular past tense marking even where control subjects or individuals

with SLI would not, for example, overgeneralizing the regular -ed ending

on irregular verbs (e.g., falled ) (Clahsen, Ring, and Temple 2004 ). Assuming

Words, dictionaries, and the mental lexicon 23

that the genetic anomalies associated with these disorders affect different

parts of the brain, we can explain this pattern of behavior.

2.3.6 Evidence from imaging studies Early imaging studies of normal subjects, such as those done with PET

(positron emission tomography) scans seemed also to yield results that

support the ‘words and rules’ hypothesis. PET scans measure the level of

blood flow to different parts of the brain, which in turn shows us areas of

activation in those parts. Jaeger et al. ( 1996 ) reported that there are parts

of the brain that are activated when subjects are asked to read regularly

inflected past tenses that are distinct from those activated in reading or

producing irregular past tenses. Several other studies using a variety of

imaging techniques seemed to come to the same conclusion. But these

results are not uncontroversial: in a more recent study Desai et al. ( 2006 )

point out that imaging studies do not agree on which parts of the brain are

associated with regulars and irregulars, which is somewhat troubling. And

in a study using MEG imaging Stockall and Marantz ( 2006 ) argue against

the ‘words and rules’ hypothesis, suggesting that subjects do not access

irregular past tenses as unanalyzed wholes in their mental lexicons, but

rather activate both the irregular form and the root of the verb – suggest-

ing that some analysis is going on early in the process of recognizing the

past tense. It seems safe to say that this is an extremely important and

interesting area of study, but that we are not able to draw firm conclu-

sions yet about the organization of the mental lexicon.

2.3.7 Reprise: is it really a word? We have spent some time in this chapter contrasting the dictionary with

the mental lexicon in order to understand the question “Is xyz really a

word?” We are now in a position to understand this question better, at

least from the point of view of morphologists. Most morphologists would

say that xyz is a word if it can be formed by the rules of word formation in

a particular language. So words like wordhood or re-reprise that you might

never have seen before you read this chapter really are words, even though

you won’t find them in any dictionary. They are words because they follow

the rules of English word formation. It is the rules of word formation that

we know that most distinguish our mental lexicon from the dictionary.

The dictionary does not need to list all the words that we know or that we

could create, because once we know word formation rules we can produce

and understand potentially infinite numbers of new words from the mor-

phemes available to us. The remainder of this book will be an attempt to

work out in some detail what those rules are.

2.4 More about dictionaries

In section 2.2 we considered all the reasons why morphologists don’t look

upon dictionaries as the ultimate arbiters of ‘wordhood’ in English,

or indeed in any language. You may not need more convincing of this

issue, but for those of you who have a fondness for dictionaries (most

24 INTRODUCING MORPHOLOGY

morphologists do!), it’s worth knowing something about how dictionaries

have developed. I’ll again concentrate on English here, as our common

language, but the history of dictionary-making for other languages can be

equally fascinating.

Let’s start with a thought experiment. Look at the next Challenge.

Challenge

Suppose that a great catastrophe has occurred and every single

written or online dictionary has disappeared from the face of the

earth. You and your classmates have survived the catastrophe

(perhaps in a hidden concrete tunnel beneath the building in which

you are now sitting), and have been delegated the task by other

survivors of creating the first post-catastrophe dictionary of your

language.

How would you start?

Your first instinct would probably be to make a list of words that you

would need to define. Assuming that there were no surviving books to use

as dictionary-fodder, a good way to begin would be by thinking of catego-

ries, and listing everything you could in each one. After you’ve listed all the

animals, plants, and types of furniture you could think of, you’d come up

with a list of hairstyles ( crewcut, bob, beehive, bun, buzz cut, duck’s ass,

cornrows, mullet . . . ) and condiments ( ketchup, soy sauce, mustard, horseradish,

wasabi, sambal oelek . . . ), and so on, and eventually you’d come to articles ( a,

the, this, that . . . ), prepositions ( in, on, above, during, for . . . ) and the other

small words that form the grammatical glue that holds sentences together.

But along the way, you’d discover a number of problems. First, you’d

have a suspicion that you’d be forgetting things (what, for example, was

the name for that women’s hairstyle that was the rage in the seventies?).

Second, you and your classmates would get into constant arguments over

this word or that: is it worth putting the word mullet in the dictionary as

the name of a hairstyle? Wasn’t that slang? Does slang go in the diction-

ary? What IS slang, anyway? Is it too vulgar to put duck’s ass in the diction-

ary as a name of a 1950s hairstyle? What about really raunchy words? Is

sambal oelek a word for a condiment in English, or is it just something

we’ve borrowed from another language (what other language, though?)?

What this thought experiment does is to put you in the shoes of a lexi-

cographer. In reality, it’s been centuries since lexicographers have had to

start from scratch in creating a dictionary – and perhaps they’ve never

really done so. As the lexicographer Sidney Landau has said about the

tradition of dictionary-making in English (2001: 43), “The history of

English lexicography usually consists of a recital of successive and often

successful acts of piracy.” For years and years, each succeeding dictionary-

maker has consulted already existing dictionaries to come up with a base

list of words, often adding new ones and sometimes deleting words for

Words, dictionaries, and the mental lexicon 25

various reasons. But at least at first, lexicographers did have to decide

one by one on each of the English words to include. Of course, there

were manuscripts and books available to suggest words that needed to

be included, and in fact, the earliest English lexicographers did rely on

the words they found in books as the material from which they built their

dictionaries.

2.4.1 Early dictionaries It was not until the early seventeenth century that anything we would

recognize as a monolingual dictionary could be found for the English

language. Dictionaries or glossaries for translating Latin to English date

from a century or so earlier, and in the sixteenth century lists of so-called

hard words could be found for English, explaining words which largely

had been adapted from Latin . The first real dictionary of English is gener-

ally acknowledged to be Robert Cawdrey’s (1604) A Table Alphabeticall of

Hard Words . The tradition of lexicographical piracy goes at least as far back

as Cawdrey, who is said to have used an available Latin–English dictionary

of his day to help come up with the words to define. The first dictionaries

going beyond the tradition of defining only ‘hard’ words to include ordi-

nary, everyday words began to appear in the early eighteenth century;

Landau ( 2001 : 52) cites John Kersey’s (1702) A New English Dictionary as the

earliest of these, followed by Nathaniel Bailey’s (1721) An Universal

Etymological English Dictionary .

2.4.2 Johnson’s dictionary A more significant milestone in the history of English lexicography for

our purposes was Samuel Johnson’s Dictionary of the English Language , pub-

lished in 1755. It contains more than 42,000 entries – even then, only a

small fraction of English vocabulary – and took seven years to write, an

astonishing feat for a single individual. Johnson’s dictionary was not only

the most comprehensive English dictionary of his time, but it was also

among the first dictionaries to include illustrative quotations on a large

scale.

What is most interesting for our purposes, though, are the idiosyncra-

sies of Johnson’s dictionary: what he included, what he left out, and how

he defined various words in odd ways. Henry Hitchings ( 2005 : 110) notes

that:

dictionaries are fraught with submerged ideas, narratives and histories.

Johnson’s is no exception. It offers no overarching system of knowledge,

but it is a literary anthology, a compendium of quotable nuggets, and a

mine of information – some trivial, some considerable – on subjects as

diverse as heraldry and hunting, rhetoric and pharmacy, oracles and

literary style, the zodiac and magic, law and mathematics, ignorance

and politics, the art of conversation and the benefits of reading.

Johnson’s dictionary, in other words, contains a lot about Johnson him-

self – both his interests and his prejudices. It was quite a comprehensive

dictionary in its time. But Hitchings notes that Johnson still left out

Some

Johnsonian

definitions:

urim: Urim and

thummim were

something in

Aaron’s breastplate;

but what, criticks

and commentators

are by no means

agreed.

trolmydames: [Of

this word, I know

not the meaning.]

worm (v.): To

deprive a dog of

something, nobody

knows what, under

his tongue, which is

said to prevent

him, nobody knows

why, from running

mad.

network: Anything

reticulated or decus-

sated, at equal dis-

tances, with inter-

stices between the

intersections.

pastern: The knee of

an horse.

26 INTRODUCING MORPHOLOGY

entries for such words as ultimatum, irritable, zinc, engineering, athlete, and

annulment , even though he actually used some of those words in his defi-

nitions. On the other hand, he included such words as ariolation, clancular,

deuteroscopy, and incompossiblity , which even the nineteenth-century

American lexicographer Noah Webster considered dubious. And it has

often been pointed out that some of Johnson’s definitions were odd,

unhelpful, and occasionally downright wrong.

For example, the word urim is used by Milton, and therefore Johnson

judged it important enough to be included even though he was unable to

discern the meaning of the word from its literary context. Similarly, trol-

mydames is used in Shakespeare, and therefore it merited inclusion for

Johnson – although, again, he had no idea what it meant. And what

can we say about the definitions for network and worm ? If you don’t

already know what they mean, you won’t be enlightened by Johnson’s

definitions!

As Hitchings implies in the passage quoted above, we can learn a lot

about Johnson’s interests from his dictionary. For example, we can tell

from Johnson’s entry for pastern that he had no particular knowledge of

or interest in horses: he defines the pastern as the knee of a horse. People

who are interested in horses know that a pastern is part of a horse’s foot.

Similarly, as Hitchings points out, Johnson apparently had no interest in

music. His definitions for a number of stringed instruments ( viola, lute,

guitar ) are precisely the same: “a stringed instrument.” Furthermore, the

definition of violin suffers from the cardinal lexicographical sin of circu-

larity: the entry for violin sends one to the entry for fiddle , which in turn

sends one back to violin .

These examples are not intended to imply that Johnson’s dictionary was

incompetent – far from it, it was an amazing achievement for one man

working alone for seven years. Much of it still holds up to twenty-first

century scrutiny. For every entry that is obscure, weird, or unhelpful,

there are a hundred that are brilliant and insightful. I devote this much

attention to its deficiencies merely to point out that dictionaries are fal-

lible, and often reflect the foibles of their makers.

2.4.3 Webster’s dictionary Johnson’s dictionary was followed in 1828 by Noah Webster’s dictionary –

billed as the first American dictionary. Webster’s agenda in writing his

dictionary was at least partly political; through the dictionary he sought

to establish American English as a national language. His dictionary

included not only new words but also new meanings that had developed

for old words in the context of American life, for example, words relevant

to the newly minted form of democracy, such as congress and senate .

Webster is also credited with promoting the spelling differences which

even today distinguish American from British English – color instead of

colour , center instead of centre , tire instead of tyre , and so on.

Webster was not particularly skilled at etymology (the study of where

words come from); Baugh and Cable (2013: 356) suggest that his sense of

nationalism caused him to ignore advances in historical and comparative

Words, dictionaries, and the mental lexicon 27

linguistics that were taking place in Europe at that time. However, his

definitions are excellent. Not surprisingly, though, some definitions in

Webster’s dictionary are pirated directly from Johnson. Note, however,

that not all of Webster’s contemporaries shared his desire to distinguish

American English from British English. Joseph Worcester , for example,

published his own Comprehensive Pronouncing and Explanatory Dictionary in

1830, in which he took a far more conservative approach to Americanisms

and spelling.

2.4.4 The Oxford English Dictionary By the mid-nineteenth century, members of the English Philological

Society had come to feel that Johnson’s dictionary was inadequate. As we

saw above, Johnson had missed many words, and even if he had not, over

the course of a century many new words are added to a language and

many old words come to be used in new ways. After much deliberation

and a number of false starts, the Philological Society chose James Murray ,

a Scottish schoolmaster, to edit the New English Dictionary . Oxford University

Press contracted to publish it, and by 1895 it had come to be known as the

Oxford English Dictionary . Murray began work on the dictionary in 1879,

hoping to finish it within ten years. But it would be almost fifty years

before the first edition of the dictionary was finished, during which time

three more editors were added, and Murray himself died.

The OED took so long to compile because the goals of its originators

were so ambitious. Murray and his colleagues sought to create a diction-

ary that would not only give current meanings of words, but also trace

those words back as far into the history of English as they could, taking

note of all the spelling variants and meaning changes along the way.

Following Johnson’s dictionary, all senses of words would be illustrated

with quotations from literary works. Words that were already archaic or

obsolete by the late nineteenth century would still be included, as long as

they had not died out before 1250 ce . The dictionary was to be comprehen-

sive in both breadth and depth, a task which turned out to be far more

challenging than anyone in 1879 could have anticipated. The first edition

of the OED ran to ten large volumes and contained almost a quarter of a

million main entries. By the time the last volume was finished, the early

volumes were already obsolete; one supplement was added in 1933, and a

second one in 1972. A second edition of twenty volumes was issued in

1989, incorporating all of the supplements into the original volume.

Today, work continues on the third edition, with segments issued online

on a quarterly basis, as they are finished. Since the first edition, the OED

has grown to include more than half a million entries; in its online form,

size and space are no longer as much of a concern as they once were.

James Murray was well aware both of the weight his lexicographical deci-

sions carried and of his potential fallibility in making those decisions –

after all, most people do look to the dictionary to determine whether xyz

really is a word. Perhaps Murray put it best when he noted in the

Introduction to the first edition that:

28 INTRODUCING MORPHOLOGY

The Vocabulary of a widely-diffused and highly-cultivated living

language is not a fixed quantity circumscribed by definite limits.

That vast aggregate of words and phrases which constitutes the

Vocabulary of English-speaking men presents, to the mind that

endeavours to grasp it as a definite whole, the aspect of one of those

nebulous masses familiar to the astronomer, in which a clear and

unmistakable nucleus shades off on all sides, through zones of

decreasing brightness to a dim marginal film that seems to end

nowhere, but to lose itself imperceptibly in the surrounding

darkness.

In other words, it’s impossible to pin down the vocabulary of English

(and we might add, any other language). Murray illustrates his point

with a diagram ( Figure 2.1 ), reproduced from the Introduction to the

first edition of the dictionary. His idea is that there is a core of words

whose place in the dictionary nobody would dispute, encompassing

what he called “common,” “literary,” and “colloquial” words. Common

words are words that occur in all registers of English, like mother, dog,

walk, apologetic, wiggle, if, and, to, in, that , and so on. Literary words are

words that we might recognize when we read, but would not necessar-

ily use in daily conversation, words, for example, like omnipotent, not-

withstanding, heretical, hegemony, and ambulatory. And also among the

core words would be colloquial words, ones that we use frequently in

spoken language, but far less frequently in written or formal language,

for example, grubby, pooch, and mad (in the sense of ‘angry’). But there

is no clear dividing line between these words and words which are

perhaps too technical or scientifically specialized ( circumfix, triptan ),

not quite assimilated enough into English ( tchachka, sambal oelek ), too

bound to a specific dialect ( frappe, black ice ), or too informal, imperma-

nent, or bound too narrowly to a particular time or a particular seg-

ment of society ( groovy, homie ). Deciding which of these uncommon

words merit inclusion in the dictionary is a judgment call, often based

more on practical considerations – the size of the dictionary, its

FIGURE 2.1 Reproduced with permission of Oxford University Press, James A.H. Murray et al. 1888. A New English Dictionary on Historical Principles, p. xvii

SCIENTIFIC

TECHNICAL COMMON

LITERARY

COLLOQUIAL

SLANG

FOREIGN

DIALECTAL

Of Obscure

Meaning in the

OED :

These entries for

smazky and val-

dunk are among

the 87 entries that

the OED designates

as “meaning

obscure” or “of

obscure meaning.”

Did they deserve to

be in the OED? You

decide!

smazky , a . Obs.

(Meaning obscure.)

1599 MIDDLETON

Micro-cynicon A5,

Auant, . . Ile anger

thee inough, And

fold thy firy-eyes in

thy smazkie snufe.

val-dunk Obs.

(Meaning obscure.)

1631 R. BRATHWAIT

Whimzies, Wine-

soaker 102 By this

time his cause is

heard, and now this

val-dunke growne

rampant�drunke,

would fight if hee

knew how.

Words, dictionaries, and the mental lexicon 29

intended audience – than on strict linguistic principles. All lexicogra-

phers face this conundrum, and each one makes a slightly different

decision.

The OED is certainly the gold standard for English dictionaries today.

Nevertheless, it has its own idiosyncrasies. For example, it contains a

number of nonce words , words that are attested only once. Indeed, there

are quite a few nonce words that the OED includes, even though it is

unable to define them. A search, using the key words “meaning obscure”

and “of obscure meaning,” turns up 87 words so labeled, including

smazky, squirgliting, val-dunk, vezon, uncape, and umbershoot . Each bears an

entry illustrated with one quotation, which unfortunately does not illu-

minate the word’s meaning. Nevertheless, these and 81 others like them

made it into the OED !

2.4.5 Modern dictionaries Today, there are dozens of dictionaries available for English – unabridged

dictionaries, college dictionaries, children’s dictionaries, specialized dic-

tionaries of music or architecture, an official Scrabble dictionary, not to

mention online dictionaries in many varieties. Each one of these is edited

by a team of individuals who make the judgment call whether xyz

deserves to be in the dictionary. The decision is made on a number of

grounds:

• the size of the dictionary, which determines the number of words it can hold;

• the intended audience of the dictionary (adults, children, language learners, etc.);

• whether a word has a sufficiently broad base of usage; • whether it’s likely to last; • whether it’s too specialized or technical for the intended audience; • for a word borrowed from another language, whether it’s assimilated

enough to be considered part of English.

With respect to size, the number of words and the depth of entries (wheth-

er etymologies and illustrative quotes are included, for example) in print

dictionaries are determined by the number of pages and the font size of the

print used. Online dictionaries do not have the sort of space constraints

that print dictionaries do. As for audience, a dictionary intended for college-

age adults will probably have more learned and technical words than a

dictionary for children. On the other hand, words that a native speaker is

unlikely to need defined might be more of a focus in a dictionary for

English language learners; the meanings of prepositions and their idiomat-

ic uses come to mind here. The type of dictionary also determines how

broad a base of usage a word needs to have in order to be included.

Dictionaries of slang, dialect, or specialized fields obviously contain more

narrowly used words than general dictionaries do (although you might be

surprised at how much slang and technical terminology can be found in

general dictionaries).

30 INTRODUCING MORPHOLOGY

Perhaps the trickiest issue is how long a word has to have been around

to merit inclusion in the dictionary. These days, words can appear in dic-

tionaries fairly quickly, especially in online dictionaries. The OED already

has entries for the words f lexitarian , hashtag , and selfie ; the first citation for

the last of these was as far back as 2002. The word bouncebackability –

allegedly coined by British sportscaster Iain Dowie in 2004 (Hohenhaus

2006 ) – already had a draft OED entry by June, 2006 (although, interest-

ingly, the OED has traced the word as far back as 1961!).

And there is more to consider in deciding whether a word goes into the

dictionary. Take, for example, words formed with various prefixes and suf-

fixes. If happy is in the dictionary (as it certainly would be), do we need to

have an entry as well for happiness ? Similarly, if sad has an entry, do we

need sadness ? If our audience is a learner of English, perhaps yes, but for

native speakers who know intuitively how the suffix -ness is used, is there

any need for these extra entries? Interestingly, dictionaries are often quite

inconsistent on how many and which derivatives with particular suffixes

get entries. The online OED has entries for redness, blueness, pinkness, green-

ness, and yellowness , but not orangeness . The word purpleness is used in the

definition of the word purplely , but does not have its own entry. And not

surprisingly, there is no entry for mauveness or beigeness . What is more

surprising is that there are so many entries for color words with the suffix

-ness attached.

Certainly, if a word derived with a prefix or suffix takes on an idiosyn-

cratic or lexicalized meaning, the dictionary needs to include it. Take, for

example, the word transmission , which can have the transparent meaning

‘the act of transmitting’ but probably more often is used to denote a part

of a car. This second meaning probably deserves to be in the dictionary.

But is it necessary to include all derived words whose meanings are

perfectly clear from the meaning of the base plus the meaning of the

affix ? Probably not.

Until the last decade of the twentieth century lexicographers made

their decisions by reading materials of all sorts, and in more recent

decades by listening to radio, TV, and talk in general. Potential entries

would be recorded with their context on small slips of paper. These slips

would then be filed, and when a critical mass of usages accumulated for

a word, it might be considered for entry in the dictionary. These days

lexicographers are aided by corpora (singular corpus ), large computerized

databases that can be searched for words in the context of their use, and

by the internet, which might be viewed as a vast corpus. Indeed the rise

(and sometimes fall) of a new word can be traced by searching for its use

on the internet. We will see in the next chapter that corpora can also be

invaluable in the study of word formation rules.

Perhaps the most interesting recent development in lexicography is

the rise of Wiktionary – an online collaborative dictionary created not

by professional lexicographers, but by users themselves ( www.wiktion-

ary.org/ ). In the instructions for submitting entries, Wiktionary asks

that words be attested, by which it means they must be in widespread

use, available in well-known works or refereed publications, used at least

Words, dictionaries, and the mental lexicon 31

3. According to the New Columbia Encyclopedia, Lillian Virginia Mountweazel lived from 1942 to 1973. A

fountain designer and photographer, she was supposedly well known for taking pictures of rural American

mailboxes. She died tragically in an explosion while she was on assignment for Combustibles magazine.

three times in at least three sources over more than a year. It does, how-

ever, have a category of what it calls ‘protologisms’ for “terms defined in

the hopes that they will be used, but which are not actually in wide use.”

One final note about the vagaries of dictionaries. Lest you think that

lexicographers are humorless (“harmless drudges” as Johnson calls them

in his dictionary), let’s consider the issue of mountweazels mentioned

briefly above. As Henry Alford reveals in the August 29, 2005 issue of The

New Yorker, the editors of the New Oxford American Dictionary (2001) planted

the non-existent word esquivalience (defined as “the willful avoidance of

one’s official responsibilities”) among the entries for the letter “e” to

catch potential dictionary pirates. Such false words are called ‘mountwea-

zels’, from the fictitious entry for Lillian Virginia Mountweazel in the New

Columbia Encyclopedia . 3 What is most interesting for our purposes is that

once these fake words have been coined, they take on lives of their own.

As of June 2014, there were 52,300 hits for esquivalience and 13,200 for

mountweazel on Google, leading me to wonder whether these fakes have

now become real words .

2.4.6 And other languages I have concentrated here on the history of dictionary-making in English,

but the same points might be made with respect to dictionaries of French,

Italian, Russian, Chinese, or Central Alaskan Yup’ik. All dictionaries are

products of individuals and all display the choices and idiosyncrasies of

those individuals in some way or another.

Dictionaries of other languages might be organized quite differently

from those of the Indo-European languages that we are most familiar

with, however. For example, dictionaries of Mandarin Chinese are not

alphabetized in the way that dictionaries of English and French are,

because Chinese is not written in the Roman alphabet. Instead, the writ-

ing system (or orthography ) of Chinese is logographic or word-based.

Each word in Chinese is represented by a single character (or sometimes a

combination of two characters). When you look up a word in a Chinese

dictionary, you need to know how many strokes or lines make up that

character. Dictionaries are organized from those characters made up of

the fewest strokes to those containing the most strokes.

Dictionaries of other languages might include many fewer complex

words than English dictionaries typically do. For example, if a language

has very regular rules of word formation such that both the form and the

resulting meaning of a complex word are perfectly predictable, the dic-

tionary will have no need to list all complex words in separate entries. All

it needs to do is list individual morphemes with their meanings (and

perhaps some indication of how they combine). But the less predictable

the form and meaning of complex words are, the greater the need to put

them in the dictionary.

32 INTRODUCING MORPHOLOGY

Summary In this chapter we have been concerned with the question of what constitutes a word. We have contrasted dictionaries with the mental

lexicon. Dictionaries are written constructs that record words, along

with their pronunciations, meanings, etymologies, and perhaps

examples of use. On the one hand, they do not and cannot contain

everything that a native speaker would recognize as words of her

language – dictionaries have no need to record regularly inflected

forms of words and words derived by very active rules of word

formation, for example. On the other hand, dictionaries may include

items that perhaps don’t deserve to be considered real words – for

example, nonce words that are undefinable, or artificially created

words put in to check for copyright violations.

Our mental lexicons are something different, however. High-school

educated adults may have vocabularies of 60,000 words. We acquire

these words rapidly, and sometimes our mental representations are

sketchy or incomplete. The evidence we have looked at from aphasia

and genetic disorders, as well as studies using various imaging tech-

niques, allows us to begin to develop a picture of how these vast num-

bers of words are organized in our minds, although the picture is still

quite unclear. Unlike dictionaries that list words alphabetically, our

mental lexicon is organized as a complex web of entries that are linked

in various ways, along with a system of rules for combining listed

forms. It appears that entries and rules are at least to some extent

wired into different parts of the brain.

Exercises 1. Go to the OED Online website and search for words that are in the

dictionary but have no known definition. To do this, click on Advanced Search (look towards the top of the OED home page), and type into the first open box “meaning obscure” or “of obscure meaning.” Then choose three words and read through their entries. Do you think the OED was justified in including these words? If so, why? If not, why not?

2. Make a list of five words that you consider to be slang . Now look them up in your dictionary (you may use any dictionary at hand, whether print or online). First note whether or not you find them. If you do, is the dictionary definition the one that you had in mind? Does your dic- tionary list them as slang? If not, speculate on why they might not be listed as slang.

3. Make a list of at least ten words that come to mind that end in the suffix - less . Look these words up in a dictionary (you may use a standard col- lege desk dictionary like the American Heritage Dictionary or you may use the online OED ). How many of your words are in the dictionary? Is there any pattern that you can discern with respect to the words that are listed, as opposed to the words that are not?

Words, dictionaries, and the mental lexicon 33

4. Visit the Word Spy website ( www.wordspy.com ). Look at the list of new words and decide which ones, if any, are part of your own mental lexi- con. If some of them are, compare your understanding of them with the definition that Word Spy gives.

5. Of all the original words created by J.K. Rowling in the Harry Potter books, the only one to make its way into the online Oxford English Dictionary is the word muggle . Look up the ‘Harry Potter’ meaning of muggle and speculate on why this word alone has made it into the OED .

6. At the 2015 meeting of the American Dialect Society, the word #blacklivesmatter was voted the Word of the Year for 2014. Discuss the issues that are raised by this form for the definition of ‘word’. If you’d like more information, go to the American Dialect Society homepage ( www.americandialect.org ) and click on “2014 Word of the Year.”

KEY TERMS derivation

affixation

compounding

conversion

coinage

blending

backformation

corpus

formative

cran morph

extender

sound symbolism

splinter

CHAPTER OUTLINE

In this chapter you will learn about common ways of

creating new lexemes.

♦ We will look at derivational affixation, considering the

distinction between affixes and bases, and between free

and bound bases.

♦ We will learn how to segment words into morphemes,

how to formulate word formation rules, and how to

determine the structure of words.

♦ We will consider what morphemes mean.

♦ Beyond affixation, in this chapter we will learn about

processes of compounding, conversion, and other ways

of creating new words.

♦ And you will learn to use corpora to search for your own

morphological data.

3 C H A P T E R

Lexeme formation: the familiar

36 INTRODUCING MORPHOLOGY

3.1 Introduction

Take a look at the words below:

• unwipe (v.) • conversate (v.) • googlical (a.) • fiberhood (n.) • nympho-ology (n.) • snorgle (v.) • crankypants (n.) • swapoutable (a.)

Have you ever heard these words before? Can you imagine what they

mean?

Chances are that you haven’t heard or read them before. Nevertheless,

you probably didn’t have much trouble figuring out at least roughly what

their meanings might be. Assuming that you know the noun conversation ,

you can surely guess that conversate must mean ‘to have a conversation’.

Googlical must mean ‘pertaining to Google’, and in the phrase Googlical

proportions , metaphorically must mean ‘huge’. The verb snorgle is harder,

but you might guess that it’s probably onomatopoeic. 1 And crankypants,

fiberhood , and swapoutable are pretty transparent if you know what the

individual morphemes mean.

The reason you can make educated guesses about these words is that

they follow the rules of word formation in English. Once you know what

the base – the central bit of the word – means, you can often figure out

everything else. In this chapter, we’re going to look at the most common

ways of forming new lexemes in English and in other languages of the

world. You’ll learn how to analyze words into their component parts, see

how those parts are organized, and how the various parts contribute to

their meanings.

3.2 Kinds of morphemes

Most native speakers of English will recognize that words like Googlical ,

crankypants , or fiberhood are made up of several meaningful pieces, and

will be able to split them into those pieces:

(1) un / wipe

Google / ical

cranky / pants

fiber / hood

As you learned in Chapter 1 , these pieces are called morphemes , the

minimal meaningful units that are used to form words. Some of the mor-

phemes in (1) can stand alone as words: wipe, Google, cranky, pants, fiber .

1. It’s the noise that a small dachshund makes when clearing its throat. Thanks to my 2012 Morphology

students for these examples!

Lexeme formation: the familiar 37

These are called free morphemes . The morphemes that cannot stand

alone are called bound morphemes . In the examples above, the bound

morphemes are - ical, un- , and -hood . Bound morphemes come in different

varieties. Those in (1) are prefixes and suffixes ; the former are bound

morphemes that come before the base of the word, and the latter bound

morphemes that come after the base. Together, prefixes and suffixes can

be grouped together as affixes. 2

New lexemes that are formed with prefixes and suffixes on a base are

often referred to as derived words, and the process by which they are

formed as derivation . The base is the semantic core of the word to which

the prefixes and suffixes attach. For example, wipe is the base of unwipe ,

and Google is the base of Googlical . Frequently, the base is a free morpheme,

as it is in these two cases. But stop a minute and consider the data in the

next Challenge box.

Challenge

Divide the following words into morphemes:

• pathology

• psychopath

• dermatitis

• endoderm

Chances are that you recognize that there are two morphemes in

each word. However, neither part is a free morpheme. Do we want to

call these morphemes prefixes and suffixes? Would this seem odd to

you?

If you said that it would be odd to consider the morphemes in our

Challenge as prefixes and suffixes, you probably did so because this would

imply that words like pathology and psychopath are made up of nothing but

affixes!

Morphologists therefore make a distinction between affixes and bound

bases . Bound bases are morphemes that cannot stand alone as words, but

are not prefixes or suffixes. Sometimes, as is the case with the morphemes

path or derm , they can occur either before or after another bound base:

path precedes the base ology , but follows the base psych(o) ; derm precedes

another base in dermatitis but follows one in endoderm . This suggests that

path and derm are not prefixes or suffixes: there is no such thing as an

affix which sometimes precedes its base and sometimes follows it. But not

all bound bases are as free in their placement as path ; for example, psych(o)

and ology seem to have more fixed positions, the former usually preceding

another bound base, the latter following. Similarly, the base -itis always

follows, and endo- always precedes another base. Why not call them respec-

tively a prefix and a suffix, then?

2. We will see in Chapter 5 that there are other types of affi xes as well.

38 INTRODUCING MORPHOLOGY

One reason is that all of these morphemes seem in an intuitive way to

have far more substantial meanings than the average affix does. Whereas

a prefix like un- ( unhappy, unwise ) simply means ‘not’ and a suffix - ish ( red-

dish, warmish ) means ‘sort of ’, psych(o) means ‘having to do with the mind’,

- ology means ‘the study of ’, path means ‘sickness’, derm means ‘skin’, and

-itis means ‘disease’. Semantically, bound bases can form the core of a

word, just as free morphemes can. Figure 3.1 summarizes types of mor-

phemes. We’ll look more carefully at the meanings of affixes in

section 3.3 .

FIGURE 3.1 Types of morphemes

morphemes

boundfree

affix base

prefix suffix other types

to be introduced

in Chapter 5

frog

ride

re-

un-

-ness

-ize

derm

endo

Another reason to believe that bound bases are different from prefixes

and suffixes is that prefixes and suffixes tend to occur more freely than

bound bases do. For example, any number of adjectives can be made nega-

tive by using the prefix un- , but there are far fewer words with the bound

base psych(o) . This is perhaps not the best way of distinguishing between

bound bases and affixes, though, as there are a few bound bases – -ology is

one of them – that occur with great freedom, and there are some prefixes

and suffixes that don’t occur all that often (e.g., the -th in width or health ).

So we’ll stick with the criterion of ‘semantic robustness’ for now. We’ll

return in the next chapter to the question of how freely various mor-

phemes are used in word formation.

With regard to bases, another distinction that’s sometimes useful in

analyzing languages other than English is the distinction between root

and stem . In languages with more inflection than English, there is often

no such thing as a free base : all words need some sort of inflectional end-

ing before they can be used. Or to put it differently, all bases are bound.

Consider the data below from Latin :

(2) Latin 1st sg am + o ‘I love’ pl am + a + mus ‘we love’

dic + o ‘I say’ dic + i + mus ‘we say’

In the singular, an ending signaling the first person (“I”) can sometimes

attach to the smallest bound base meaning ‘love’ or ‘say’; this morpheme

is the root . In the first person plural, and in most other persons and num-

bers, however, another morpheme must be added before the inflection

Lexeme formation: the familiar 39

3. Some speakers will fi nd the forms in (3b) odd, and will question their acceptability, but they are all attested

in the Corpus of Contemporary American English and discussed in Bauer, Lieber, and Plag ( 2013 ).

goes on. This morpheme (an a for the verb ‘love’ and an i for the verb ‘say’)

doesn’t mean anything, but still must be added before the inflectional

ending can be attached. The root plus this extra morpheme is the stem .

Thought of another way, the stem is usually the base that is left when the

inflectional endings are removed. We will look further at roots and stems

in Chapter 6 , when we discuss inflection more fully.

3.3 Affixation

3.3.1 Word formation rules Let’s look more carefully at words derived by affixation . Prefixes and suf-

fixes usually have special requirements for the sorts of bases they can

attach to. Some of these requirements concern the phonology (sounds) of

their bases, and others concern the semantics (meaning) of their bases –

we will return to these shortly – but the most basic requirements are often

the syntactic part of speech or category of their bases. For example, the

suffix -ness attaches freely to adjectives, as the examples in (3a) show and

sometimes to nouns (as in (3b) ), but not to verbs (3c) :

(3) a. - ness on adjectives: redness, happiness, wholeness, commonness,

niceness

b. - ness on nouns: appleness, babeness, couch-potatoness 3

c. - ness on verbs: *runness, *wiggleness, *yawnness

The prefix un- attaches to adjectives (where it means ‘not’) and to verbs

(where it means ‘reverse action’), but not to nouns:

(4) a. un - on adjectives: unhappy, uncommon, unkind, unserious

b. un- on verbs: untie, untwist, undress, unsnap

c. un- on nouns: *unchair, *unidea, *ungiraffe

We might begin to build some of the rules that native speakers of English use

for making words with -ness or un- by stating their categorial requirements:

(5) Rule for -ness (first version): Attach - ness to an adjective or to a noun.

Rule for un- (first version): Attach un- to an adjective or to a verb.

Of course, if we want to be as precise as possible about what native speak-

ers know about forming words with these affixes, we should also indicate

what category of word results from using these affixes, and what the

resulting word means. So a more complete version of our - ness and un-

rules might look like (6) :

(6) Rule for - ness (second version): - ness attaches to adjectives or nouns

‘X’ and produces nouns meaning ‘the quality of X’.

Rule for un- (second version): un- attaches to adjectives meaning ‘X’

and produces adjectives meaning ‘not X’; un - attaches to verbs

meaning ‘X’ and produces verbs meaning ‘reverse the action X’.

40 INTRODUCING MORPHOLOGY

If we’re really trying to model what native speakers of English know about

these affixes, we might try to be even more precise. For example, un- does

not attach to all adjectives or verbs, as you can discover by looking at the

next Challenge box.

Challenge

Look at the following words and try to work out more details of the

rule for un- in English. The (a) list contains some adjectives to which

negative un- can be attached and others which seem impossible or at

least somewhat odd. The (b) list contains some verbs to which

reversative un- can attach and others which seem impossible. See if

you can discern some patterns:

(a) unhappy, *unsad, unlovely, *unugly, unintelligent, *unstupid

(b) untie, unwind, unhinge, unknot, *undance, *unyawn, *unexplode,

*unpush

What the (a) examples in the Challenge box seem to show is that the negative

prefix un- in English prefers to attach to bases that do not themselves have

negative connotations. This is not true all of the time – adjectives like unself-

ish or unhostile are attested in English – but it’s at least a significant tendency.

As for the (b) examples, they suggest that the un- that attaches to verbs pre-

fers verbal bases that imply some sort of result, and moreover that the result

is not permanent. Verbs like dance, push, and yawn denote actions that have

no results, and although explode implies a result (i.e., something is blown up),

it’s a result that is permanent. In contrast, a verb like tie implies a result

(something is in a bow or knot) which is temporary (you can take it apart).

We have just constructed what morphologists call a word formation

rule , a rule which makes explicit all the categorial, semantic, and phono-

logical information that native speakers know about the kind of base that

an affix attaches to and about the kind of word it creates. We might now

state the full word formation rules for negative un- as in (7) :

(7) Rule for negative un- (final version): un- attaches to adjectives,

preferably those with neutral or positive connotations, and creates

negative adjectives. It has no phonological restrictions.

Now let’s look at two more affixes. In English we can form new verbs by

using the suffixes -ize or -ify . Both of these suffixes attach to either nouns

or adjectives, resulting in verbs:

(8) -ize on adjectives: civilize, idealize, finalize, romanticize,

tranquillize

-ize on nouns: unionize, crystallize, hospitalize, caramelize,

animalize

-ify on adjectives: purify, glorify, uglify, moistify, diversify

- ify on nouns: mummify, speechify, classify, brutify, scarify,

bourgeoisify

Lexeme formation: the familiar 41

We might state the word formation rules for -ize and -ify as in (9) :

(9) Rule for -ize (first version): -ize attaches to adjectives or nouns that

mean ‘X’ and produces verbs that mean ‘make/put into X’.

Rule for -ify (first version): -ify attaches to adjectives or nouns that

mean ‘X’ and produces verbs that mean ‘make/put into X’.

But again, we can be a bit more precise about these rules. Although -ize

and -ify have almost identical requirements for the category of base they

attach to and produce words with roughly the same meaning, they have

somewhat different requirements on the phonological form of the stem

they attach to. As the examples in (8) show, -ize prefers words with two or

more syllables where the final syllable doesn’t bear primary stress (e.g.,

trá nquil, hó spital ). The suffix -ify , on the other hand, prefers monosyllabic

bases ( pure, brute, scar ), although it also attaches to bases that end in a - y

( mummy, ugly ) or bases whose final syllables are stressed ( divé rse, bourgé ois ).

Since we want to be as precise as possible about our word formation rules

for these suffixes, we will state their phonological restrictions along with

their categorial needs:

(10) Rule for -ize (final version): -ize attaches to adjectives or nouns of

two or more syllables where the final syllable does not bear

primary stress. For a base ‘X’ it produces verbs that mean ‘make/put

into X’.

I leave it to you to come up with the final version of the word formation

rule for - ify.

3.3.2 Word structure When you divide up a complex word into its morphemes, as in (11) , it’s

easy to get the impression that words are put together like the beads that

make up a necklace – one after the other in a line:

(11) unhappiness = un + happy + ness

But morphologists believe that words are more like onions than like neck-

laces: onions are made up of layers from innermost to outermost. Consider

a word like unhappiness . We can break this down into its component mor-

phemes un + happy + ness , but given what we learned above about the

properties of the prefix un- and the suffix - ness we know something more

about the way in which this word is constructed beyond just its constitu-

ent parts. We know that un- must first go on the base happy . Happy is an

adjective, and un- attaches to adjectives but does not change their catego-

ry. The suffix -ness attaches only to adjectives and makes them into nouns.

So if un- attaches first to happy and -ness attaches next, the requirements

of both affixes are met. But if we were to do it the other way around, -ness

would have first created a noun, and then un- would be unable to attach.

We could represent the order of attachment as if words really were

onions, with the base in the innermost layer, and each affix in its own

succeeding layer: see Figure 3.2 .

42 INTRODUCING MORPHOLOGY

But linguists, not generally being particularly artistic, prefer to show

these relationships as ‘trees’ that look like this :

(12)

un happy ness

Similarly, we might represent the structure of a word like repurify as in (13) :

(13)

re pure ify

In order to draw this structure, we must first know that the prefix re-

attaches to verbs (e.g., reheat, rewash , or redo ) but not to adjectives ( *repure,

*rehappy ) or to nouns ( *rechair, *retruth ). Once we know this, we can say that

the adjective pure must first be made into a verb by suffixing -ify , and only

then can re- attach to it.

FIGURE 3.2 Words are like onions

Challenge

In English, the suffix -ize attaches to nouns or adjectives to form

verbs. The suffix -ation attaches to verbs to form nouns. And the suffix

-al attaches to nouns to form adjectives. Interestingly, these suffixes

Lexeme formation: the familiar 43

3.3.3 What do affixes mean? When we made the distinction between affixes and bound bases above, we

did so on the basis of a rather vague notion of semantic robustness; bound

bases in some sense had more meat to them than affixes did. Let us now

attempt to make that idea a bit more precise by looking at typical mean-

ings of affixes.

In some cases, affixes seem to have not much meaning at all. Consider

the suffixes in (14) :

(14) a. -(a)tion examination, taxation, realization, construction

- ment agreement, placement, advancement, postponement

-al refusal, arousal, disposal

b. -ity purity, density, diversity, complexity

-ness happiness, thickness, rudeness, sadness

Beyond turning verbs into nouns with meanings like ‘process of X-ing’ or

‘result of X-ing’, where X is the meaning of the verb, it’s not clear that the

suffixes -(a)tion, -ment , and -al add much of any meaning at all. Similarly

with -ity and -ness , these don’t carry much semantic weight of their own,

aside from what comes with turning adjectives into nouns that mean

something like ‘the abstract quality of X’, where X is the base adjective.

Affixes like these are sometimes called transpositional affixes , meaning

that their primary function is to change the category of their base with-

out adding any extra meaning.

Contrast these, however, with affixes like those in (15) :

(15) a. -ee employee, recruitee, deportee, inductee

b. -less shoeless, treeless, rainless, supperless

c. re- reheat, reread, rewash

These affixes seem to have more semantic meat on their bones, so to

speak: -ee on a verb indicates a person who undergoes an action; -less

means something like ‘without’; and re- means something like ‘again’.

Languages frequently have affixes (or other morphological processes, as

we’ll see in Chapter 5 ) that fall into common semantic categories. Among

those categories are:

• personal or participant affixes: These are affixes that create ‘people nouns’ either from verbs or from nouns. Among the personal affixes

in English are the suffix -er which forms agent nouns (the ‘doer’ of

can be attached in a recursive fashion: convene → convention →

conventional → conventionalize → conventionalization .

First draw a word tree for conventionalization . Then see if you can

find other bases on which you can attach these suffixes recursively.

What is the most complex word you can create from a single base

that still makes sense to you? Are there any limits to the complexity

of words derived in this way?

44 INTRODUCING MORPHOLOGY

the action) like writer or runner and the suffix -ee which forms patient

nouns (the person the action is done to). Also among this class of

affixes are ones that create other ‘people nouns’ other than the agent

or the patient in an event, for example, inhabitants of a place (like

Manhattanite or Bostonian ).

• locative affixes: These are affixes that designate a place. For example, in English we can use the suffix - ery or - age to denote a place where

something is done or gathered (like eatery or orphanage ).

• abstract affixes: These are affixes that create abstract nouns that denote qualities (like happiness or purity ) or statuses ( puppydom, advisor-

ship, daddyhood ) or even aspects of behavior ( buffoonery ).

• negative and privative affixes: Negative affixes add the meaning ‘not’ to their base; examples in English are the prefixes un- , in- , and non-

( unhappy, inattentive, non-functional ). Privative affixes mean something

like ‘without X’; in English, the suffix -less ( shoeless, hopeless ) is a priva-

tive suffix, and the prefix de- has a privative flavor as well (e.g., words

like debug or debone mean something like ‘cause to be without bugs/

bones’).

• prepositional and relational affixes: Prepositional and relational affixes often convey notions of space and/or time. Examples in

English might be prefixes like over- and out- or pre - and post - ( overfill,

overcoat, outrun, outhouse, preschool, preheat, postwar, postdate ).

• quantitative affixes: These are affixes that have something to do with amount. In English we have affixes like -ful ( handful, helpful )

and multi - ( multifaceted ). Another example might be the prefix re-

that means ‘repeated’ action ( reread ), which we can consider quan-

titative if we conceive of a repeated action as being done more

than once. Other quantitative affixes that we have in English

denote collectives or aggregates of individuals (e.g., acreage or

knickknackery ).

• evaluative affixes: Evaluative affixes consist of diminutives , affixes that signal a smaller version of the base (e.g., in English -let as in

booklet or droplet ) and augmentatives , affixes that signal a bigger ver-

sion of the base. The closest we come to augmentative affixes in

English are prefixes like mega- ( megastore , megabite ). The Native

American language Tuscarora (Iroquoian family) has an augmenta-

tive suffix - ʔ o ʔ y that can be added to nouns to mean ‘a big X’; for

example, takó:θ- ʔ o ʔ y means ‘a big cat’ (Williams 1976 : 233). Diminutives and augmentatives frequently bear other nuances of

meaning. For example, diminutives often convey affection, or

endearment, as we find in some words with - y or - ie in English (e.g.,

sweetie, kitty ). Augmentatives sometimes have pejorative overtones

(the Niger-Congo language Fula has a pejorative diminutive, for

example).

Note that some semantically contentful affixes change syntactic category

as well; for example, the suffixes -er and -ee change verbs to nouns, and the

prefix de- changes nouns to verbs. But semantically contentful affixes

Lexeme formation: the familiar 45

need not change syntactic category. The suffixes - hood and -dom , for

example, do not ( childhood, kingdom ), and by and large prefixes in English

do not change syntactic category.

So far we have been looking at suffixes and prefixes whose meanings

seem to be relatively clear. Things are not always so simple, though. Let’s

look more closely at the suffix -er in English, which we said above formed

agent nouns. Consider the following words:

(16) a. writer

skater

b. printer

freighter

c. loaner

fryer (i.e., a kind of chicken)

d. diner

All of these words seem to be formed with the same suffix. Look at each

group of words and try to characterize what their meanings are. Does -er

seem to have a consistent meaning?

It’s rather hard to see what all of these have in common. The words in

(16a) are indeed all agent nouns, but the (b) words are instruments; in

other words, things that do an action. In American English the (c) words

are things as well, but things that undergo the action rather than

doing the action (like the patient -ee words discussed above): a loaner is

something which is loaned (often a car, in the US), and a fryer is some-

thing (a chicken) which is fried. And the word diner in (d) denotes a loca-

tion (a diner in the US is a specific sort of restaurant). Some morphologists

would argue that there are four separate suffixes in English, all with the

form -er . But others think that there’s enough similarity among the

meanings of -er words in all these cases to merit calling -er a single affix,

but one with a cluster of related meanings. All of the forms derived with

-er denote concrete nouns, either persons or things, related to their base

verbs by participating in the action denoted by the verb, although some-

times in different ways. This cluster of related meanings is called affixal

polysemy .

Affixal polysemy is not unusual in the languages of the world. For

example, it is not unusual for agents and instruments to be designated by

the same suffix. This occurs in Dutch , as the examples in (17a) show (Booij

and Lieber 2004 ), but also in Yoruba (Niger-Congo family), as the examples

in (17b) show (Pulleyblank 1987 : 978):

(17) a. Dutch

spel-er ‘player’ ( spelen ‘play’)

Maai-er ‘mower’ ( maaien ‘mow’)

b. Yoruba

a-pànìà ‘murderer’ ( pa ‘kill’, ènìà ‘people’)

a-bẹ ‘razor, penknife’ ( bẹ ‘cut’)

46 INTRODUCING MORPHOLOGY

The Dutch suffix -er is in fact quite similar to the -er suffix in English in

the range of meanings it can express. The Yoruba prefix a- also forms both

agents and instruments.

3.3.4 To divide or not to divide? A foray into extenders , formatives , crans , and other messy bits In Chapter 1 we defined a morpheme as the smallest unit of language that

has its own meaning. We have now looked at affixes and bases, both free

and bound, and considered their meanings and how they combine into

complex structures. For the most part, the examples we have looked at are

simple and their analysis has been relatively clear. But often the closer we

look at the morphology of a language, the more complex it becomes.

There are many ways we can illustrate this with the morphology of

English , but we will choose just a few points in this section that compli-

cate our initial picture. I should point out in advance that my examples

here will come from English, but we should expect to find similar twists

and turns in any language whose morphology we examine in detail.

Consider the words in (18) :

(18) a. report, import, transport, deport, comport, export

b. cranberry, huckleberry, raspberry

c. Platonic, tobacconist, spasmodic, egotist

d. sniffle, snort, snot, snout

e. eggitarian, pizzatarian, pastatarian, fruitarian, flexitarian

Let’s start with the ones in (18a) . We have assumed so far that words can

be broken down into morphemes, which are pieces that have meaning.

The words in (18a) certainly look like they might be broken down because

they have recurrent parts, but if they are, what do the pieces mean? In

fact, English has dozens of words that are similar to what we might call

the - port family. See how many cells of Table 3.1 you can fill in.

Table 3.1

in- ex- con- re- trans- de-

-port

-mit

-ceive

-duce

-cede

-fer

-scribe

-gress

-sist

Lexeme formation: the familiar 47

One reason for our dilemma in analyzing these forms is that they are

not native to English. They were borrowed from Latin (or from French ,

which in turn is descended from Latin), where they did have clear mean-

ings: -port comes from the verb portare ‘to carry’, - mit from the verb mittere

‘to send’, -scribe from the verb scribere ‘to write’, and so on. But English

speakers (unless they’ve studied Latin!) don’t know this. Morphologists are

left with an unsatisfying sense that the words above somehow ought to be

treated as complex, but are nevertheless reluctant to give up the strict

definition of morpheme. One way of dealing with these pieces is to

acknowledge that they seem to be independent and recombinable in some

way, but that they are not morphemes in the normal sense. Bauer, Lieber,

and Plag ( 2013 : 16) call elements like these formatives , which they define

as “elements contributing to the construction of words whose semantic

unity or function is obscure or dubious.”

The items in (18b) illustrate a different type of formative that are some-

times called cran morphs , from the first bit of the word cranberry . The

second part of the word cranberry is clearly a free morpheme. But when we

break it off, what’s left is a piece that doesn’t seem to occur in other words

(except in recent years, words like cranapple that are part of product

names), and doesn’t seem to mean anything independently. There are

quite a few of these cran morphs in the names of other types of berries:

rasp- in raspberry , huckle- in huckleberry . In cases such as these we are even

more tempted than we were with - port, -ceive , and the like to divide words

into morphemes, even though we know that one part of the word isn’t

meaningful in the way morphemes usually are.

The examples in (18c) also display a puzzling characteristic. If we try to

break these words down into their component morphemes, what we find

is that each one consists of two obvious morphemes plus an extra sound

or two:

(19) Plato + n + ic (compare icon+ic )

tobacco + n + ist (compare accordian+ist )

spasm + od + ic (compare Celt+ic )

ego + t+ ist (compare clarinet+ist )

The question is what the extra bit is. Is it part of the base of the word or part

of the affix or part of neither? It seems pretty clear that it doesn’t mean

anything. And why do we get an / n / in Platonic , but / od / in spasmodic , and noth-

ing between the base and the suffix in heroic , or an /n/ in tobacconist , but a /t/

in egotist ? Morphologists don’t have a clear answer to these questions – part

Challenge

Do you think that units like -port, -mit, -ceive , and the like should be

considered morphemes? If so, what problems do they present for our

definition of morpheme ? If not, what should we do about the

intuition that native speakers of English have that such words are

complex?

48 INTRODUCING MORPHOLOGY

of the fun of doing morphology is that we can argue about the possibilities,

after all! – but we can at least give these puzzling bits a name. Following

Bauer, Lieber, and Plag ( 2013 ), we will call them extenders .

Next, let’s look at the examples in (18d) . These exhibit what is called

sound symbolism . All of the words begin with the consonant cluster /sn/

and seem to have something to do with the nose, but the sequence of

sounds /sn/ doesn’t mean anything by itself. Here morphologists are rela-

tively agreed that sound symbolic words cannot be broken into parts. For

one thing, the sequence /sn/ doesn’t refer to ‘nose’ everywhere it occurs

(consider words like snail, snap , or snit ), and for another, if we were to seg-

ment /sn/ in the words in (18) , what would be left over would neither have

meaning by itself nor recur elsewhere in English.

Our final group of odd bits is illustrated in (18e) . In these examples, the

first part of each word is clearly a free morpheme, but the second part is

not. Rather, it is what Bauer, Lieber, and Plag ( 2013 ) call a splinter ,

something which is split off from an original word, but which is not

really (yet!) a true suffix. In the examples in (18e) , the splinter is tarian ,

which is a bit broken off from the word vegetarian , and then used to create

new words meaning ‘one who eats X’. English has lots of splinters, among

them tastic , as in funktastic or fishtastic , which is used to form mostly ironic

words meaning ‘excellent or great in reference to X’, originally from fan-

tastic , or licious , as in bagelicious or bootielicious , which is used to form words

meaning ‘appealing in reference to X’, originally from the word delicious .

The difference between a splinter and a true suffix is that speakers under-

stand splinters in relation to the original word from which the ending

splits off. If these bits survive and continue to give rise to new forms,

though, they might someday be real suffixes!

3.4 Compounding

So far we have concentrated on complex words formed by derivation, spe-

cifically by affixation. Derivation is not the only way of forming new

words, of course. Many languages also form words by a process called

compounding. Compounds are words that are composed of two (or more)

bases, roots, or stems. In English we generally use free bases to compose

compounds, as the examples in (20) show:

(20) English compounds

compounds of two nouns: windmill, dog bed, book store

compounds of two adjectives: icy cold, blue-green, red hot

compounds of an adjective and a noun: greenhouse, blackboard,

hard hat

compounds of a noun and an adjective: sky blue, cherry red, rock hard

3.4.1 When do we have a compound? How do we know that a sequence of words is a compound? Surprisingly,

it’s not that easy to come up with a single criterion that works in all cases.

Lexeme formation: the familiar 49

In English, spelling is no help at all; unlike German , where compounds

are always written as one word, 4 in English there is no fixed way to spell

a compound word. Some, like greenhouse, are written as one word, others

like dog bed, as two words, and still others, like producer-director are written

with a hyphen between the two bases.

A better criterion is stress; compounds in English are often stressed on

their first or left-hand base, whereas phrases typically receive stress on

the right. Compare, for example, a gré enhouse, which is the place where

plants are grown, to a green hó use, that is, a house that’s painted green.

But it’s not always the case that compounds are stressed on the left. For

example, most people pronounce apple pí e with stress on the second base,

but á pple cake with stress on the left one. Yet we have the feeling that both

are compounds; it seems illogical to consider one a compound and not

the other.

There is, however, one test for identifying compounds that is fairly reli-

able: we can test for whether a sequence of bases is a compound by seeing

if a modifying word can be inserted between the two bases and still have

the sequence make sense. If a modifying word cannot sensibly be inserted,

the sequence of two words is a compound. This test confirms that both

apple pie and apple cake are compounds, in spite of their differing stress. In

neither case can we insert a modifier like delicious between the two stems;

*apple delicious pie and *apple delicious cake are equally peculiar!

3.4.2 Compound structure We can look at compounds as having internal structure in precisely the

same way that derived words do, and we can represent that structure in

the form of word trees. The compounds windmill and hard hat would have

the structures in (21) :

(21)

A N

hard hat

N N

wind mill

Compounds, of course, need not be limited to two bases.

Compounding is what is called a recursive process, in the sense that a

compound of two bases can be compounded with another base, and

this compounded with still another base, so that we can eventually

obtain very complex compounds like paper towel dispenser factory build-

ing committee report . As with derived words, it is possible to show the

internal structure of complex compounds using word trees . Assuming

that this compound is meant to denote a report from the building

committee for the paper towel dispenser factory, we might give it the

structure in (22) :

4. English-speaking students of German often delight in compounds like Donaudampfschifffahrtsgesellschaft-

skapitänsmütze (Neef 2009 ), which means ‘Cap of the captain of the Danube steam ship company’.

50 INTRODUCING MORPHOLOGY

(22)

N N N N N N N

paper towel dispenser factory building committee report

Some compounds can be ambiguous, and therefore can be represented by

more than one structure. For example, the compound paper cat container,

might have this structure:

(23)

N N N

paper cat container

The way we’ve drawn this tree, the compound paper cat has been

compounded with the noun container to make a complex compound.

The compound as a whole then must mean ‘a container for paper cats’.

But if the compound paper cat container were intended to mean ‘a paper

container for cats’, the structure of the tree would be that in (24) ,

where cat container is first compounded, and then paper added in: 5

(24)

N N N

paper cat container

Often, the more complex the compound is, the greater the possibility of

multiple interpretations, and therefore multiple structures.

5. These two interpretations are sometimes distinguished in spoken speech by placement of stress.

Challenge

The compound paper towel dispenser factory building committee report

could in fact have more than one meaning. See how many different

meanings you can come up with, and draw a tree that corresponds to

each of those meanings.

Lexeme formation: the familiar 51

Languages other than English frequently construct compounds on free

bases just as English does, although we can see in the French and

Vietnamese examples in (25) that the order of elements in the compound

is sometimes different from that in English, a fact we will return to in the

next section:

(25) a. French:

timbre poste ‘stamp-post = postage stamp’

chêne liège ‘oak cork = cork oak’

b. Dutch:

boekhandel ‘book shop’

zakgeld ‘pocket money’

c. Vietnamese:

nhá thuong ‘establishment be-wounded’ = ‘hospital’

nguói ỏ ̣’ ‘person be-located’ = ‘servant’

As we saw above, English has bound bases as well as free bases, and

when we put two of them together, as in the examples in (26) , we might

call these forms compounds as well. Some linguists call them neo-

classical compounds , as the bound bases usually derive from Greek and

Latin:

(26) English compounds on bound bases: psychopath, pathology, endoderm,

dermatitis

In languages like Latin where, as we saw, word formation often operates

on roots or stems, rather than on free forms, all compounds are formed

from bound bases. Specifically, the first parts of the compounds in (27) are

formed from the roots of the nouns ala ‘wing’ and capra ‘goat’ (respec-

tively al- and capr -) plus a vowel -i- linking the two parts of the compound

together:

(27) Latin compounds:

ali-pes ‘wing-footed’

capri-ficus ‘goat fig’ = ‘wild fig’

The -i- that occurs between the two roots has no meaning, and is not the

vowel that usually precedes the inflections (for these two nouns, that

vowel would be -a ). It is there solely to link the parts of the compound

together, and might be seen as another example of the bits called

‘extenders’ that we discussed in section 3.3.4 . Other morphologists call

this bit a linking element or interfix .

3.4.3 Types of compounds In English and other languages there may be a number of different ways

of classifying compounds. Traditionally, English compounds have been

classified as either root (also known as primary ) compounds or synthetic

(also known as deverbal ) compounds. Synthetic compounds are composed

of two lexemes, where the second element is derived from a verb, and the

52 INTRODUCING MORPHOLOGY

6. We will go into arguments in more depth in Chapter 8 . For now, it’s enough to know that the arguments of

the verb are its subject and its complements (direct object, indirect object, and so on).

7. Or sometimes more, as we saw in the last section.

first is interpreted as an argument of that verb. 6 Dog walker, hand washing,

and home made are all synthetic compounds.

(28)

N N

V N

dog walk er

Root compounds , in contrast, are made up of two lexemes, 7 which may

be nouns, adjectives, or verbs; the second lexeme is typically not derived

from a verb. The interpretation of the semantic relationship between the

head and the non-head in root compounds is quite free as long as it’s not

the relationship between a verb and its argument. Compounds like wind-

mill, ice cold, hard hat, and red hot are root compounds.

Although this is a traditional way of classifying English compounds,

some morphologists like to use more fine-grained systems of classifications

that focus on several distinct aspects of compound structure, and this is

what we will do here. One advantage of doing this is that it makes it easier

for us to compare compounds in English with compounds in other lan-

guages. In this section, we will discuss three different ways of classifying

compounds: first, according to the part of speech or category of the words

that make them up; second, according to the semantic relationships that

they express; and third, according to the placement of what’s called the

head of the compound, a term that we will define shortly.

3.4.3.1 Classifying compounds according to category One simple way of looking at compounds is to see what category or part

of speech the parts of the compound belong to, and what part of speech

the compound as a whole exhibits. In English , as we saw in (20) , we can

form compounds from any combination of nouns and adjectives, but it’s

harder to form compounds using verbs or prepositions:

(29) compound category of

elements compound examples

N + N N dog bed, file cabinet, apple pie

N + A A sky blue, stone cold, bone dry

A + A A blue-green, icy cold

A + N N blackbird, greenhouse, fast food

N + V V hand wash, brainwash, babysit

V + N N pick pocket, sell sword, think tank

V + V V stir-fry, slam dunk, blow dry

A + V V hot glue, slow dance

V + A A go-slow

Lexeme formation: the familiar 53

8. Bauer et al. ( 2013 ) give this as an example of an N+P compound, but don’t say what the resulting category

is. In the case of most compounds, it’s clearly that of the right-hand category, but with this combination of

categories it’s not at all clear!

N + P Adv? year-in (Bauer et al. 2013 : 452) 8

P + N N afterbirth, backseat

V + P N breakdown

P + V V downgrade

A + P A tuned-in

P + A A inbuilt

P + P P into

In the cases with prepositions and verbs, it is difficult to find plausible

examples, and some morphologists would question whether we really can

form compounds composed of a noun plus a preposition or a verb plus an

adjective in English, for example.

3.4.3.2 Classifying compounds according to semantic relationships We can also classify compounds more closely according to the semantic

and grammatical relationships holding between the elements that make

them up. One useful classification is that proposed by Bisetto and Scalise

( 2005 ), which recognizes three types of relation. The first type is what

might be called an attributive compound . In an attributive compound

the non-head acts as a modifier of the head. So snail mail is (metaphori-

cally) a kind of mail that moves like a snail, and a windmill is a kind of mill

that is activated by wind. With attributive compounds the first element

might express just about any relationship with the head. For example, a

school book is a book used at school, but a yearbook is a record of school

activities over a year. And a notebook is a book in which one writes notes.

With a new compound (one I’ve just made up) like mud wheel , we are free

to come up with any reasonable semantic relationship between the two

bases, as long as the first modifies the second in some way: a wheel used

in the mud, a wheel made out of mud, a wheel covered in mud, and so on.

Some interpretations are more plausible than others, of course, but none

of these is ruled out.

In coordinative compounds , the first element of the compound does

not modify the second; instead, the two have equal weight. In English,

compounds of this sort can designate something which shares the denota-

tions of both base elements equally, or is a mixture of the two base ele-

ments, or expresses a relationship between the two elements:

(30) Coordinative compounds: producer-director, prince consort, blue-

green, doctor–patient

A producer-director is equally a producer and a director, a prince consort at

the same time a prince and a consort. In the case of blue-green the

compound denotes a mixture of the two colors. Finally, there are also

coordinative compounds that denote a relation between the two bases

54 INTRODUCING MORPHOLOGY

(like doctor–patient in doctor–patient confidentiality ). We will return to these

below.

We find a third kind of semantic/grammatical relationship in subordi-

native compounds . In subordinative compounds one element is inter-

preted as the argument of the other, usually as its object. Typically this

happens when one element of the compound either is a verb or is derived

from a verb, so the synthetic compounds we looked at at the very begin-

ning of this section are subordinative compounds in English. Some more

examples are given in (31) :

(31) with - er truck driver, mystery writer, lion tamer

with - ing truck driving, food shopping, hand holding

with - ation meal preparation, home invasion

with - ment cost containment

It is easy to see that subordinative compounds are interpreted in a very

specific way: that is, the first element of the compound is most often

interpreted as the object of the verb that forms the base of the deverbal

noun: for example, a truck driver is someone who drives trucks , food prepara-

tion involves preparing food , and so on.

Synthetic compounds are not the only subordinate compounds, how-

ever. A second type of subordinate compound is poorly represented in

English, but occurs with great frequency in Romance languages like

Spanish , French , and Italian :

(32) English pickpocket

Italian lava piatti (lit. ‘wash dishes’ = ‘dishwasher’)

Spanish saca corcho (lit. ‘pull cork’ = ‘corkscrew’)

In these compounds the first element is a verb, and the second bears an

argumental relationship to the first element, again typically the object

relationship. We will return to these shortly.

3.4.3.3 Classifying compounds according to headedness A third way of classifying compounds concerns whether or not they have

what is called a head, and if they do have a head, according to where the

head is located. We start with a definition: the head of the compound is

the element that serves to determine both the part of speech and the

semantic kind denoted by the compound as a whole. For example, in

English the base that determines the part of speech of compounds such as

greenhouse or sky blue is always the second one; the compound greenhouse is

a noun, as house is, and sky blue is an adjective as blue is. Similarly, the

second base determines the semantic category of the compound – in the

former case a type of building, and in the latter a color. English com-

pounds are therefore said to be right-headed . In other languages, howev-

er, for example French and Vietnamese, the head of the compound can be

the first or leftmost base. For example a timbre poste (33a) is a kind of

stamp, and a nguói ọ̉ ’ (33b) is a kind of person. French and Vietnamese can therefore be said to have left-headed compounds:

Lexeme formation: the familiar 55

(33) a. French:

timbre poste ‘stamp-post = postage stamp’

chêne liège ‘oak cork = cork oak’

b. Vietnamese:

nhá thuong ‘establishment be-wounded’ = ‘hospital’

nguói ỏ ̣’ ‘person be-located’ = ‘servant’

Using the notion of head, we can further divide compounds into endocen-

tric or exocentric varieties. In endocentric compounds, the referent of the

compound is always the same as the referent of its head. So a windmill is a

kind of mill, and a truck driver is a kind of driver. Endocentric compounds

are illustrated in (34) :

(34) Endocentric compounds

Atrributive: windmill, greenhouse, sky blue, icy cold

Subordinative: truck driver, meal preparation

The Dutch, French, and Vietnamese compounds in (33) are endocentric, as

well, although as we pointed out above, the head occurs on the left in

these compounds.

Coordinative compounds like producer-director or blue-green present a bit

of a complication for us here. On the one hand, it’s certainly true that a

producer-director is a kind of director, but he or she is also a kind of pro-

ducer, and blue-green is sort of a mixture between blue and green. We might

reasonably say then that these compounds have two semantic heads, in

which case we might still be justified in considering them endocentric.

Compounds may be termed exocentric when the denotation of the com-

pound as a whole is not the denotation of the head. For example, the English

attributive compounds in (35) all refer to types of people – specifically stupid

or disagreeable people – rather than types of heads, brains, or clowns, respec-

tively. So an air head is a person with nothing but air in her head, and so on.

(35) Exocentric compounds

Attributive: air head, meat head, bird brain, ass clown

Coordinative: parent-child, doctor-patient

Subordinative: pickpocket, cutpurse, lava piatti (Italian, lit. ‘wash

dishes’)

In coordinative compounds like parent–child or doctor–patient the compound

as a whole denotes a relationship between its elements; when we speak of

a parent–child conversation or a doctor–patient conference , the compound does

not have a single referent, and some would argue that it’s not even a

noun, even though it is made up of two nouns. So the coordinative exam-

ples in (35) would be exocentric as well. Finally, we saw examples of exo-

centric subordinative compounds from English, Spanish , and Italian in

(32) . English has only a few examples: a pickpocket is not a type of pocket,

but a sort of person (who picks pockets). Romance languages have many

compounds of this type, however.

The different types of compounds are summarized in Figure 3.3 .

56 INTRODUCING MORPHOLOGY

3.5 Conversion

Although we often form new lexemes by affixation or compounding, in

English it is also possible to form new lexemes merely by shifting the cat-

egory or part of speech of an already existing lexeme without adding an

affix. This means of word formation is often referred to as conversion or

functional shift . In English, we often create new verbs from nouns, as the

examples in (36a) show, but we also do the reverse (36b) , and sometimes

we can even create new verbs from adjectives (36c) :

(36) a. table to table

bread to bread

fish to fish

b. to throw a throw

to kick a kick

to fix a (quick) fix

c. cool to cool

yellow to yellow

When we create new verbs from nouns, the resulting verbs may have a

wide range of meanings. For example, to bread is ‘to put bread (crumbs) on

something’, but to fish is ‘to take fish from a body of water’. And to clown is ‘to

act like a clown’ rather than to put a clown somewhere or take a clown from

somewhere! Going in the opposite direction, the meaning of the new word

is somewhat more predictable; that is, when we turn a verb into a noun, the

result often means something like ‘an instance of X-ing’, where X is the deno-

tation of the verb. So for example, a throw is ‘an instance of throwing’.

English is, of course, not the only language with conversion. Noun to

verb conversion occurs frequently in German and Dutch as well, as the

examples in (37a–b) show, and verb to noun conversion is said to occur in

French , as the examples in (37c) show:

(37) a. German

Antwort ‘answer’ antwort-en ‘to answer’

Strick ‘cord, string’ strick-en ‘to knit’

b. Dutch

fiets ‘bicycle’ fiets-en ‘to bicycle’

hamer ‘hammer’ hamer-en ‘to hammer’

winkel ‘shop’ winkel-en ‘to shop’

compounds

attributive coordinative subordinative

endo exo endo exo endo exo

windmill

dog bed

air head

bird brain

producer-director

blue-green

truck driver

cost containment

pickpocket

cutpurse

parent–child FIGURE 3.3 Types of compounds

Lexeme formation: the familiar 57

c. French

gard-er ‘to guard’ garde ‘guard’

visit-er ‘to visit’ visite ‘visit’

There may appear to be a suffix added in the derivation of the verbs in

the examples in ( 37a–b ) and one deleted in (37c) . But the -en suffix in

German and Dutch and the -er suffix in French do not derive the verbs per

se – they are inflectional morphemes that signal the infinitive form of the

verb. If we assume that conversion involves only the base or root, these

examples count as conversion.

Morphologists have been divided on how to analyze conversion. Some

argue that conversion is just like affixation, except that the affix is pho-

nologically null – that is, it is unpronounced. When analyzed this way,

conversion is called zero affixation . It might be represented structurally

as in (38) .

(38)

chair ∅

Other morphologists argue that conversion is different from affixation,

and treat it simply as change of category with no accompanying change of

form, as we have done here. With this analysis, converted verbs like to chair

would not have any internal structure, but would simply be regarded as

having been relisted or recategorized in our mental lexicons. We will not

decide between these analyses here.

Challenge

Is it possible in English for already compounded or affixed words to

undergo conversion? Try to think of examples of words with prefixes

or suffixes or compound words that can function as more than one

part of speech (e.g., as both nouns and verbs).

3.6 Marvelous intricacies: how affixation, compounding, and conversion interact

You might think that in creating new words we might be content to take

a base or two and make a compound or add a prefix or suffix and leave it

at that. But native speakers of English (and of course other languages) are

remarkably exuberant in the way that they combine different types of

word formation strategies to create wonderfully complex words. Consider,

for example, the words in (39) .

(39) a. multiple suffixes : weaklingdom, ducklingish, institutionalize

b. multiple prefixes : hypermultilateralize, super-mega-omni-prayer,

micronanosecond

58 INTRODUCING MORPHOLOGY

c. multiple prefixes and suffixes : reinstitutionalization,

counterreformational

d. suffixes on compound words : bestsellerdom, basketballer, dotcomist,

girlfriendless

e. prefixes on compound words : ex-ballplayer, mega-earthquake,

unfootnoted

f. compound words that have undergone conversion : to blacklist, to

breakfast, a stir-fry,

Such words show us that we can form new words with several prefixes or

suffixes or a mixture of the two and that both simple bases and com-

pound bases may take prefixes and suffixes. And we can even take com-

pounds of one category and change them to another using conversion, as

the examples in (39f) suggest.

3.7 Minor processes

Affixation, compounding, and conversion are the most common ways of

forming new words, at least in English (we will see in Chapter 5 that there

are other means of word formation that languages other than English

use). In addition, there are a number of less common ways in which new

lexemes may be formed. We provide a survey of them here, without going

into great depth on any one of them.

3.7.1 Coinage It is of course possible to make up entirely new words from whole cloth, a

process called coinage . However, it turns out that we rarely coin complete-

ly new words, choosing instead to recycle bases and affixes into new com-

binations. New products are sometimes given coined names like Kodak,

Xerox , or Kleenex , and these in turn sometimes come to be used as common

nouns: kodak was at one time used for cameras in general, and xerox and

kleenex are still used respectively for copiers and facial tissue by some

American English speakers. But it’s relatively rare to coin new words. In

hundreds of new words archived on the Word Spy website ( www.wordspy.

com ), I was able to find only the following four apparent coinages:

(40) blivet ‘an intractable problem’

mung ‘to mess up, to change something so that it no longer

works’

grok ‘to understand in a deep and exhaustive manner’ (from

Robert Heinlein’s Stranger in a Strange Land )

mongo ‘objects retrieved from the garbage’

The verb snorgle , among the words with which we began this chapter, is

probably another example. Why are there so few coinages? Perhaps

because the words themselves give no clue to their meaning. Context often

clarifies what a word is intended to mean, but without a context to suggest

meaning, the words themselves are semantically opaque. It is no wonder

that many of the pure coinages that creep into English come from original

Lexeme formation: the familiar 59

product names: the association of the coined word with the product makes

its meaning clear, and occasionally the word will then be generalized to

any instance of that product, even if manufactured by a different company.

3.7.2 Backformation Generally, when we derive words we attach affixes to bases; in other words,

the base exists before the word derived by affixation. For example, we start

with the verb write and form the agent noun writer . Sometimes, however,

there are words that historically existed as monomorphemic bases, but

which ended in a sequence of sounds identical to or reminiscent of that of

certain affixes. When native speakers come to perceive these words as

being complex rather than simple, they create what is called a backforma-

tion . For example, historically the word burglar was monomorphemic. But

because its last syllable was phonologically identical to the agentive -er

suffix, some English speakers have understood it to be based on a verb to

burgle . Arguably for those speakers, then, burglar is no longer a simple

word. Similarly, the verb surveil has been created from surveillance and the

verb liaise from liaison . At least at first, some native speakers will find the

backformations odd-sounding or objectionable. Some years ago I heard the

then governor of Iowa, Tom Vilsack, use the verb incent on National Public

Radio; in context, it clearly was a backformation from the noun incentive ,

and it sounded quite odd at the time. But with time, that feeling of odd-

ness will disappear. Indeed speakers are sometimes surprised to learn that

the verb did not exist before the corresponding noun, so ordinary-sound-

ing has the verb come to be. Such is the case for peddle and edit , both of

which are historically backformations from peddler and editor respectively.

3.7.3 Blending Blending is a process of word formation in which parts of lexemes that

are not themselves morphemes are combined to form a new lexeme.

Familiar examples of blends (sometimes also called portmanteau words)

are words like brunch , a combination of breakfast and lunch , or smog , a com-

bination of smoke and fog. While not one of the major ways of forming new

words, blending is used quite a bit in English in advertizing, product-

naming, and playful language. The Word Spy website lists these blends:

(41) skitch ‘to propel oneself while on a skateboard or

in-line skates by hanging onto a moving

vehicle’ (combination of skate and hitch )

spime ‘a theoretical object that can be tracked

precisely in space and time over the lifetime of

the object’ (combination of space and time )

splog ‘a fake blog’ (combination of spam and blog )

vortal ‘a vertical portal’

bagonize ‘to wait anxiously for one’s bag to appear on

the carousel at the airport’ (combination of bag

and agonize )

Chrismukkah ‘a holiday celebration that combines elements

of Christmas and Hanukkah’

60 INTRODUCING MORPHOLOGY

Indeed, the sheer number of words of this sort that can be found in the

Word Spy archives suggests the vitality of this process. Note that while most

of the time the parts that are fused together to form blends are not them-

selves morphemes, sometimes a whole base or affix will be used; for exam-

ple, Word Spy also lists the word celeblog (‘a blog written by a celebrity’)

which is made up of the chunk celeb from celebrity and the word blog ; the

latter part has become a free morpheme in English in the last few years.

3.7.4 Acronyms and initialisms When the first letters of words that make up a name or a phrase are used

to create a new word, the results are called acronyms or initialisms .

In acronyms, the new word is pronounced as a word, rather than as a

series of letters. For example, A cquired I mmune D eficiency S yndrome

gives us AIDS , pronounced [eidz]. And s elf- c ontained u nderwater b reath-

ing a pparatus gives us scuba . Note in the case of scuba , the acronym has

become so familiar to English speakers that many do not know that it’s

an acronym! My favorite current acronym is the DUMP, a term univer-

sally used in Durham, New Hampshire, to refer to a local supermarket

with the unwittingly unfortunate name ‘the Du rham M arket P lace’.

Initialisms are similar to acronyms in that they are composed from the

first letters of a phrase, but unlike acronyms, they are pronounced as a

series of letters. So most people in the US refer to the F ederal B ureau of

I nvestigation as the FBI pronounced [εf bi ɑI]. Other initialisms are PTA for Parent Teacher Association, PR for either ‘public relations’ or ‘personal

record’, and NCAA for National College Athletic Association.

3.7.5 Clipping Clipping is a means of creating new words by shortening already existing

words. For example, we have info created from information , blog created

from web log , or fridge from refrigerator . Universities are fertile grounds for

the creation of clippings: students study psych, anthro, soc, and even ling

with one prof or another, and if they’re taking a science class, may spend

long hours in the lab, which might or might not involve running some

stats . Although clippings are often used in a colloquial rather than a for-

mal register, some have attained more neutral status. The word lab , for

example, is probably used far more frequently in the US than its longer

version laboratory . The word mob is a seventeenth-century clipping from

the Latin term mobile vulgus ‘the fickle common people’; the Latin phrase

has long been forgotten, but the clipping persists as the normal word for

an unruly throng of people.

3.8 How to: finding data for yourself

One of the most exciting parts of learning about morphology is that we

have all kinds of data at our fingertips, and with only a bit of training we

can begin to find these data and come up with hypotheses of our own. In

this section I will introduce you to a technique you can use to explore the

Lexeme formation: the familiar 61

morphology of English in more depth and begin to make discoveries of

your own.

Morphologists have always used their own intuitions to find data – for

example, trying to think of words that end in a particular prefix or suffix,

or trying to form particular kinds of compounds. But sometimes it’s hard

to think of pertinent examples and sometimes the examples we come up

with might sound funny or not quite right. We can of course go to diction-

aries to check whether those words occur, or to try to find other words

with the characteristics we’re searching for, but we already know that the

complex words we find in dictionaries are not necessarily representative

of the complex words we can form.

Challenge

Take five minutes and see how many words you can come up with

off the top of your head with the suffix - ity . Then take another

five minutes to page through a dictionary (you’ll need an old-

fashioned paper dictionary for this challenge!) and see how many

more - ity words you can come up with. How many words do you

have?

Chances are that no matter how many words you came up with, you

might have a nagging feeling that you don’t have enough words to really

see how the suffix - ity works. But how to get more data?

One classic way is to look in a backwards word list like Lehnert ( 1971 ).

A backwards word list gives words alphabetized starting with the last let-

ter, rather than the first. So all the words with - ity can be found together.

A few of the - ity words to be found in Lehnert ( 1971 : 584) are shown in the

sidebar. You’ll notice that using a backwards word list is not a perfect

tool: such word lists simply alphabetize words from the end to the begin-

ning, so any word ending in the letters ity will occur in the list, not just

words that really have the sufix - ity . In the list I give here, in addition to

real - ity words like oddity and rancidity , we find ‘junk’ like rumti-iddity

(spelled two different ways!); a bit earlier in the list we would have found

the word city which of course also ends in the sequence of letters ity but

certainly doesn’t contain a suffix. So if you work with a backwards word

list, be prepared to go through it word by word and check whether you

really have an example of the suffix you’re looking for.

The backwards word list is sometimes an improvement over picking our

own brains, but these days we have a much better resource available to us in

the form of vast databases of spoken and written language called corpora

(singular corpus ) that can be searched in quite sophisticated ways. For

English, several corpora are freely available on the internet, and with just a

bit of training you can use them to find all kinds of nifty data on contempo-

rary English word formation. The largest is the Corpus of Contemporary

American English , often referred to by its acronym COCA (Davies 2008 ) with

(as of 2015) 450 million words of American English from 1990 to 2012, taken

paucity

raucity

caducity

rumti-iddity

rumpt-iddity

quiddity

oddity

heredity

rabidity

morbidity

turbidity

acidity

subacidity

placidity

nonacidity

hypoacidity

hyperacidity

f laccidity

rancidity

62 INTRODUCING MORPHOLOGY

FIGURE 3.4 COCA website interface

from spoken language, newspapers, magazines, academic journals, and fic-

tion. The second is the British National Corpus or BNC, which contains 100

million words of British English from 1980 to 1994, again with both spoken

and written components. And if you’re interested in the development of

English over the last two centuries, you can look at the Corpus of Historical

American English (COHA). All three of these can be found at http://corpus.byu.

edu/ . One of the things that makes corpus data much more useful and inter-

esting than data from a backwards word list is that it’s possible to see words

in context, so we get a clue as to what they mean and how they were used.

Here’s how to use them to find examples of prefixed or suffixed words.

We’ll use COCA for this demonstration. Once you’ve found the COCA website

and entered it, you’ll see a user interface that looks the one in Figure 3.4 .

To find words with a particular prefix or suffix, you’ll need to do

what’s called a wildcard search. In a wildcard search, you use the asterisk

symbol (*) to mean ‘anything at all’. For example, if you want to search for

words with the suffix - esque , you would enter * esque in the search box,

which means anything at all ending in the sequence of letters esque . For

a prefix like mega - you would enter mega * to mean anything at all begin-

ning with the sequence of letters mega , and so on. If you hit the search

button now using the default settings, the corpus will give you the 100

examples ending in that sequence of letters that occur most frequently

in the corpus. This may give you a lot to work with, but for reasons that

we will learn in the next chapter, morphologists are often interested in

the examples that occur the least frequently rather than those that occur

most frequently. So it’s good to learn how to fine-tune your search.

To find more examples and less frequently occurring ones, you will have

to change the default settings. First click on the line at the bottom left

that reads ‘Click to see options’, and then change the number in the box

labeled ‘frequency’. You can set this for a much higher number than the

default 100; I usually set this at 6,000, so unless there are a really huge

number of ‘hits’, you should get all possible words with the prefix or suf-

fix you’re looking at. If you’d prefer to get your results in alphabetical

order rather than in order of frequency, you can click on the box labeled

‘sorting’ and click on ‘alphabetical’.

Lexeme formation: the familiar 63

Now you can search. But that’s not the end of the story. Unfortunately,

what you will get when you search for almost any affix is a long list that

needs a lot of cleaning up. There will be misspelled words, items with

funky punctuation, and lots of items that just happen to end or begin

with the same sequence as the affix you’re searching for, but don’t really

have that affix (as was the case in our backwards word list). So you’ll need

to get rid of a lot of junk before you’re left with only data on the affix your

studying. To do the cleaning, you can copy your results to a spreadsheet,

or just copy down the examples that have the affix that you’re studying,

if you want a low-tech method.

Now you get to look at your data. The payoff of what may seem like a lot

of grunt work is that for most English affixes you’ll come up with many

more examples than you might have thought of yourself, and some of

them will undoubtedly be quite interesting and surprising. And from here

on, you get to hatch your own theories! Note that if you see an interesting

example and you want to see the context in which it’s used, you can click

on the word. An excerpt of the text in which the word is found will appear

at the bottom of your screen. If you want to see a bit more of the context,

click on the name of the source, which can be found to the left of the

excerpt. Looking at the context will help you figure out the meaning of

the word and its part of speech, all of which will help you in formulating

word formation rules like the ones we worked on in section 3.3.1 .

A couple of final notes. First, if you want to use one of these corpora

more than a few times, you’ll need to register. This is free, and takes just

a couple of minutes. Second, the same kind of work can be done on lan-

guages other than English. Corpora exist for many languages these days.

However, they are not always freely available, and the user interface will

differ from one to the next.

Summary In this chapter we have looked at a number of ways in which new words may be formed in languages. Affixed words are formed by word

formation rules that make explicit the categorial, semantic, and

phonological requirements of particular affixes, and specify the

categorial, semantic, and phonological properties of the resulting

words. Words formed by affixation have internal structure that may be

represented in the form of trees. Similarly, compound words – words

composed of two or more free morphemes or bound bases – have

internal structure that can be represented in trees. Compounds may be

classified according to the parts of speech of the elements that make

them up, or according to the semantic relations they display

(attributive, coordinative, or subordinative) or according to their

headedness. We have also looked at conversion, a shift in the category

of a lexeme with no accompanying change in form, as well as a

number of forms of word formation – coinage, blending, clipping,

backformation, acronyms and initialisms, that play a minor role, at

least in English. Finally, we have learned some techniques for searching

for our own morphological data.

64 INTRODUCING MORPHOLOGY

Exercises 1. Divide the following words into morphemes and label each morpheme

as a prefix, suffix, free base, or bound base.

hypoallergenic non-morphological telephonic overcompensation reheatability monomaniacal

2. On p.41 we gave the word formation rules for -ize. Now consider the words below and discuss what other sorts of restrictions we would have to add to our rules for -ize.

catechize, evangelize, antagonize, metabolize, epitomize

3. Using the data below, try to write a word formation rule for the suffix -able. Consider what category it attaches to, and what part of speech the resulting words belong to. Does it seem to have any phonological or semantic restrictions? Then draw the word trees for the words unwashable and rewashable.

washable *yawnable dryable *arriveable heatable *fallable readable *blinkable loveable knowable

4. The word unwindable is potentially ambiguous. What are its two possible meanings? Draw two tree structures and show which meaning goes with each structure.

5. The linguist Laurence Horn has argued (2002) that the prefix un- really does attach to nouns, contrary to what we said in section 3.3 . He has collected such examples as undeath, uncountry, uncopier, unphilosophy, and unpublicity. Can you think of or find other examples where un- has attached to nouns? What do you think these un- nouns mean? (You can use a dictionary to help think of examples.)

6. In section 3.3 , we discussed the meanings (or lack thereof) of bases like -ceive, -mit, and -port, but not the meanings of the prefixes with which they combine. Consider the prefixes re- and de- in words like report, deport, receive, deceive, remit, and demit. Do these seem to be the same prefixes as the re- and de- in rewash, rewind, reload or debug, de-ice, derail ? Why or why not?

7. How many meanings can you come up with for the complex compound miniature poodle groomer manual ? Try to draw the trees that correspond to each meaning you’ve come up with.

8. Classify the compounds below as attributive, coordinative, or subordinative, and as either endocentric or exocentric. Example: book shelf is an endocentric attributive compound; truck driver is an endocentric subordinate compound.

Lexeme formation: the familiar 65

oil burner lighthouse blue blood hell raiser scholar athlete blue-eyed pickpocket house-hunting

9. Many languages use compounding as a strategy for forming new words. Consider the data below and try to determine: (a) which element is the head, (b) whether the resulting compounds are endocentric or exocentric.

a. Kannada (Dravidian) (Sridhar 1990 ) a: Du-ma:tu ‘speak word’ ‘colloquial speech’ siDi-maddu ‘explode chemical’ ‘explosive (i.e., chemical

that explodes)’

maduve a:gu ‘marriage become’ ‘to get married’ santo:Sa paDu ‘happiness feel’ ‘rejoice’ kittaLe haNNu ‘orange fruit’ ‘tangerine’

b. Maori (Polynesian) (Bauer 1993 ) ipu para ‘container waste’ ‘rubbish bin’ apuru teepu ‘cushion table’ ‘desk pad’ wai mangu ‘water black’ ‘ink’ whaka-koi pene ‘cause.sharp pen’ ‘pencil sharpener’

10. Consider the following noun/verb conversion pairs in English. In each case decide whether the noun was converted from the verb or vice versa. Give arguments based on meaning to support your choices.

bug to bug kick to kick saddle to saddle howl to howl yawn to yawn book to book (e.g., a table in a restaurant)

11. Take a look at the words you (and your classmates) have collected so far in your Word Logs. Can you classify them according to the means of word formation used to create them? Does any one means of word formation predominate? If so, think about why this might be.

12. Following the directions in section 3.8 , use COCA to search for the suffix - eer (as in auctioneer or mountaineer ). Make a list or a spreadsheet of the data that you find, and use those data to formulate a word formation rule for - eer .

13. In this chapter, we used tarian , licious , and tastic as examples of splinters. See if you can think of other examples of splinters in English and search for novel words using those splinters in COCA.

14. Consider the semantic categories of affixes discussed in section 3.3.3 . First look at the data below from the Yuman language Hualapai , spoken in northern Arizona. Assume that a process of affixation creates the pairs

66 INTRODUCING MORPHOLOGY

of verbs in this language. What semantic category would you place these affixes in? (Data from Watahomigie et al. 1982 : 280.)

d-akk ‘to throw toward the speaker’ d-amk ‘to throw from the speaker’

e:kk ‘to give/receive (toward me)’ e:mk ‘to send’ ha:kk ‘to look this way’ ha:mk ‘to look over that way’ jiyu:kk ‘to send one person/animal

toward the speaker’ jiyu:mk ‘to send one person/

animal away’ ‘u:kk ‘to come and see’ ‘u:mk ‘to go and see’ vo:kk ‘to come home’ vo:mk ‘to go home’

KEY TERMS productivity

transparency

lexicalization

compositionality

frequency

creativity

hapax legomenon

CHAPTER OUTLINE

In this chapter you will learn about productivity – the extent

to which word formation rules can give rise to new words.

♦ We will consider what factors contribute to productivity,

what restricts the productivity of word formation

processes, and how we can measure productivity.

♦ We will look at how the productivity of a word formation

process can change over time.

♦ And we will consider how speakers of a language can

use even unproductive word formation processes to

create new words for humorous or playful effects.

4 C H A P T E R

Productivity and creativity

68 INTRODUCING MORPHOLOGY

4.1 Introduction

Consider the examples in (1) :

(1) a. warm warmth

true truth

b. modern modernity

pure purity

c. happy happiness

dark darkness

In each case, we have adjectives and nouns that are derived from them (all

cases of transposition, by the way). As a first pass, we might hypothesize

the three rules of lexeme formation in (2) :

(2) a. Rule for -th : -th attaches to adjectives, and creates nouns. For a

base meaning ‘X’, the derived noun means ‘the state of being X’.

b. Rule for -ity : -ity attaches to adjectives, and creates nouns. For a

base meaning ‘X’, the derived noun means ‘the state of being X’.

c. Rule for -ness : -ness attaches to adjectives, and creates nouns. For a

base meaning ‘X’, the derived noun means ‘the state of being X’.

Now consider the list of adjectives in (3) . If you had to make a noun from

each of these, which of the three suffixes would you choose (note that you

might be able to use more than one in some cases)?

(3) lovely

cool

crude

evil

googleable

rustic

musty

inconsequential

feline

toxic

bovine

Chances are that there are some of these words that you would choose to

use -ity with (I choose crude, toxic, googleable, rustic, inconsequential , maybe

feline ), and others that you would use -ness with (for me, lovely, cool, evil,

musty, probably bovine ). Your choices might be slightly different from

mine, but I’d be willing to predict that you didn’t choose to use -th with

any of these adjectives.

What does this mean? In some cases, we can look at words, decide that

they are complex, and isolate particular affixes. But when it comes to

using those affixes to create new lexemes, we have the sense that they

are no longer part of our active repertoire for forming new words. We

have no trouble using other affixes, however, even if we’ve never seen

them on particular bases; for example, you may never have seen a noun

Productivity and creativity 69

form of the word bovine , but you have no trouble forming the word

bovineness (or maybe bovinity , or maybe even both). Processes of lexeme

formation that can be used by native speakers to form new lexemes are

called productive . Those that can no longer be used by native speakers

are unproductive; so although we might recognize the -th in warmth as a

suffix, we never make use of it in making new words. The suffixes -ity

and -ness , on the other hand, can still be used, although perhaps not to

the same degree. Most morphologists agree that productivity is not an

all-or-nothing matter. Some processes of lexeme formation, like affixa-

tion of -th , are truly unproductive, but for those processes that are pro-

ductive, we have the sense that some are more productive than others.

In this chapter we will explore in some detail what we mean by produc-

tivity, and look at a number of factors that contribute to productivity.

We will also look at several ways in which productivity can be measured.

4.2 Factors contributing to productivity

A number of factors contribute to the degree to which we can use morpho-

logical processes to create new lexemes (see Figure 4.1 ).

transparency

frequency

of base

productivity

usefulness

FIGURE 4.1 Factors contributing to productivity

One factor is what is called transparency . Words formed with transpar-

ent processes can be easily segmented, such that there is a one-to-one

correspondence between form and meaning. In other words, when we

attach an affix to a base, the phonological form (the pronunciation) of

both morphemes stays the same, and the meaning of the derived word is

exactly what we would expect by adding the meaning of the affix to that

of the base. Let’s look further at the case of -ness and -ity , this time consid-

ering the additional examples in (4) :

70 INTRODUCING MORPHOLOGY

(4) a. candid candidness

pink pinkness

hardy hardiness

common commonness

ticklish ticklishness

cunning cunningness

horrible horribleness

pure pureness

odd oddness

b. crude crudity

odd oddity

pure purity

dense density

rustic rusticity

timid timidity

grammatical grammaticality

local locality

available availability

senile senility

In all the -ness examples in (4a) , it is easy to divide the complex words into

base and suffix. The base is always pronounced in the derived word as it

is in isolation. And the suffix always creates a noun meaning ‘state of

being “adjective”’, whatever the adjective. Words formed with -ness are

perfectly transparent. The suffix -ity is somewhat less transparent.

Although you don’t see this when words are written in English orthogra-

phy, when you pronounce them, you see that -ity often has the effect of

changing the phonological form of its base – sometimes its stress pattern,

and sometimes both stress and phonological segments in the base. So

timid in isolation is pronounced with stress on the first syllable ( tí mid ), but

when -ity is added, stress shifts to its second syllable ( timí dity ). And with

the base rustic , in addition to a shift in stress from first to second syllable,

the final [k] of the base becomes [s] when -ity is added. Further, some of the

words formed with -ity have meanings that cannot be arrived at by com-

bining the meaning of the base with that of the suffix. An oddity , for

example, is not merely ‘the state of being odd’ (we would probably prefer

the word oddness for that meaning), but a person or thing that is odd. And

a locality is not ‘the state of being local’, but a place or area. Finally, con-

sider the examples in (5) :

(5) verity

dexterity

authority

In the first two examples, -ity occurs on bound bases ver, dexter . In the

third, it’s not clear exactly how to analyze the derived word. Although it

appears to be a combination of author and -ity , there are two problems

with this analysis. First, as a free base author is a noun, and -ity typically

attaches to adjectives, rather than nouns. And second, it’s not clear what

Productivity and creativity 71

the independent meaning of the base is; certainly the meaning ‘profes-

sional writer’ does not seem to be part of the meaning of authority . We

never find -ness , however, on bound bases. All of this adds up to a conclu-

sion that the suffixation of -ness is a much more transparent process than

the suffixation of -ity , and this in turn suggests that -ness is a more produc-

tive affix than -ity .

Hand in hand with the notion of transparency comes the related notion

of lexicalization . When derived words take on meanings that are not

transparent – that cannot be made up of the sum of their parts – we say

that the meaning of the word has become lexicalized . Meanings of com-

plex words that are predictable as the sum of their parts are said to be

compositional . Lexicalized words have meanings that are non-composi-

tional . So the words oddity and locality that we looked at above have devel-

oped lexicalized or non-compositional meanings. Sometimes the mean-

ings of derived words have drifted so far from their compositional mean-

ings that it’s quite difficult to imagine the compositional meaning for

them. Consider, for example, the word transmission , which denotes a part

of a car. It takes a bit of thought to realize that the car part in question is

so called because it transmits power from the engine to the wheels.

Transparency is not the only factor that contributes to productivity.

Another factor that is important is what we might call frequency of base

type . By this, I mean the number of different bases that might be available

for affixes to attach to, thus resulting in new words. If an affix attaches

only to a limited range of bases, it has less possibility of giving rise to lots

of new words, and it will therefore be less productive. Consider, for exam-

ple, the suffix -esque in English, which means something like ‘having the

style of ’ (Marchand 1969 : 286).

Challenge

Do a COCA search for the suffix - esque following the instructions in

section 3.8 . What kinds of bases does - esque attach to? Look at the

category (part of speech) of the bases, the semantic types of those

bases, and the phonological properties of those bases.

You’ll probably find that - esque attaches to nouns, but mostly to concrete

ones ( mountainesque ), and in fact, most often to proper names ( Kafkaesque,

Reaganesque ). Indeed, although it attaches pretty freely to names, it seems

most comfortable on names that have at least two syllables ( ?Bushesque,

?Blairesque ). Compared to a suffix that could attach to any noun at all,

-esque would be less productive.

The final factor that contributes to productivity is what we might call

usefulness . A process of lexeme formation is useful to the extent that

speakers of a language need new words of a particular sort. It’s always

useful, for example, to be able to form a noun meaning ‘the state of being

X’ from an adjective, whatever X means, so both -ness and -ity are highly

useful affixes. On the other hand, consider the suffix -ess in English. It

72 INTRODUCING MORPHOLOGY

used to be useful to be able to coin words referring to jobs performed by

women or positions held by women ( stewardess, murderess, authoress ). But

with the rise of feminism and efforts to promote gender-neutral language,

such words have fallen into disuse, and the need for new words using this

suffix has almost died out. Consequently the affix has become far less

productive.

4.3 Restrictions on productivity

As we saw above, the more limitations there are on the bases available to

a lexeme formation process, the less productive it will be. In this section,

we will explore different kinds of restrictions that may apply to lexeme

formation processes.

We have actually looked at some such restrictions in Chapter 3 (sec-

tion 3.2), when we learned how to write word formation rules. We learned

that there could be different sorts of restrictions on what sorts of base an

affix might attach to, including:

• categorial restrictions: Almost all affixes are restricted to bases of specific categories. For example, -ity attaches to adjectives, -ize

attaches to nouns and adjectives, or un- attaches to adjectives or

verbs.

• phonological restrictions: Sometimes affixes will attach only to bases that fit certain phonological patterns. For example, -ize prefers nouns

and adjectives that consist of two or more syllables, where the final

syllable does not bear primary stress. The suffix -en, which forms

verbs from adjectives, attaches only to bases that end in obstruents

(stops, fricatives, and affricates). So we can get darken, brighten, and

deafen but *slimmen and *tallen, which end in sonorant consonants, are

impossible.

• the meaning of the base: For example, negative un- prefers bases that are not themselves negative in meaning. We find unlovely but not *unugly,

unhappy but not *unsad .

To these sorts of restrictions we might add:

• etymological restrictions: Some affixes are restricted to particular subclasses of bases. For example, there are affixes in English that

prefer to attach to bases that are native – for example the suffix -en

that forms adjectives from nouns ( wooden, waxen but not * metalen or

*carbonen ). On the other hand, another suffix -ic that forms adjectives

from nouns ( parasitic, dramatic ) will not attach to native bases, only to

bases that are borrowed into English from French or Latin.

• syntactic restrictions: Sometimes affixes are sensitive to syntactic properties of their bases. For example, the suffix -able generally

prefers to attach to transitive verbs (verbs that have a subject and an

object), specifically verbs that can be passivized. So from the

transitive verb love we can get loveable, but from the intransitive verb

snore there is no *snorable .

Productivity and creativity 73

• pragmatic restrictions: Bauer ( 2001 : 135) gives the following example. In Dyirbal, there is a suffix -ginay that means ‘covered with’. Although

there might conceivably be a use for a word meaning something like

‘covered with honey’, in fact, the suffix occurs in Dyirbal only on

bases that denote things that are “dirty or unpleasant” (Dixon 1972 :

223), like gunaginay, which means ‘covered with feces’. What’s

considered dirty or unpleasant might to some extent be a function of

cultural beliefs.

We might expect there to be an inverse correlation between the number

of restrictions and the productivity of a lexeme formation process: the

more restrictions apply, the fewer bases it will have available to it, and the

fewer words it will be able to derive.

The restrictions above pertain to inputs to lexeme formation rules. But

it’s also possible for there to be restrictions specifically on the output of

rules. For example, certain sorts of complex words can be restricted in

“jocular and often slightly pejorative personal names.” For example,

from the adjective bang ‘afraid’ we get bangerd ‘fraidy-cat’ and from dik

‘fat’ we get dikkerd ‘fatty’. Baayen points out that although there are a lot

of adjectives that might give rise to pejorative names for people, words

formed with the suffix are confined to use in spoken, as opposed to writ-

ten, language and therefore the output of this lexeme formation process

is restricted.

4.4 Ways of measuring productivity

We have seen that the productivity of lexeme formation processes

depends on a variety of factors, including restrictions on possible bases,

usefulness of the words formed, and the transparency of the process.

Looking at these factors can give us some sense of how productive a pro-

cess might be, but can we do better and actually measure productivity? Is

it possible to compare the productivity of different processes? If so, how

might we go about making such measurements?

Challenge

One conceivable way of measuring the productivity of a lexeme

formation process might be to count up all the items formed with

that process that can be found in a good dictionary. Most

morphologists think that this is not a good way of measuring

productivity. Think of as many reasons as you can why they should

think so.

It’s not hard to think of reasons why counting items in a dictionary

wouldn’t be an accurate way of estimating productivity. For one thing,

counting items that are already in the dictionary doesn’t really tell us

anything about how many new words might be created with a lexeme

74 INTRODUCING MORPHOLOGY

formation process, and it’s the possibility of creating new forms that’s

most important in making processes productive. Further, the most pro-

ductive of lexeme formation processes are ones that are phonologically

and semantically transparent. If the words resulting from these processes

are perfectly transparent in meaning, then it’s unlikely that dictionaries

will need to record them! On the other hand, less productive processes, as

we’ve seen, frequently have outputs that are less transparent (more lexi-

calized), and therefore have more need to be listed in the dictionary. So

simple counting might give a paradoxical result: less productive processes

would be represented by more entries in the dictionary than more produc-

tive processes!

Morphologists have therefore tried hard to come up with other ways of

measuring productivity. One suggestion (Aronoff 1976 ) was to make a

ratio of the number of actual words formed with an affix to the number

of bases to which that affix could potentially attach.

Challenge

Consider the suffix -esque that we mentioned in section 4.2 above.

Based on the data that you gathered yourself, do you see any problem

in counting the number of bases to which - esque could potentially

attach?

Most morphologists would say that we might be able to count the words

we find in a given dictionary with a particular affix, but would this num-

ber be equivalent to the number of words in any speaker’s mental lexicon?

How could we even guess at this? And further, how do we count potential

bases? In the case of - esque , this would require knowing how many names

the suffix might attach to, which is, of course, not possible!

A somewhat more sophisticated – but still not perfect – measure of

productivity proposed by Baayen ( 1989 ) capitalizes on what we know

about the token frequency of derived words. Remember from Chapter 1

the distinction between types and tokens: if we’re counting types in a

corpus or language sample we look for each different word and count it

once, no matter how many times it appears, but if we’re looking at tokens

we count up all the separate occurrences of that word in a particular cor-

pus. The number of separate occurrences of a word in the corpus is the

token frequency of that word.

An important observation that has been made about lexeme formation

processes is that the less productive they are, the less transparent the

words formed by those processes, and the less transparent the words, the

higher their mean token frequency in a corpus. In other words, words

formed with less productive suffixes are often more lexicalized in mean-

ing and will often display many tokens in a corpus. The more productive

a process is, the more new words it will give rise to and the more chance

that these items will occur in a corpus with a very low token frequency,

sometimes only once. Remember that in section 3.8 I suggested that we

Productivity and creativity 75

might be more interested in derived words that occur with a very low

frequency, rather than ones that occur with a high frequency in a corpus?

This is why: we can make use of this observation by counting up all tokens

of all words formed with a particular affix, and then seeing how many of

those words occur only once in the corpus (a type with token frequency of

one in a corpus is called a hapax legomenon or sometimes just a hapax ).

The ratio of hapaxes to all tokens tells us something about the probability

of finding new forms using that affix: the higher the probability of low

frequency items, the likelier those items are to be new formations. Baayen

( 1989 ) gives us this formula:

(6) Baayen’s productivity formula

P = n 1

where N equals the total number of tokens and n1 the number of hapaxes

For example, using my own data from COCA, I count 5,738 total tokens

for words using the suffix - esque , of which 665 types occur only once. This

gives a P score of .12 for - esque . To know exactly what this number means,

we need to look at it relative to other suffixes in contemporary English.

Take, for example, the suffix - dom , which attaches primarily to nouns and

forms nouns meaning ‘the state or quality of Noun’, as in chiefdom , fandom ,

and stardom . Again, using my own data from COCA, I count 64,345 total

tokens and 257 hapaxes. That gives a P score of .004. You can see that by

comparison, - esque appears to be far more productive than - dom .

4.5 Historical changes in productivity

It should not come as a surprise at this point that lexeme formation pro-

cesses may change their degree of productivity over time . Particular pre-

fixes or suffixes can come into vogue and flourish for a few decades or

centuries and then fade out of existence. The words formed with them

may persist for a long time, but we won’t find any new ones being formed.

In this section we will see how it’s now possible to track the productivity

of affixes in English.

How can we measure changes in productivity? Until recently, the best

way that we had to gauge the historical productivity of a prefix or suffix

was by using dictionaries. For example, the Oxford English Dictionary gives

us information about when particular words are cited for the very first

time in English, and from looking at those first citations we can see some-

thing about how many new words have been formed in a given century

with a particular affix. If we know how many first citations there are for

words with a given affix in a specific century and if we also know the total

number of citations there are in the OED for each century (not every cen-

tury has the same number of citations), we can calculate what percentage

of them are first citations with that affix.

Let’s see how this works with the suffix -dom . For example, if the OED

gives 28,698 citations dating from the thirteenth century, and seven of

them are the first citations of words with - dom , then the -dom citations

76 INTRODUCING MORPHOLOGY

1. Figure 4.2 starts with the twelfth century because that is the fi rst century for which we have the number of

citations in the OED. Many thanks to Charlotte Brewer of Oxford University for sharing these data with me.

2. The careful reader will note that the P score for - dom in the last two decades of the twentieth century

looks higher than the fi gure I calculated for - dom using COCA above. The reason for the discrepancy is that

COHA uses a different sample of texts than COCA. The overall number of words per decade is smaller, and

all excerpts are from written English, so fi nding a small discrepancy in P fi gures should not be especially

troubling.

represent 0.0243% of all those citations. I’ve calculated these percentages

for each century, and then plotted them on a graph, as you can see in

Figure 4.2 .

The suffix -dom is a very old one, going back to the beginnings of

English, and indeed further back into the Germanic branch of the Indo-

European family, from which English descends. 1 We can see that after an

initially very productive period in the twelfth century, -dom seems to have

dropped off in productivity from the fourteenth through the eighteenth

centuries. But its productivity rises again in the nineteenth century. These

days, if we don’t want to trace the productivity of an affix over its whole

history, but only over the last two or so centuries, we can make use of

Baayen’s P formula. To do this, we can use another corpus, called the

Corpus of Historical American English (COHA) . This corpus can be found

at the same website as COCA and BNC, and it uses the same user interface

as those corpora. But COHA breaks the data down by decade, starting with

1810 and ending with 2000. As we did with COCA, we can use COHA to see

how many tokens overall there are with a particular affix and how many

of those tokens are hapaxes. And from these two numbers you can calcu-

late the P score for that affix decade by decade. As you can see from

Figure 4.3 , the P score for - dom generally rises from the beginning of the

nineteenth century to the end of the twentieth century, although it’s not

a completely smooth rise – there’s a mysterious dip in the 1860s. We can

only wonder about the decline in the popularity of the suffix during the

decade from 1860 to 1870! 2

Exactly why the productivity of this suffix should start to rise

after centuries of minimal productivity is unclear. Wentworth ( 1941 )

notices the same trend that I’ve shown here, and points out that particular

FIGURE 4.2 Comparative percentages of first citations of - dom per century

0.01

0.02

0.03

0.04

0.05

0.06

12th 13th 14th 15th 16th 17th 18th 19th

Productivity and creativity 77

nineteenth-century authors seem especially prone to coin new words with

the suffix: Thomas Carlyle and William Makepeace Thackeray in Britain,

Mark Twain and Sinclair Lewis in the United States. But whether they are

the cause of the rise or a reflection of something that was happening

in the language at large is impossible to say. Bauer ( 2001 ) points out that

in the nineteenth century, the kind of bases available to the suffix -dom

seemed to expand drastically. Where it was confined for many centuries

to bases referring to important types of people ( lord, king, master, pope, earl,

but also martyr and witch ) or a few adjectives ( wise, rich, free ), in the nine-

teenth century it began to appear with more frequency on names for

animals ( puppy, dog, butterf ly, centaur ) and a wide variety of common nouns

( school, twaddle, leaf, magazine, jelly, cotton, fossil ). But why, exactly, its range

of bases expanded at this point is still a mystery.

Note that we can also use COHA to compare the relative productivity of

affixes over the last two centuries. Figure 4.4 illustrates this with the suf-

fixes - dom and - esque . Remember that we calculated P for both suffixes on

the basis of the contemporary data from COCA and found that - esque

appeared to be much more productive than - dom . The data from COHA

FIGURE 4.3 Productivity of - dom using Baayen’s P measure and data from COHA

18 10

18 20

18 30

18 40

18 50

18 60

18 70

18 80

18 90

19 00

19 10

19 20

19 30

19 40

19 50

19 60

19 70

19 80

19 90

20 00

0.001

0.002

0.003

0.004

0.005

0.006

0.007 Productivity of -dom

FIGURE 4.4 Comparative productivity of - esque and - dom using data from COHA

1 8 1 0

1 8 2 0

1 8 3 0

1 8 4 0

1 8 5 0

1 8 6 0

1 8 7 0

1 8 8 0

1 8 9 0

1 9 0 0

1 9 1 0

1 9 2 0

1 9 3 0

1 9 4 0

1 9 5 0

1 9 6 0

1 9 7 0

1 9 8 0

1 9 9 0

2 0 0 0

0.05

0.1

0.15

0.2

0.25

0.3

comparing -esque and -dom

dom

esque

78 INTRODUCING MORPHOLOGY

enriches our understanding of the relative productivity of these two suf-

fixes. Consider Figure 4.4 .

You can see from the graph that compared to - dom , the suffix - esque not

only is much more productive overall, but that the trajectory of its produc-

tivity is different. The suffix - esque seems to have taken off in productivity

starting in the 1940s, and has risen dramatically since 1980, whereas the

rise in productivity of - dom is comparatively modest.

Challenge

Try your hand at tracing the historical productivity of the prefixes

mini - and nano - in English. Can you pinpoint when each one started to

gain in productivity?

4.6 Productivity versus creativity

Some morphologists make a distinction between morphological produc-

tivity and morphological creativity . When processes of lexeme formation

are truly productive, we use them to create new words without noticing

that we do so. Similarly, when hearers are exposed to a productively

formed complex word, they understand it, but usually don’t note that it’s

a new word (at least for them). This is not to say that speakers and hearers

never notice productively formed new words, just that often such words

slip by without notice. Morphological creativity, in contrast, is the

domain of unproductive processes like suffixation of -th or marginal

lexeme formation processes like blending or backformation . It occurs

when speakers use such processes consciously to form new words, often

to be humorous or playful or to draw attention to those words for other

reasons.

For example, speakers might use the unproductive suffix -th to form an

adjective like coolth (in contrast to warmth ), consciously trying to be clever

or witty. Another example might be the suffix -some that occurs in English

in words like twosome, threesome, and foursome . Theoretically this suffix

might be infinitely productive because its bases are cardinal numbers. But

it’s really only attached to the numbers two through four or five . We would

probably only coin a new word like seventeensome if we were trying to be

funny. Such a use would be creative, rather than a productive use of this

lexeme formation process.

Let’s now look more closely at the case of blending in English. Blending,

as we saw in Chapter 3 , is the creation of new words by putting together

parts of words that are not themselves morphemes. Relatively few blended

words have become lexicalized words in English ( brunch, smog ), but the

technique is frequently used for coining words by advertizers and the

media, precisely because such words are noticeable. McDonald’s, for

example, creates a word like menunaire from menu and millionaire to catch

your eye (or ear), and make you pay attention to their pitch.

Productivity and creativity 79

Websites that track new words often have a disproportionate number of

blends, and most of those words are culled from the popular press. For

example, in the new words posted on the Word Spy website ( www.wordspy.

com ) in April and May of 2014, there were five blends:

(7) toebesity blend of toe and obesity. ‘Excessive fat on one or

more toes’.

listicle blend of list and article. ‘Article or news story that

consists primarily of a thematic list of short items’.

clicktivism blend of click and activism. ‘Activism that consists

solely of performing quick and simple online tasks’.

paralpinism blend of paraglider and alpinism. ‘A sport that

involves climbing a mountain and then descending

using a paraglider or similar craft’.

practocalypse blend of practice and apocalypse. ‘A drill that simu-

lates a disastrous situation, particularly a shutdown

of the internet’.

All five of these words were found in popular media – newspapers such

as the Washington Post, magazines such as the New Yorker , and online media

outlets such as Slate and the Huffington Post . All were intended to catch the

reader’s eye and therefore make for lively reading, and we might deem

them successful because they found their way to Word Spy. In contrast,

websites like Word Spy don’t pick up other new words with -ness . Word-

spotters are far less likely to notice new forms that come from truly pro-

ductive lexeme formation processes than new blends or the sporadic cre-

ative coinages that still come from unproductive processes.

As Bauer ( 2001 ) points out, however, it is not always possible to draw a

sharp line between productivity and creativity. Take, for example, the

diminutive suffix -let in English ( booklet, wavelet, eyelet ). A look at the Oxford

English Dictionary and COHA suggests that this suffix enjoyed at best a small

degree of productivity during the nineteenth and twentieth centuries.

COHA gives us a handful of apparently novel forms, that is, ones that occur

only once in the corpus and are not captured in the OED: peaklet, gravelet,

adlet, auntlet, boatlet, commissionlet, cliff let, cluster-bomblet, choplet, czarlet, dicta-

torlet, f lowerlet, f lylet, footnotelet, faunlet, dudelet, statelet, porchlet, sniff let, squab-

let, splotchlet, sky-let, shoelet . Most of these forms are used in contexts which

suggest not just small size, but also disparagement and often irony. To

these I can add a further example that comes from a biography of Julia

Child (great TV chef and cookbook author), in which her husband refers to

her as his wifelet – surely meant to be funny, as Julia was over six feet tall! 3

This last example was clearly meant to draw attention to itself, but are all

these - let formations self-conscious enough that they should be seen as

creative formations, or rather is - let just a minimally productive suffix

that’s taken on a special ironic or pejorative flavor in English? Perhaps

there is no good way of answering this question, but looking at examples

in the corpora gives us a new way of considering the issue for ourselves.

3. Laura Shapiro, Julia Child . Viking, 2007.

80 INTRODUCING MORPHOLOGY

Summary In this chapter we have explored the notion of productivity – the extent to which lexeme formation processes can be used to create new

words. We have seen that several factors contribute to productivity:

the phonological and semantic transparency of the process, the size of

the pool of bases it can apply to, and its usefulness. We have seen as

well that there can be different sorts of restrictions on lexeme

formation processes that result in a decrease in productivity. Among

these are categorial, phonological, semantic, syntactic, etymological,

and pragmatic restrictions. We have looked as well at ways in which

we can measure productivity. Finally, we have seen that even

unproductive and marginal processes can still give rise to occasional

new formations, a phenomenon that we called creativity, but that

discerning the line between productivity and creativity may be a

subjective matter.

Exercises 1. Which of the following derived words with the suffix -ity have lexicalized

(non-compositional) meanings. Hint: some have both. Fill in the grid below:

Compositional?

Yes/No

Compositional

meaning

Non-compositional

meaning

a. curiosity

b. solidity

c. publicity

d. sexuality

e. visibility

f. facility

2. Consider the examples in (a)–(c) below. Each set involves a lexeme formation process that takes nouns as base and produces adjectives. On the basis of these examples, compare the three lexeme formation processes in terms of their transparency. Remember that transparency involves both compositionality of meaning and the phonological stability of the base (i.e., the base is pronounced the same way in isolation and in the derived word):

a. - ish girlish kittenish sheepish loutish babyish

b. - ic cyclic metallic economic totemic organic

Productivity and creativity 81

c. - al herbal global homicidal glacial clinical

3. In this chapter, we have looked exclusively at productivity as it concerns derivational processes. We can, however, also compare the productivity of various types of compounding. English has compounds that consist of two nouns ( dog bed, windmill ), two adjectives ( bittersweet, blue-green ), and two verbs ( blow dry, stir fry ). Are all three types of compounding equally productive? (Hint: one way to start is by thinking of examples of NN, AA, and VV, and seeing which type gives you the most difficulty.) Give as much evidence as you can for your answer.

4. Look at the words you’ve collected in your Word Log. How many of them are formed with affixes? Which affixes? How many are formed by compounding or conversion? What does this tell you about the productivity of various processes in present-day English?

5. The graph in Figure 4.5 shows percentages of first citations in the OED with the suffixes -dom, -esque, -ship , -let, and -hood . Make some observations on the patterns that you observe in the graph. How good a view of the comparative productivity of these suffixes do you think this chart gives? Take into consideration what we have said in this chapter about basing estimates of productivity on material in a dictionary.

FIGURE 4.5 Percentages of first citations in the OED of suffixes -dom, -esque, -ship, -let, -hood .

6. Using COHA search for words ending in the suffix - eer (as in charioteer or mountaineer ). You’ll first need to clean the data, as you did for the exercises in Chapter 3 . Then look for all the - eer words which occur with a frequency of 1. Check which decade each word appeared in and see if you can make a graph showing how many hapaxes occurred in each

82 INTRODUCING MORPHOLOGY

decade. (You don’t need to do a P calculation – just count items with a frequency of 1 in each decade.) Can you see any trends in the productivity of the suffix - eer ?

7. Do the same for the prefix mega - . Discuss the trends that you find.

8. Go back to the splinters you found for exercise 13 in Chapter 3 . Try to determine when each one began to give rise to new forms in English. Do you see any trends in the formation of splinters?

KEY TERMS infix

circumfix

internal stem change

ablaut

consonant mutation

templatic morphology

reduplication

hypocoristic

CHAPTER OUTLINE

In this chapter you will begin to look in detail at data

from less familiar languages and learn about kinds of

morphological processes that are uncommon or don’t

exist at all in English and other familiar languages.

♦ We will look at kinds of word formation that may be

new to you: infixation, circumfixation, ablaut, umlaut,

and consonant mutation, reduplication, and templatic

morphology.

♦ And you will get further practice in morphological

analysis.

5 C H A P T E R

Lexeme formation: further afield

84 INTRODUCING MORPHOLOGY

5.1 Introduction

So far we have looked at the way lexeme formation works in English and

a few other relatively familiar languages. Of course, morphologists are

interested in how word formation (and inflection, which we will get to in

Chapter 6 ) works across all the languages of the world. Cross-linguistic

investigation allows us to see the ways in which English word formation

is typical and the ways in which it is unusual. From our perspective,

English morphology might seem utterly typical, but it’s only by compari-

son with other languages that we can begin to gain a wider perspective.

What is typical in English might very well turn out to be atypical in the

context of other languages.

In this chapter we will first learn how to work with data from less fam-

liar languages and then move on to kinds of word formation that are rare

or completely non-existent in English. Just to give you a taste of what’s to

come, take a look at the data in (1) :

(1) a. Tagalog (Schachter and Otanes 1972 : 356)

ganda ‘beauty’ gumanda ‘become beautiful’

hirap ‘difficulty’ humirap ‘become difficult’

b. Manchu (Haenisch 1961 : 34)

haha ‘man’ hehe ‘woman’

ama ‘father’ eme ‘mother

amila ‘cock’ emile ‘hen’

c. Samoan (Mosel and Hovdhaugen 1992 : 227)

a’a ‘kick’ a’aa’a ‘kick repeatedly’

‘etu ‘limp’ ‘etu‘etu ‘limp repeatedly’

fo’i ‘return’ fo’ifo’i ‘keep going back’

These examples should look quite different from the kinds of morphol-

ogy that we’ve concentrated on so far: prefixation , suffixation , compound-

ing , and conversion . In (1a) , it looks like a morpheme has been inserted

right into a base to form a verb. In (1b) , vowels have changed to form the

female correlates of male nouns, and in (1c) , segments of the base are

repeated to form what’s called the frequentative form of the verb (for a

verb meaning X, this form means ‘X repeatedly’). Prefixation, suffixation,

compounding, and conversion may be the main ways of forming new

words in English and many other languages, but there’s a much wider

world out there, and there are types of morphology that do not figure in

English at all, or figure only in the most minor ways.

In this chapter we’ll expand our horizons by surveying a number of

morphological processes that we have not yet encountered: different

kinds of affixes, internal stem changes to consonants and vowels, redupli-

cation, and templatic morphology. Our concentration will be on the struc-

tural aspects of morphology – the kinds of rules that languages can make

use of to form new words – as opposed to the semantic or grammatical

aspects. Our aim here is to characterize a sort of universal toolbag of rules

which languages may make use of in word formation.

Lexeme formation: further afield 85

5.2 How to: morphological analysis

Before we look at different kinds of word formation, student morpholo-

gists need to be comfortable working with data from less familiar lan-

guages. When you look at English , you may have an intuitive sense of how

to divide words into morphemes. But you might not have the same sense

when you encounter a language you know nothing about. Surprisingly,

with a few analytical tools at your disposal it’s possible to learn quite a lot

about how words are formed in a less familiar language. In this section I’ll

take you through the steps of developing a working hypothesis about how

the morphology works in a language you’ve probably never looked at

before.

How do linguists go about deciding what words are complex in a less

familiar language, what sorts of processes are involved in creating com-

plex words, and how to analyze individual words? Consider the words in

(2) , from the language Dyirbal, a language of the Pama-Nyungan family,

formerly spoken in Australia, but now, according to Ethnologue, nearly

extinct; the data are taken from Dixon ( 1972 : 222–33):

(2) a. �alŋgaŋunu ‘from a boy’

b. ya�aŋaru ‘like a man’

c. gugulaŋaru ‘like a platypus’

d. banabaḑun ‘proper water’

e. waŋalbaḑun ‘proper boomerang’

f. ya�abaḑun ‘proper man’

g. ya�agabun ‘another man’

h. ya�aḑaran ‘two men’

i. baŋguyḑaran ‘two frogs’

j. yugubila ‘with a stick’

k. waŋalḑaranbila ‘with two boomerangs’

l. miḑagabunŋunu ‘from another camp’

m. gugulabaḑunŋaru ‘like a proper platypus’

n. ya�agabunḑaran ‘two other men’

Just by looking at the Dyirbal words (2a) and (2b) and their glosses, you

really can’t tell anything. They might be simple or complex, but there’s no

way of knowing, because there are no parts of the two words that seem to

overlap. But as soon as you look at example (2c) and its gloss, you will

notice some overlap with (2b) . Both examples share the gloss ‘like a’, and

both have some characters at the end that overlap ( aŋaru ). So you might make a tentative hypothesis that these words are complex, and that they

can be broken down into two morphemes, ya� + aŋaru and gugul + aŋaru , respectively. You might also hypothesize that ya� means ‘man’, gugul

means ‘platypus’, and aŋaru means ‘like a’. This is a good first guess, but you should always be prepared to revise your analysis as you look at more

data.

If you then move on and look at examples (2d–f), you’ll notice that

they all share part of their meanings (‘proper’), and the end of each

86 INTRODUCING MORPHOLOGY

word has the sequence baḑun . It’s therefore reasonable to make the

hypothesis that baḑun means ‘proper’, and that what’s left over means

‘water’ in (2d) , ‘boomerang’ in (2e) , and ‘man’ in (2f) . But now, we need

to look back at our analysis of (2b) , because our first hypothesis was that

ya� meant ‘man’, and what’s left over in (2f) is not ya� but ya�a . We

therefore need to go back and revise our analysis of examples (2b) and

(2c) to be consistent with what we’ve learned from examples (2d–f). This

means that (2b) should be divided into ya�a + ŋaru and (2c) should be divided into gugula + ŋaru . What we’ve discovered so far is summarized in (3) :

(3) ya�a ‘man’ ŋaru ‘like a’ gugula ‘platypus’ baḑun ‘proper’

bana ‘water’ gabun ‘another’

waŋal ‘boomerang’ ḑaran ‘two’

We can now build on this hypothesis to analyze some more data. One

strategy that’s often good to use is to look for other words in which you

already recognize a piece. Indeed it looks like examples (2g) and (2h) both

have the piece ya�a and a gloss that includes ‘man’. If we subtract this

piece, we are left with two more bits we can now identify: gabun probably

means ‘another’ and ḑaran ‘two’. This in turn suggests that we can iden-

tify the piece baŋguy as meaning ‘frog’. At this point, we have a good idea how to analyze examples (2b–i), but we

still haven’t cracked (2a) . Example (2j) is still a problem as well, as so far, it

doesn’t overlap with any of the other examples. But as soon as we go on to

(2k) , we start to get a clue, because there are two morphemes that we can

now recognize in this word ‘boomerang’ and ‘two’, leaving only the final

bit bila which therefore must mean ‘with’. We can now go back to (2j) and

determine that yugu must mean ‘stick’. Example (2l) finally leads us back

to example (2a) : since we can identify a stretch in the middle of (2l) – the

morpheme gabun , which we decided means ‘another’– we can guess that

miḑu means ‘camp’ and ŋunu means ‘from’. Why not the opposite, by the way? The reason is that so far it looks like the more semantically content-

ful morphemes, like ‘man’ and ‘frog’, always come first, and the less con-

tentful come after; we might therefore hypothesize that morphemes like

‘from’ and ‘proper’ are suffixes in Dyirbal. And now we can finally go back

to example (2a) and decide that the morpheme for ‘boy’ is �alŋga . I leave it

to you to analyze the last two examples, and check that our analysis so far

is right.

I say ‘so far’ because we have only a tiny bit of data to work with here,

and every morphological analysis is provisional on checking it against

further data. Sometimes there are loose ends left after we’ve analyzed

our data as much as we can. One loose end you might notice in our

Dyirbal analysis is that it looks like there is no morpheme in any of our

data that corresponds to a word like a in English. It’s impossible to know

from this little data set whether Dyirbal has anything that corresponds

to indefinite articles .

Lexeme formation: further afield 87

Challenge

Look at the following data from the language San Miguel Chimalpa

Zoque a Mixe-Zoquean language spoken in Mexico and try your hand

at breaking it down into individual morphemes. What do you think

each morpheme means? (Johnson 2000 : 413)

yədə ‘this’ yəhi ‘there’ yəhə ‘here’ yəhənaŋ ‘towards here’ yəhə́ŋ ‘over here, towards here’ yəhí ŋ ‘to this point, no farther’ yey ‘soon, right now’

dedə ‘that’ dehi ‘there’

dehé ŋ ‘towards there’ dehí ŋ ‘from there’ dey ‘now, then’

5.3 Affixes: beyond prefixes and suffixes

Now that you’ve gotten your feet wet analyzing complex words in languages

that might not be familiar to you, we can begin to look beyond prefixation

and suffixation to more ‘exotic’ morphological processes. As we saw in

Chapter 3 , prefixes and suffixes are types of affixes that respectively go

before or after a base, but these are not the only positions in which affixes

can occur. This section will look at these different sorts of affixes.

5.3.1 Infixes Infixes are affixes that are inserted right into a root or base. We saw an

example in (1a) above. In Tagalog , a Malayo-Polynesian language spoken in

the Philippines, it is possible to form intransitive verbs meaning ‘become

X’ from adjectives by inserting the morpheme <um> after the first conso-

nant of the adjective root. Example (4) shows how the words can be broken

down, using the convention that the infix is enclosed in angled brackets

(< and >):

(4) g<um>anda become

beautiful

h<um>irap become

difficult

Another example of infixation can be found in Karok (a nearly extinct

Hokan language, formerly spoken in northern California). In Karok, a

form of verb called the intensive is created by infixing the morpheme -eg-

after the first consonant or cluster of consonants in the root, as in (5) :

88 INTRODUCING MORPHOLOGY

(5) Karok (Garrett 2001 : 269)

Base verb Intensive

la:y- ‘to pass’ l<eg>a:y

�kyork w - ‘to watch’ �ky<eg>ork w

koʔmoy- ‘to hear’ k<eg>oʔmoy

trahk- ‘to fetch water’ tr<eg>ahk

Both examples of infixation that we’ve seen so far have had the infix

right after the first consonant or consonant cluster of the base, but some-

times infixes can come near the end of the base as well. As the examples

in (6) illustrate, in Hua , a Trans-New Guinea language, the negative infix

<’ a > comes before the last syllable of the verb root:

(6) Hua (Haiman 1980 : 195)

zgavo ‘embrace’ zga<’a>vo ‘not embrace’

harupo ‘slip’ haru<’a>po ‘not slip’

rvato- ‘be nigh’ rva<’a>to ‘not be nigh’

The Oceanic language Hoava shows another interesting characteristic

that sometimes appears with infixation. In Hoava an infix <in> is used to

create nouns from adjectives or verbs. The <in> morpheme must be

infixed if the base it attaches to begins with a consonant, but it appears

as a prefix if the base begins with a vowel:

(7) Hoava (Davis 2003 : 39; Blevins 2014 : 137)

a. to ‘alive’ t<in>o ‘life’

b. va-bobe ‘fill’ v<in>abobe ‘filled object’

c. ta-poni ‘be given’ t<in>aponi ‘gift’

d. asa ‘grate’ <in>asa ‘pudding of grated cassava’

e. edo ‘happy’ <in>edo ‘happiness’

Challenge

The following data are from Ulwa , a Misumalpan language spoken in

Nicaragua. What is the morpheme that marks the possessed form and

what do you notice about where it is placed in the word? (Davis and

Tsujimura 2014 : 203)

Base Gloss Possessed form

sana ‘deer’ sanaka

bas ‘hair’ baska

kii ‘stone’ kiika

al ‘man’ alka

siwanak ‘root’ siwakanak

karasmak ‘knee’ karaskamak

anaalaaka ‘chin’ anaakalaaka

Infixation in English?

English doesn’t have any productive processes of infixation, but

there’s one marginal process that comes close, which is affectionately

Lexeme formation: further afield 89

referred to by morphologists as ‘ fuckin ’ infixation’. In colloquial

spoken English, we will often take our favorite taboo word or

expletive – in American English fucking , goddam , or frigging , in British

English bloody – and insert it into a base word:

abso-fuckin-lutely

fan-bloody-tastic

Ala-friggin’-bama

This kind of infixation is used to emphasize a word, to make it

stronger.

What’s particularly interesting is that we can’t insert fuckin just

anywhere in a word. Indeed, there are rather strict phonological

restrictions on the insertion of expletives. Try inserting your favorite

expletive into the following words:

Winnepesaukee

elementary

onomatopoeia

Next, try it on the following words:

Texas

Georgia

Can you do ‘fuckin’ infixation’ with these words?

Now think up some other words and try ‘ fuckin ’ infixation’. Can you

figure out what conditions the placement of the expletive?

5.3.2 Circumfixation and parasynthesis Another type of affix that occurs in languages is the circumfix . A circum-

fix consists of two parts – a prefix and a suffix that together create a new

lexeme from a base. We don’t consider the prefix and suffix to be separate,

because neither by itself creates that type of lexeme, or perhaps anything

at all. This kind of affixation is a form of parasynthesis , a phenomenon

in which a particular morphological category is signaled by the simulta-

neous presence of two morphemes.

One example of a circumfix can be found in Dutch , although Booij

( 2002 : 119) says that it’s no longer productive. In Dutch, to form a collec-

tive noun from a count noun, the morpheme ge- is affixed before the base

and -te after the base:

(8) berg ‘mountain’ ge-berg-te ‘mountain chain’

vogel ‘bird’ ge-vogel-te ‘flock of birds’

Neither geberg nor bergte alone forms a word – it’s only the presence of

both parts that signals the collective meaning. Another example can be

found in Tagalog (Malayo-Polynesian), where adding ka before and an after

a noun base X makes a noun meaning ‘group of X’:

90 INTRODUCING MORPHOLOGY

(9) Tagalog (Schachter and Otanes 1972 : 101)

Intsik ‘Chinese person’ ka-intsik-an ‘the Chinese’

pulo ‘island’ ka-pulu-an ‘archipelago’

Tagalog ‘Tagalog person’ ka-tagalog-an ‘the Tagalogs’

Again, neither ka + noun, nor noun + an , has its own meaning in these

words.

Two more examples of circumfixation can be found in (10) , from

Cavineña (a Tacanan language of Bolivia) and (11) , from Kambera (an

Austronesian language spoken in Indonesia):

(10) Cavineña (Guillaume 2008 : 435–6; Bauer 2014 : 127)

jutu ‘to dress s.o.’ e-jutu-ki ‘cloth’

sama ‘to cure’ e-sama-ki ‘medicine’

taru ‘to stir’ e-taru-ki ‘paddle’

teri ‘to close sth’ e-teri-ki ‘door’

(11) Kambera (Klamer 1998 : 245; Bauer 2014 : 127)

ngùru ‘murmur’ ka-ngùru-k ‘to make a murmuring sound’

ndùru ‘thunder’ ka-ndùru-k ‘to make the sound of thunder’

mbàti ‘drip’ ka-mbàti-k ‘to make a dripping sound’

hètu ‘sniff ’ ka-hètu-k ‘to make a sniffing sound’

Bauer ( 2014 : 128) points out that Cavineña does have a prefix e - and a

suffix - ki , but the meanings of these affixes when they are used alone are

quite different from the meaning that we find in (10) .

Challenge

Remember that in Chapter 3 we learned to draw word trees. Review

Chapter 3 , section 2, and think about how we would have to draw

word trees for words with circumfixes.

Circumfixes are only one sort of parasynthetic affix. Take a look at the

data in (12) from the Mayan language Tzutujil and try to analyze them

yourself. What are the two parts of the affix?

(12) Tzutujil (Dayley 1985 : 176; Bauer 2014 : 128)

tik- ‘sow’ tijkoʔm ‘sowing’

loq’ ‘buy’ lojq’oʔm ‘item bought’

Here, it looks like we have an affix that consists of an infix and a suffix.

Again, the critical thing that distinguishes parasynthesis from other sorts

of affixation is that neither part of the affix can be used alone – at least

with the relevant meaning.

5.3.3 Other kinds of affix Occasionally in the literature on morphology we find reference to several

other types of affix. For the most part, in this book we use different terms

for these particular morphological processes, so here I will just mention the

terms and refer you to the sections of this book where they are discussed:

Lexeme formation: further afield 91

1. The exception here is conversion, which we discussed in Chapter 3.

• interfixes: These are what we have called linking elements . See Chapter 3 , section 3.4.

• simulfixes: This is another term for internal stem changes, which we will discuss in section 5.4 .

• transfixes: These are what we will call templatic morphology . See section 5.6 below.

5.4 Internal stem change

Most of the forms of lexeme formation that we’ve looked at so far have

involved adding something to a base (or combining bases). 1 Some languages,

however, have means of lexeme formation that involve changing the quality

of an internal vowel or consonant of a base, root, or stem; sometimes this

internal change occurs alone, and sometimes in conjunction with affixation

of some sort. Such processes are called internal stem change or apophony .

5.4.1 Vowel changes: ablaut and umlaut Example (13) gives some words where internal vowels change:

(13) a. Manchu (Haenisch 1961 : 34)

haha ‘man’ hehe ‘woman’

ama ‘father’ eme ‘mother’

amila ‘cock’ emile ‘hen’

b. Muskogee (Haas 1940 : 143)

nis ‘to buy it’ Stem class I

ní:s ‘to buy it’ Stem class III

ni:s ‘to buy it’ Stem class IV

c. German (Lederer 1969 : 25)

Bruder ‘brother’ Brüderlein ‘brother-dimin.’

Frau ‘woman’ Fräulein ‘woman-dimin.’

In Manchu, in forming the female equivalent of a male noun, back vow-

els become front vowels. In Muskogee, verb stems have five forms each of

which can be used in a number of verbal contexts (completive, incomple-

tive, durative, and so on). Three of these forms are differentiated by the

length and tonal patterns on their vowels. For class I, stems have short

vowels and no special tonal accent. Class III stems have long vowels and

falling tonal accents, and class IV have long vowels and no special accent.

Morphological processes that affect the quality, quantity, or tonal pat-

terns of vowels are often referred to as ablaut .

A note on

English

Ablaut figures in a

minor way in the

morphology of

English as well, as

we can see in the

past and past parti-

ciple forms of verbs

like sing (past

sang, past pple.

sung ) or sit (past

and past pple. sat ).

Since ablaut figures

only in inf lectional

forms of English,

though, we will post-

pone further discus-

sion of it until

Chapter 6 .

In German , when certain suffixes like the diminutive suffix -lein are

added to a stem, the stem vowel becomes a front vowel. Historically, this

fronting was a phonological process that occurred when a following suf-

fix itself contained a front vowel; this process is called umlaut . Over time,

the front vowels were lost in some suffixes or became back vowels, as is

the case with the diminutive suffix -lein (pronounced [lɑιn]).

92 INTRODUCING MORPHOLOGY

5.4.2 Consonant mutation In some languages morphological processes are signaled by changes in

consonants rather than vowels in the base, root, or stem. Such processes

are called consonant mutations . As with the vowel processes noted above,

consonant mutations may occur alone or in conjunction with prefixes or

suffixes.

(14) a. Seereer-Siin (McLaughlin 2000 : 335)

odon ‘mouth’ o n don ‘mouth-dimin.’

okawul ‘griot’ o ŋ gawul ‘griot-dimin.’

opa� ‘slave’ o m ba� ‘slave-dimin.’

b. Chemehuevi (Press 1979 : 21–2)

punikai ‘see’ navunika ‘see-reflexive’

t ka ‘eat’ nar ka ‘eat-reflexive’

koa ‘cut’ naγoa ‘cut-reflexive’

In (14a) , in the West Atlantic language Seereer-Siin, some noun diminu-

tives are formed by replacing the first stop consonant in the stem, for

example [p, k, d] in the words in (14a) , with the corresponding prenasal-

ized stop ([ m b, ŋ g, n d]). And in Chemehuevi (a Uto-Aztecan language), illus-

trated in (14b) , the reflexive of a verb is formed by prefixing na- and chang-

ing the initial stop consonant of the root to a voiced continuant ([p]

becomes [v], [t] becomes [r], and [k] becomes [γ]).

Challenge

The following data come from a language isolate called Nivkh , which

is spoken in Siberia (Spencer 1991 : 19; Davis and Tsujimura 2014 : 210):

Transitive Instransitive Gloss

r�ŋz�l�d t�ŋz�l�d ‘weigh’ χavud q�avud ‘warm up’ γesqod kesqod ‘burn something/oneself ’

vakzd pakzd ‘lose/get lost’

rad t�ad ‘bake’

Use the IPA chart at the beginning of the book to help you figure out

how the consonant mutation that relates the transitive and intransi-

tive forms works.

Another note

on English

You might be

interested to know

that a historical

process of umlaut is

responsible for such

singular/plural pairs

in English as foot ∼

feet or goose ∼

geese. In earlier

stages of English, the

singular of the noun

‘foot’ was fot and its

plural foti (similarly

gos sg. ∼ gosi pl.).

The high front vowel

of the plural suffix

caused the vowel of

the base to become

front (so the singular

and plural were then

fot ∼ feti and gos ∼

gesi ). The plural

suffix was eventually

lost, leaving the

difference in vowels

as the only signal of

plurality in those

words.

5.5 Reduplication

Reduplication is a morphological process in which all or part of a word (typi-

cally the base, but not always) is repeated. Some examples are given in (15) :

(15) a. Hausa (Newman 2000 : 42)

bāya ‘behind’ bāya bāya ‘a bit behind’ gàba ‘forward’ gàba gàba ‘a bit forward’

asà ‘below’ asà asà ‘a bit below’

Lexeme formation: further afield 93

b. Samoan (Mosel and Hovdhaugen 1992 : 229)

‘apa ‘beat, lash’ ‘apa’apa ‘wing, fin’

au ‘flow on, roll on’ auau ‘current’

solo ‘wipe, dry’ solosolo ‘handkerchief ’

(15a) and (15b) illustrate full reduplication , a process by which an entire

base is repeated. In the case of Hausa, full reduplication is used to form

what’s called an attenuative , which is a form meaning ‘sort of ’ or ‘a little

bit’. In Samoan full reduplication is used to form nouns from verbs.

Some languages have what is called partial reduplication in which only

part of the base is repeated. The example in (16) comes from Samoan.

(16) Samoan (Mosel and Hovdhaugen 1992 : 223)

lafo ‘plot of land’ lalafo ‘clear land’

lago ‘pillow, bolster’ lalago ‘rest, keep steady’

pine ‘pin, peg’ pipine ‘secure with pegs’

In (16) you can see that partial reduplication in Samoan repeats the first

consonant and vowel of the base; this process derives verbs from nouns.

Mokilese , an Austronesian language spoken in Micronesia, also has partial

reduplication, but in this case what is reduplicated is a heavy syllable,

that is, a syllable that either has a long vowel or ends in a consonant: 2

(17) Mokilese (Harrison 1973 : Inkelas 2014 : 170)

pɔdok ‘plant’ pɔd-pɔdok ‘planting’

soorɔk ‘tear’ soo-soorɔk ‘tearing’

diar ‘find’ dii-diar ‘finding’

andip ‘spit’ and-andip ‘spitting’

Partial reduplication need not repeat the initial part of a base; it may

also in some languages repeat the final part of the base, as the example

from Teton Dakota in (18) illustrates:

(18) Dakota (Teton) (Shaw 1980 : 321)

wa+ksà ‘cut with sawing motion’ wa+ksà-ksà ‘slice up’

wač h í ‘dance’ wač h íč h i ‘jump up and down’

In this dialect of Dakota, the final syllable of the verb root can be redu-

plicated to indicate iterative or repetitive action.

Another sort of process that can be considered a form of reduplication

is one where a base is fully or partially repeated, but with the substitution

of one or more segments, something which Inkelas ( 2014 : 171) refers to as

‘echo’ reduplication . One dialect of English actually has a process that

illustrates this. In Yiddish-English you can replace the beginning of a word

with the sequence [ʃm] to express irony or derision, as illustrated in (19) :

(19) book schmook, fix schmix, Jennifer schmennifer

Reduplication is an interesting process in terms of structure, as we have

seen above, but it is also interesting semantically in that it sometimes

expresses meanings that can be considered iconic. By ‘iconic’, we mean

2. We will look more at the interaction of reduplication and syllable structure in Chapter 9.

94 INTRODUCING MORPHOLOGY

cases in which the form of the derived word suggests its meaning. When

reduplication is semantically iconic, it can express plurality, intensity,

augmentation, repetition, or other meanings suggesting ‘more-ness’. It is

important to point out, though, that the meaning conveyed by reduplica-

tion need not be iconic. Cross-linguistically, processes of reduplication can

be found which express exactly the opposite of ‘more-ness’, diminutives,

for example, as well as meanings that have nothing to do with size or quan-

tity. Indeed, as the examples in (15b) and (16) illustrate, reduplication can

be used for canonical derivational purposes such as forming nouns from

verbs or verbs from nouns.

Challenge

Does English have any other sorts of reduplication? Consider the

following examples from English:

willy-nilly

hocus-pocus

mumbo-jumbo

hanky-panky

hodge-podge

handy-dandy

hoity-toity

helter-skelter

Can you think of more examples of this sort? Do you think that words

like these express a process of reduplication in English? If so, is it

productive? If not, why not?

5.6 Templatic morphology

Consider the data in (20) :

(20) Arabic (McCarthy 1979 : 244; 1981 : 374)

katab ‘wrote’

kattab ‘caused to write’

kaatab ‘corresponded’

ktatab ‘wrote, copied’

kutib ‘was written’ (perfective passive)

All of the words in (20) have something to do with writing, and all share

the consonants ktb , although in a couple of the forms, there’s more than

one t . All the active verb forms have the vowel a ; the passive verb form has

the vowels ui . Each word has a different pattern of vowels and consonants,

and each expresses a slightly different concept. What we find in Arabic is

called templatic morphology or root and pattern morphology.

In Arabic, the root of a word typically consists of three consonants (like

ktb ), the triliteral root, which supply the core meaning. These three

Lexeme formation: further afield 95

consonants may be interspersed with vowels in a number of different

ways to modify the meaning of the root. The precise pattern of conso-

nants and vowels – sometimes called the template – can be associated

with specific meanings. For example, the pattern CVCVC is the non-

causative non-reciprocal form of the verb, but the pattern CVCCVC con-

veys a causative meaning, and the pattern CVVCVC a reciprocal meaning

(we can take correspond to mean something like ‘write to each other’).

Each of these template patterns is called a binyan (a term which comes

from traditional Hebrew grammar). The specific vowels that get inter-

spersed between the consonants in these patterns can contribute inflec-

tional meanings; so the vowel a is used in active forms, and the vowels ui

in passive forms. Roots in Arabic are occasionally called transfixes

because some morphologists look at them as affixes that occur discon-

tinuously across the word.

Root and pattern morphology is very characteristic of the Semitic fam-

ily of languages, which includes Arabic and Hebrew. But it can also be

found in other languages, for example the Uto-Aztecan language Cupeño ,

a nearly extinct language of Southern California. Cupeño verbs can have

a form called the habilitative , which means something like ‘can V’:

(21) Cupeño (McCarthy 1984 : 309)

a. čál ‘husk’ čáʔaʔal ‘can husk’ tə́éw ‘see’ tə́ʔeʔew ‘can see’ həl y ə́p ‘hiccup’ həl y ə́ʔəʔəp ‘can hiccup’

kəláw ‘gather wood’ kəláʔaʔaw ‘can gather wood’

b. páčik ‘leach acorns’ páčiʔik ‘can leach acorns’ čás. pəl ‘mend’ čás. pəʔəl ‘can mend’

One way of looking at the habilitative forms in Cupeño is that they

conform to templates like those in (22) , where the vowel that is bolded is

the stressed vowel:

(22) Template for Cupeño habilitatives

a. (CV)C V ʔVʔVC

b. C V C(C)VʔVC

If the only or the second vowel is the stressed vowel, as is the case in

the examples in (22a) , the habilitative form adds two more syllables,

each of which start with [ʔ]. The final vowel of the stem is spread to the

new syllables. The situation is slightly different if the first vowel of two

is the stressed vowel, as the examples in (22b) show. In that case, the

template has only one more syllable than the stem, again with the

glottal stop as the consonant, and the vowel supplied by the last stem

vowel.

A final example of templatic morphology comes from another Native

American language, Sierra Miwok (Penutian family, spoken in

California). In this language, new words can be derived by adding a suf-

fix which then supplies a specific template for the base. Consider the

examples in (23) :

96 INTRODUCING MORPHOLOGY

(23) Sierra Miwok (Smith 1985 : 365, 371–2)

a. pet. ja ‘drop several things’ peet. aj -tee-ny ‘string out’

halki ‘hunt’ haalik -tee-ny ‘hunt along a trail’

b. hulaw ‘forget’ hulwaw-we ‘be late’

ʔokiih ‘beg for food’ ʔokhih-he ‘be pitiful’

c. hywaat- ‘run’ hywattatt ‘run around’

hyleet ‘fly, be in the air’ hylettett ‘flop about (fish)’

The examples in (23a) show that the suffix - tee-ny , which forms what

Smith calls the ‘linear distributive’, makes the verb stem conform to a

template of the form CVVCVC. The forms in (23b) have a suffix with the

form Ce, where the C is the last consonant of the verb stem. In addition,

the verb stem is made to conform to the pattern CVCCVC. Finally, in (23c)

verb stems are made into derived forms that mean something like ‘X

around’ just by making them conform to a template that looks like

CVCVCCVCC, with no suffix added.

Challenge

Consider the following data from modern Hebrew (Davis and Tsujimura

2014 : 195):

lamad ‘learn’ limed ‘teach’

takan ‘be straight’ tiken ‘repair’

yavan ‘Greece’ yiven ‘Hellenize’

targum ‘translation’ tirgem ‘translate’

Describe the process that is used to derive the verbs in the second

column from their bases in the first column.

5.7 Subtractive processes

Generally we think of lexeme formation as an additive process. We start

out with a base and add affixes (whether prefixes, suffixes, infixes, or

circumfixes) or compound two or more bases or reduplicate part of the

base, resulting in a derived word that is longer than the original base.

Not all lexeme formation is additive, however. There are some languages

that have processes that we might call subtractive in the sense that the

derived lexeme consists of a base from which something has been

removed. One example can be found in the Muskogean language Koasati :

(24) Koasati (Martin 1988 : 230–1)

singular plural

pitáf-li-n pít-li-n ‘to slice up the middle’

tiwáp-li-n tíw-wi-n ‘to open something’

icoktaká:-li-n icokták-li-n ‘to open one’s mouth’

misíp-li-n mís-li-n ‘to wink’

The plural is formed by subtracting the final vowel and consonant of

the base.

Lexeme formation: further afield 97

Challenge

Why do we say that the process in (24) is subtractive? Consider an

alternative hypothesis in which the singular is formed from the

plural bases ( tí w , pí t , etc.) by adding a morpheme. Why would this not

be a good hypothesis?

Exotic as subtractive morphology may seem to speakers of English , we

actually have something comparable. It’s common in English to form

nicknames ( hypocoristics , if you want the technical term!) by subtracting

part of the full name. Consider the names in (25) with their hypocoristics:

(25) Patrick Pat

Barbara Barb

Mitchell Mitch

Alan Al

Benjamin Ben

To form a nickname in English, we typically subtract all but the first

CVC sequence of the name. Sometimes we add the suffix - y or -ie as well,

as in Patty from Patricia or Allie from Allison . So we can think of nickname

formation as a sort of subtractive process as well, with or without accom-

panying affixation.

Summary In this chapter we have completed our survey of the different types of rules that can be used in forming new lexemes in the languages of the

world. We have gone beyond prefixation, suffixation, compounding,

and conversion to add new types of affixes (infixes, circumfixes), and

new processes like internal stem change (ablaut, umlaut, and

consonant mutation), reduplication (full and partial), templatic and

subtractive morphology.

Exercises 1. The Austronesian language Leti has a process that derives nouns

meaning ‘the act of V-ing’ from verbs. Consider the data below (from Blevins 1999 : 390):

kakri ‘cry’ kniakri ‘the act of crying’

pali ‘float’ pniali ‘the act of floating’

sai ‘climb’ sniai ‘the act of climbing’

teti ‘chop’ tnieti ‘the act of chopping’

vaka ‘ask (for)’ vniaka ‘the act of asking’

va-nunsu ‘knead’ vnianunsu ‘massage’ = ‘the act of kneading’

98 INTRODUCING MORPHOLOGY

a. Divide the Leti words in the second column into morphemes, and give the meaning of each morpheme.

b. What is the morphological rule that creates nouns from verbs in Leti? What kind of a rule is it?

c. Now consider the following forms:

atu ‘know’ niatu ‘knowledge’ = ‘the act of knowing’

odi ‘carry’ niodi ‘the act of carrying’

osri ‘hunt’ niosri ‘hunt’ = ‘the act of hunting’

Divide these new words into morphemes and discuss what changes you need to make to the morphological rule you wrote for part (b) in order to account for this new data.

2. The examples below are from the Native American language Yurok (data from Garrett 2001 : 274):

kep’e� ‘housepit’ kep’kep’e� ‘there are several housepits’

ket’ul ‘there’s a lake’ ket’ket’ul ‘there’s a series of lakes’

pegon ‘to split’ pegpegon ‘to split in several places’

siton ‘to crack’ (intrans.)

sitsiton ‘to crack several times’

tekun ‘to be stuck together’

tektekun ‘to be stuck together in several places’

Write a word formation rule that derives the Yurok forms in the second column from the corresponding base in the first column. Make sure to include both the structural and semantic effects of the rule. What kind of a morphological rule is this?

3. Consider the data below from the Dravidian language Kannada (data from Sridhar 1990 : 268):

a:Ta ‘game’ a:Ta-gi:Ta ‘games and the like’

huli ‘tiger’ huli-gili ‘tigers and the like’

sphu:rti ‘inspiration’ sphu:rti-gi:rti ‘inspiration, etc.’

autaNa ‘banquet’ autaNa-gi:taNa ‘banquet, etc.’

Try to write a morphological rule that derives the words in the second column from the bases in the first column. What kind of morphological rule is this? How does it differ from other morphological rules we’ve looked at in this chapter?

4. In the South Munda language Gtaʔ a number of different forms can be derived from a noun base, as the examples here show (data from McCarthy 1983 ):

kitoŋ ‘god’ kataŋ ‘being with powers equal to “kitoŋ”‘ kitiŋ ‘being smaller, weaker than “kitoŋ”’ kutaŋ ‘being other than “kitoŋ”‘ (e.g., spirits, ghosts) kesu ‘wrapper worn against cold’

kasa ‘cloth equivalent to “kesu” in size and texture’

kisi ‘small or thin piece of cloth’

kusa ‘any other material useable against cold’

Propose an analysis of this process. What kind of word formation rule is at work here?

Lexeme formation: further afield 99

5. The following words, taken from Yu ( 2004 : 620), are characteristic of the speech of the character Homer Simpson from the animated TV show The Simpsons. In this data, Homer Simpson seems to display a process of infixation:

saxomaphone ‘saxophone’ Missimassippi ‘Mississippi’

telemaphone ‘telephone’ Alamabama ‘Alabama’

wondermaful ‘wonderful’ diamalectic ‘dialectic’

feudamalism ‘feudalism’ Michamalangelo ‘Michaelangelo’

secrematery ‘secretary’

terrimatory ‘territory’

Is this like real cases of infixation that we saw in this chapter? If so, why? If not, why not? Try to formulate a precise rule for Homer Simpson infixation.

6. The following examples from the Semitic language Amharic illustrate a form of language disguise or play language (like Pig Latin) used by young women in the Ethiopian capital, Addis Ababa (McCarthy 1984 : 306):

g w aro ‘backyard’ g w ayrər

gιn ‘but’ gaynən

mətt’a ‘come’ mayt’ət

kιfu ‘cruel’ kayfəf

həd ‘go’ haydəd

man ‘who’ maynən

Figure out the morphological rule that creates the play language version (the third column) of the Amharic words. What kind of morphological rule is this?

7. Consider the following, from the Muskogean language Alabama (Hardy and Montler 1988 : 394):

salatli ‘slide once’ salaali ‘slide repeatedly’

haatanatli ‘turn around once’ haatanaali ‘turn around repeatedly’

noktil-ifka ‘choke once’ noktil-iika ‘choke repeatedly’

Describe the word formation process that derives the words in the second column from those in the first column. What kind of morphological process is this?

8. The following data come from the Muskogean language Koasati (Kimball 1991 : 351). Write a word formation rule for the process that they illustrate:

molápkan ‘to gleam’ molalápkan ‘to flash’

bolótin ‘to shake’ bololótin ‘to shake with fear’

wacíplin ‘to feel a stabbing pain’

wacicíplin ‘to feel repeated stabbing pains’

konótlin ‘to roll’ kononó:tlin ‘to quiver fatly’

watóhlin ‘to clabber’ watotóhlin ‘to jiggle like jello’

9. The language Semai , a Mon-Khmer language spoken in Malaysia, derives what Diffloth ( 1976 ) calls expressive forms from nouns using a process

100 INTRODUCING MORPHOLOGY

of reduplication. Consider the data below and try to formulate the rule of reduplication (data from Diffloth 1976 : 252–6):

dŋɔh dhdŋɔh ‘appearance of nodding constantly’ dyɔ̃:l dldyɔ̃:l ‘appearance of object which goes on floating

down’

taʔəh thtaʔəh ‘appearance of a large stomach constantly bulging out’

gh�:p gpgh�:p ‘irritation on skin’

10. Huastec is a Mayan language spoken in Mexico. Identify the morphemes in the examples below and try to assign them meanings. Discuss anything that strikes you as unusual in these data (data from Edmonson 1995 : 380–2).

1. ʔu ¢iʔikme:l ‘it becomes sweet’

2. ʔu ¢iʔikme: ‘it became sweet’

3. ʔu ¢iʔikme:nek ‘it has become sweet’

4. ʔu ¢iʔikme:neki¢ ‘it has already become sweet’

5. ʔin ¢iʔikme:θa:l ‘he makes it become sweet’ 6. ʔin ¢iʔikme:θaʔ ‘he made it become sweet’ 7. ʔin ¢iʔikme:θa:mal ‘he has made it become sweet’ 8. ʔan ¢iʔikme:θom ‘the one who sweetens’ 9. ¢iʔikme:θa:mehi¢ ‘it has already been made sweet’

10. ʔin ¢iʔikme:θanči ‘he made it become sweet for someone’

11. ʔin ¢iʔikme:θanča:mali¢ ‘he has already made it become sweet for someone’

11. The following data are from the Yuman language Hualapai , spoken in northern Arizona (from Watahomigie et al. 1982 : 197–8). Discuss the process that is used to form the adjectives/verbs from nouns.

chud ‘winter’ chu:dk ‘wintery’

guwí ‘cloud’ guwi:k ‘cloudy’

i’i ‘wood’ i’i:k ‘woody’

mà d- akwí d ‘whirlwind’ mà dakwi:dk ‘whirlwindy’ gí nya ‘younger sibling’ gi:nyk ‘to have a younger

sibling’

kwà sivdí v ‘fence’ kwà sivdi:k ‘to fence’

KEY TERMS person

number

gender

case

tense

aspect

inherent

contextual

paradigm

evidentiality

CHAPTER OUTLINE

In this chapter you will learn about inflection, the sort of

morphology that expresses grammatical distinctions.

♦ We will look at a wide variety of types of inflection, both

familiar and unfamiliar, including number, person,

gender and noun class, case, tense and aspect, voice,

mood and modality.

♦ We will learn what morphologists mean by a ‘paradigm’

and look at patterns within paradigms.

♦ And we will consider whether it is always clear where to

draw the line between inflection and derivation.

6 C H A P T E R

Inflection

102 INTRODUCING MORPHOLOGY

6.1 Introduction

At the outset of this book we divided morphology into two domains:

inflectional and derivational word formation. In the last three chapters,

we have concentrated on derivational word formation – types of word

formation that create new lexemes. In this chapter, we turn our attention

to inflectional word formation.

Inflection refers to word formation that does not change category and

does not create new lexemes, but rather changes the form of lexemes so

that they fit into different grammatical contexts. As we’ll see in detail

below, grammatical meaning can include information about number

(singular vs . plural), person (first, second, third), tense (past, present,

future), and other distinctions as well. In this chapter, we will first survey

different forms of inflection that can be found both in English and famil-

iar languages, and further afield in the languages of the world, and then

look at the ways in which inflection can work.

A word before we start though. We’ve seen that new lexemes can be

derived using all sorts of different formal processes of word formation:

affixation, compounding, conversion, internal stem change, reduplica-

tion, templatic morphology. Inflectional word formation makes use of

almost all of these types of word formation rules as well, with the possible

exceptions of compounding and subtractive processes. That is, just as lan-

guages may have derivational affixes that form new lexemes, they may

have inflectional affixes that make those lexemes suited for one gram-

matical context or another; similarly, languages may have rules of redu-

plication for either derivational purposes or inflectional purposes. In

other words, we might say that form (the type of rule or process) is inde-

pendent of function (derivation or inflection). Keep this in mind; many of

the examples that I’ll use to illustrate points below make use of affixation,

but in many cases I could have chosen examples with reduplication or

internal stem change as well.

6.2 Types of inflection

Native speakers of English are often surprised at the kinds of inflection

that can be found in languages – English is a language that has relatively

little inflection, as languages go. So we’ll start by surveying some of the

types of inflection that can be found in the languages of the world.

6.2.1 Number Perhaps the most familiar inflectional category for speakers of English is

number . In English, nouns can be marked as singular or plural:

(1) Singular cat, mouse, ox, child

Plural cats, mice, oxen, children

Although the vast majority of nouns pluralize in English by adding -s (or

in terms of sounds, one of the variants [s], [z], or [əz]), some nouns form

their plurals irregularly. We will return to the issue of regular versus

Inflection 103

irregular inflections shortly. In English, it is required to mark the plural

on nouns in a context in which more than one of that noun is being dis-

cussed ( I have six beagles ). This is not the case in all languages, as our

Mandarin Chinese example in Chapter 1 illustrated.

Some languages distinguish a third category of number in addition to

singular and plural. For example, in the Eskimo-Aleut language Yup’ik ,

nouns inflect not only for singular and plural, but also for what is called

dual . This is a number-marking that means ‘two’:

(2) Yup’ik (Mithun 1999 : 79)

qayaq ‘kayak’ paluqtaq ‘beaver’

qayak ‘two kayaks’ paluqtak ‘two beavers’

qayat ‘three or more kayaks’ paluqtat ‘three or more

beavers’

As we’ll see soon, some languages can make the singular/dual/plural dis-

tinction on verbs, as well as on nouns .

6.2.2 Person Students of Indo-European languages like Latin or German, know that

verbs in those languages are marked for the inflectional category of

person : that is, verbs exhibit different endings depending on whether the

subject of the sentence is the speaker (first person), the hearer (second

person), or someone else (third person); frequently number is also

expressed as well as person:

(3) a. Latin : amāre ‘to love’ Singular 1st amō (-o) Plural 1st amāmus (-mus)

2nd amās (-s) 2nd amātis (-tis) 3rd amat (-t) 3rd amant (-nt)

b. German: sagen ‘to say’

Singular 1st sage (-e) Plural 1st sagen (-en)

2nd sagst (-st) 2nd sagt (-t)

3rd sagt (-t) 3rd sagen (-en)

Speakers of Indo-European languages may, however, be less familiar with

marking person on nouns. It is not unusual for languages to mark person

on nouns to show possession, something we do in English with separate

possessive pronouns. For example, the Iroquoian language Mohawk uses

prefixes to mark person on nouns:

(4) Mohawk nouns (Mithun 1999 : 69)

Singular 1st person k-hnia’sà:ke ‘my throat’

2nd person s- hnia’sà:ke ‘your throat’

3rd person ie- hnia’sà:ke ‘her throat’

ra- hnia’sà:ke ‘his throat’

Plural 1st person iakwa- hnia’sà:ke ‘our throats’

2nd person sewa- hnia’sà:ke ‘your pl. throats’

3rd person konti- hnia’sà:ke ‘their F. throats’

rati- hnia’sà:ke ‘their M. throats’

104 INTRODUCING MORPHOLOGY

Mohawk and other languages also show another kind of person mark-

ing that we don’t have in English, making a distinction between the inclu-

sive and exclusive forms of the first person plural. In an inclusive form,

the speaker includes herself and the hearer. In the exclusive form, the

speaker includes herself and others, but not the hearer. So the inclusive

form of the first person could be thought of as a combination of first and

second person marking, and the exclusive as a combination of first and

third person marking. This distinction can be marked in Mohawk as well,

specifically with prefixes on verbs:

(5) Mohawk verbs (Mithun 1999 : 70)

1st person inclusive tewa-hià:tons ‘we all (you pl. and I) are writing’

1st person exclusive iakwa- hià:tons ‘we all (they and I) are writing’

As we mentioned above, it is also possible to mark verbs if the subject

consists of exactly two people. So in addition to the inclusive and exclusive

forms in (5) , Mohawk also has first person dual inclusive and exclusive

forms (Mithun 1999 : 70):

(6) Dual inclusive teni- hià:tons ‘we two (you and I) are writing’

Dual exclusive iakeni- hià:tons ‘we two (s/he and I) are writing’

English verb forms are ambiguous with respect to these distinctions. If I

say “we write,” neither the form of the verb nor the form of the pronoun

makes explicit whether the hearer is included or not, or how many other

than the speaker are involved (although of course we can make the dis-

tinction in a round-about way, if we need to!) .

6.2.3 Gender and noun class If you’ve studied French, Spanish, German, Latin, Russian, or another

Indo-European language, you’re probably familiar with the concept of

gender . In languages that have grammatical gender, nouns are divided

into two or more classes with which other elements in a sentence – for

example, articles and adjectives – must agree. We use French and German

as our examples here:

(7) a. French Masculine Feminine

homme ‘man’ femme ‘woman’

rat ‘rat’ souris ‘mouse’

bureau ‘desk’ table ‘table’

b . German Masculine Neuter Feminine

Mann ‘man’ Kind ‘child’ Frau ‘woman’

Tisch ‘table’ Pult ‘desk’ Mauer ‘wall’

Hund ‘dog’ Pferd ‘horse’ Maus ‘mouse’

French has two genders, masculine and feminine. German has three gen-

ders, masculine, feminine, and neuter. While sometimes the real-world

sex of the noun’s referent determines the grammatical gender of the

noun – that is, the class that the noun belongs to – in many more cases

nouns are assigned to genders with some degree of arbitrariness. So while

the words for ‘man’ and ‘woman’ in both languages are masculine and

Inflection 105

feminine respectively, in accordance with natural gender , the assignment

of various animal names to gender classes is quite arbitrary. Rats, mice,

dogs, and horses of course have natural gender – in the real world they

must be either male or female – but French and German grammar places

them in a gender independent of their natural genders. In French, rats are

masculine, but mice feminine. In German, dogs are masculine, horses

neuter, and mice feminine. Inanimate nouns have no natural gender, but

they are nevertheless classed as either masculine or feminine in French,

and as any of the three genders in German.

Assignment to gender classes sometimes seems completely arbitrary,

but it is not always as arbitrary as you might think. For example, in

German all nouns derived with the derivational suffixes -ung, -keit, -heit,

and -schaft are feminine, regardless of the gender of their bases. All of

these suffixes form abstract nouns, so it’s possible to say that derived

abstract nouns are always feminine. Similarly, the diminutive suffixes

- chen and -lein produce neuter nouns, regardless of the base they attach to.

In other languages, nouns may be assigned to a gender based on their

phonological shape. For example, the Afro-Asiatic language Hausa has

masculine and feminine genders. Nouns for males are masculine and

those for females are feminine in accordance with natural gender, but the

rest of the nouns are assigned to one of the classes by the phonological

form of the base: nouns that end in -aa are feminine, and everything else

is masculine (Corbett 1991 : 53).

For many nouns in French and German, neither the meaning of the

noun nor its phonological form signals its gender. In other words, there

are no suffixes or other marks right on the nouns to tell us their genders

(life would be much easier for second language learners of these languag-

es if there were!). Rather, we can tell what the gender of the noun is by

other elements in a sentence that are in agreement with a noun. So in

French, the definite article le is used with masculine nouns ( le bureau ), and

la with feminines ( la table ). Similarly, in German the definite article der is

used with masculines ( der Tisch ), das with neuters ( das Pferd ) and die with

feminines ( die Maus ).

The gender systems we are most familiar with are typical of Indo-

European languages, and occur in other language families as well, but

there are many languages outside of Indo-European that exhibit genders

or noun classes based on distinctions other than (or in addition to) mas-

culine, feminine, and neuter. Languages may have human and nonhuman

classes or classes for rational beings as opposed to everything else. In the

Algonquian family of languages, noun classes are based not on masculine

and feminine but on animacy. Words for people belong to the animate

class, but so, for example, do spirits and animals. And while most words

for inanimate things belong to the inanimate class, some belong to the

animate class; for example, the nouns for ‘snowshoe’ and ‘button’ in the

Algonquian language Ojibwa belong to the animate noun class (Corbett

1991 : 20). In other words, while there is a partial semantic basis for the

two classes, assignment to the animate and inanimate classes can still be

arbitrary.

106 INTRODUCING MORPHOLOGY

Languages are not limited to two or three classes. Consider the lan-

guage Yuchi , spoken in Oklahoma (Mithun 1999 : 103):

(8) Speaker Class

1. M singular or plural Yuchi, except certain female relatives

2. M singular female Yuchi relative, same or descending

generation (sister, daughter, niece, granddaughter)

F any female Yuchi of same or descending generation

3. F singular male Yuchi relative, same or descending

generation (brother, son, nephew, grandson)

4. M, F singular female Yuchi relative, ascending generation

(mother, aunt, grandmother)

5. F singular male unrelated Yuchi, or plural Yuchis of

same or descending generation

6. F singular male Yuchi of ascending generation (father,

uncle, grandfather, husband, or as a term of respect for

Yuchis of ascending generation)

7. M, F any non-Yuchi(s), animals

8. M, F vertical inanimate objects

9. M, F horizontal inanimate objects

10. M, F round inanimate objects

In Yuchi, nouns fall into classes based on whether they denote humans,

animals, or inanimate objects with certain shapes. For example, Class 1

contains nouns used by men to denote Yuchi people, except close female

relatives. The gender of a noun is indicated by a number of things, one of

which is the article suffix used with the noun. For example, gɔ n′t ε -n ɔ̧́ means ‘the Yuchi man’ and g ɔ n′t ε -w ə n ɔ̧́ ‘the non-Yuchi man’. ‘The tree’ is

y á -fa but ya-ʔέ is ‘the log’. Class 7 contains nouns used by either men or

women to denote animals or people who are not Yuchi. Other classes dis-

tinguish Yuchis from other humans, and Yuchis related to the speaker

from those unrelated to the speaker.

Related to gender or noun class systems are what some linguists call

classifiers , which are words or morphemes that are used in noun phrases

with quantifiers (like all, some ), number words, demonstratives ( this, that )

and the like. The classifier for a particular noun is chosen according to

various semantic criteria, that might include animacy, shape (e.g., long,

flat), consistency (e.g., wet, flexible), or kind (e.g., fruit, book) (Aikhenvald

2000 : 93; Velupillai 2012 : 172). Unlike gender or noun class, classifiers

typically don’t agree with nouns or determiners. They just serve to catego-

rize nouns in noun phrases. As we saw in Chapter 1 (examples (6) , (7) ),

Mandarin uses a system of classifiers in noun phrases, so that if we want

to talk about some number of giraffes, we need to say ‘one zhi giraffe’,

where zhi is the classifier for animals of this sort. However, if we were

counting books, we would need to use the classifier ben between the

numeral and the noun book .

6.2.4 Case Case is another grammatical category that may affect nouns (or whole

noun phrases). In languages that employ the inflectional category of case,

Inflection 107

nouns are distinguished on the basis of how they are deployed in sen-

tences, for example, whether they function as subject, direct object, indi-

rect object, as a location, time, or instrument, or as the object of a prepo-

sition. In Latin , for example, nouns must be inflected in one of five cases,

with singular and plural forms for each case:

(9) Latin :

Singular stella ‘star’ (F) puer ‘boy’ (M)

Nominative stella puer

Genitive stellae puerī Dative stellae puerō Accusative stellam puerum

Ablative stellā puerō

Plural

Nominative stellae puerī Genitive stellārum puerōrum Dative stellīs puerīs Accusative stellās puerōs Ablative stellīs puerīs

The nominative case forms are used for the subject of the sentence.

Accusative is generally used for the direct object and dative for the indi-

rect object. Genitive is used for the possessor (e.g., the boy’s shirt ). Ablative

is used for the objects of prepositions (e.g., cum ‘with’, d ē ‘from’), although some prepositions take objects in the accusative case ( ad ‘to’, post ‘after’).

Latin displays what is commonly called a nominative/accusative case system .

In this sort of system, subjects of verbs are nominative, whether the verbs are

transitive (i.e., they take an object) or intransitive (they don’t take an object):

(10) a. Puer amat puellam

boy. NOM loves girl. ACC

‘The boy loves the girl’

b. Puer it

boy. NOM goes

‘The boy goes’

Less frequent is a kind of case marking system called an ergative/

absolutive system . In this kind of system, the subject of a transitive verb

gets a case called the ergative . The subject of an intransitive verb gets a

case called the absolutive , which is also the case used for the direct object

of a transitive verb. The examples in (11) from Georgian illustrate an

ergative-absolutive case marking system (Whaley 1997 : 163): 1

(11) a. Student-i mivida

student- ABS went

‘The student went’

b. Student-ma ceril-i dacera

student- ERG letter- ABS wrote

‘The student wrote the letter’

1. Interestingly, Georgian has an ergative-absolutive case marking system in the perfect tense, but in the

present tense it has a nominative-accusative system.

108 INTRODUCING MORPHOLOGY

You can see the two systems compared schematically in (12) :

(12)

Nominative/ Accusative Ergative/ Absolutive

Subject of transitive verb Nominative Ergative

Subject of intransitive verb Nominative Absolutive

Object of transitive verb Accusative Absolutive

Ergative/absolutive case systems are less frequent in the languages of the

world than nominative/accusative systems, but they do occur in the

Pama-Nyungan languages of Australia (e.g., Dyirbal) , in the

Tsimshianic languages of North America (e.g., Sm’algyax, spoken in

British Columbia), in the language isolate Basque , as well as in

Caucasian languages like Georgian .

The Leipzig glossing rules

As you’ve already seen, morphologists frequently study word

formation phenomena from languages that we don’t speak ourselves

and may not have studied directly. To do so, we rely on grammars and

articles written by experts. Data in such sources are most helpful and

informative when they are glossed . By glossing, we refer to a

morpheme by morpheme translation, frequently written right under

an example in the language in question. The gloss is often then

followed by a colloquial translation, as in the example you saw in (11) .

Until recent years there was no agreed upon convention for what

abbreviations to use for particular inflectional or derivational

meanings, or what symbols to use to separate morphemes in the

language examples and meanings in the glosses. The Leipzig glossing

rules are an attempt to rectify this issue. These days, linguists are

encouraged to follow the guidelines for glossing that can be found at

www.eva.mpg.de/lingua/resources/glossing-rules.php . I won’t give you

the entire list of rules, but will mention just a few that are useful for

the student morphologist.

• Morphemes and their glosses are separated by hyphens and

aligned so that the gloss appears just below the morpheme it

translates.

• A colloquial translation occurs below the gloss, enclosed in single

quotes.

• Clitics (bound morphemes that are more loosely attached to their

bases than suffixes are – see Chapter 8 ), as opposed to affixes, are

separated from their host by an = sign.

• When the gloss has more parts than the morpheme it is

translating, the meanings are separated by a period, as in the

Turkish and Latin examples below.

• Grammatical morphemes such as inflections are written in small

capitals.

Inflection 109

6.2.5 Tense and aspect Tense and aspect are inflectional categories that usually pertain to verbs.

Both have to do with time, but in different ways.

Tense refers to the point of time of an event in relation to another

point – generally the point at which the speaker is speaking. In present

tense the point in time of speaking and of the event spoken about are the

same. In past tense the time of the event is before the time of speaking.

And in future tense the event time is after the time of speaking. This can

be represented schematically as in (13) , where S stands for the time of

speaking and E for the time of the event:

(13) Present S = E

Past E before S

Future S before E

In English, we mark the past tense using the inflectional suffix -ed on

verbs ( walked, yawned ), but there is no inflectional suffix for future tense.

Instead, we use a separate auxiliary verb will to form the future tense ( will

walk, will scream ). The use of a separate word to form a tense is called

periphrastic marking . Strictly speaking, periphrastic marking is a mat-

ter of syntax rather than morphology. Unlike English, Latin marks both

past and future inflectionally, that is, by means of morphology on the

verb:

(14) Present amō ‘I love’ Past amāvī ‘I loved’ Future amābō ‘I will love’

Past, present, and future are not the only possible tenses; some langu-

ages distinguish several kinds of past tense and several kinds of future

tense, based on how close or distant the event spoken about is from

the time of speaking. For example, in the nearly extinct Hokan language

Washo , there are four different past tenses and three different future

tenses (Mithun 1999 : 152–3):

Turkish çık-mak

come.out- INF

‘to come out’

Latin insul-arum

island- GEN.PL

‘of the islands’

There are many other rules available at the Leipzig website, as well as

lists of suggested abbreviations for various inflectional meanings. The

student morphologist should keep in mind, though, that up until recently

there were no agreed upon conventions for glossing, and one is apt to find

all sorts of variations on abbreviations and glossing conventions in articles

and grammars about different languages. Luckily, most authors include a

list of the abbreviations they use for the reader to consult!

110 INTRODUCING MORPHOLOGY

A past tense ‘sled’ might be a pile of junk in a shed, or something I used

to own, now owned by someone else. If your ‘wife’ is past tense, she might

be dead, or you might be divorced. A future tense ‘sled’ might also be a

pile of material yet to be assembled, or something you might buy or get as

a present. Your future tense ‘wife’ is your fiancée.

Aspect is another inf lectional category that may be marked on verbs.

Rather than showing the time of an event with respect to the point of

speaking, aspect conveys information about the internal composition

of the event or “the way in which the event occurs in time” (Bhat

1999 : 43).

One of the most frequently expressed aspectual distinctions that can be

found in the languages of the world is the distinction between perfective

and imperfective aspect. With perfective aspect, an event is viewed as

completed; we look at the event from the outside, and its internal struc-

ture is not relevant. With imperfective aspect, on the other hand, the

event is viewed as ongoing; we look at the event from the inside, as it

were. English isn’t the best language with which to illustrate this distinc-

tion, as tense and aspect are not completely distinct from one another, but

I can give you a rough example from English. In English, when we say I ate

the apple , we not only place the action in the past tense, but also look at it

(15) - lul distant past, before the lifetime of the speaker

- gul distant past, but within the speaker’s lifetime

- ayʔ intermediate past, earlier than the same day, but not

extremely distant

-leg recent past, earlier on the same day or during the previous

night

- ášaʔ near future, from the point of speaking until about an hour

from that point

- tiʔ intermediate future, after a short lapse of time (usually later

in the day)

- gab distant future, following day or later

Tense on nouns?

It may seem odd to think of putting tense marking on nouns; we

don’t do it in English, and it probably isn’t done in any of the

languages you’ve studied. But it’s not uncommon in native languages

of North America. For example, in Central Alaskan Yup’ik , both past

tense and future tense can be marked on nouns (Mithun 1999 : 154):

ikamraqa ‘my sled’

ikamralqa ‘my former sled’

ikamrarkaqa ‘my sled to be’

nuliaqa ‘my wife’

nulialqa ‘my late/ex-wife’

nuliarkaqa ‘my wife to be’

Inflection 111

as a completed whole. But if we say I was eating the apple , although the

action is still in the past, we focus on the event as it is progressing. The

Iroquoian language Seneca has a much clearer distinction between perfec-

tive and imperfective aspect (Mithun 1999 : 165):

(16) Perfective ǫkáhtaʔt ‘I got full’

Imperfective akáhtaʔs ‘I get full, I’m getting full’

Other forms of aspect focus on particular points in an event. Inceptive

aspect focuses on the beginning of an event. Continuative aspect focuses

on the middle of the event as it progresses, and completive on the end.

We can illustrate these aspects with the following sentences from the

Tibeto-Burman language Manipuri (Bhat 1999 : 52):

(17) a. Inceptive

məhak-nə phu-gət-li

he- NOM beat-start- NON.FUT

‘He began to beat it (and would continue to do so)’

b. Continuative

tombə layrik pa-rì

Tomba book read- CONT

‘Tomba is reading the book’

c. Completive

yumthək ədu yu-rəm-mì

roof that leak- COMP-CONT

‘That roof had been leaking (but not any more)’

A third category of aspectual distinction can be called quantificational .

Quantificational aspectual distinctions concern things like the number of

times an action is done or an event happens – once or repeatedly – or how

frequently an action is done. Among the quantificational aspects are

semelfactive , iterative , and habitual aspects. Actions that are done just

once are called semelfactive . The Athapaskan language Koyukon has a

special verb stem for actions that are done just once (Mithun 1999 : 168),

and West Greenlandic has suffixes that express iterative aspect for some-

thing that is done repeatedly and habitual aspect for something that is

usually or characteristically done (Fortescue 1984 : 280–2):

(18) a. Koyukon semelfactive

yeeltle�

‘she chopped it once, gave it a chop’

b. West Greenlandic iterative

quirsur-tar-puq

cough-ITERATIVE-3rd.SG.INDIC

‘He coughed repeatedly’

c. West Greenlandic habitual

qimmi-t qilut-tar-put

dog.PL bark-HABITUAL-3rd.INDIC

‘Dogs bark’

112 INTRODUCING MORPHOLOGY

There are other sorts of aspectual distinctions that can be made as well,

but these illustrate at least the main types of aspect that can be found in

the languages of the world.

English can make some of the aspectual distinctions mentioned above,

but it does so periphrastically, using a combination of extra verbs, preposi-

tions, and adverbs to convey such nuances of meaning. In other words,

many of these aspectual differences are expressed lexically rather than

inflectionally:

(19) Inceptive She began to walk.

Habitual She always/usually walks.

Continuative She keeps on walking.

Iterative She reads over and over.

Tense and aspect can be and frequently are combined in languages, so

for example, it is possible to speak of an event that is ongoing in the past

or the future. As mentioned above, tense and aspect are often combined

in English. The past tense in English is also typically perfective: when we

say she walked we generally speak of an event that is conceived of as com-

pleted. But when we use the past progressive, as in she was walking , we are

talking about something that happened in the past, but which we are

thinking of as ongoing. The present tense in English can be used to signal

the present moment ( At this very moment, a dog barks ), or also to convey

habitual aspect ( Dogs bark ) .

6.2.6 Voice Voice is a category of inflection that allows different noun phrases to be

focused in sentences. In the active voice in a sentence with an agent and a

patient, 2 the agent is focused by virtue of being the subject of the sentence:

(20) The cat chased the mouse.

But in the passive voice the patient is the subject of the sentence, and it

gets the focus:

(21) The mouse was chased (by the cat).

In English the passive is expressed periphrastically by a combination of

the auxiliary verb be plus the past participle, in example (19) chased , but in

Latin active and passive forms of the verbs are distinguished by inflec-

tional suffixes:

(22) Active

singular 1st amō plural 1st amāmus 2nd amās 2nd amātis 3rd amat 3rd amant

Passive

singular 1st amor plural 1st amāmur 2nd amāris 2nd amāminī 3rd amātur 3rd amantur

2. See Chapter 8 , section 2, for a discussion of the terms agent and patient .

Inflection 113

There is a great deal more to be said about voice distinctions like active

and passive, and we will return to them in some detail in Chapter 8 .

6.2.7 Mood and modality The inflectional categories of mood and modality have to do with a range

of distinctions that include signaling the kind of speech act in which a

verb is deployed. Speech acts are classically defined as things we can do

with words, for example, making a statement, asking a question, or giving

a command. Languages often have three moods: declarative for making

ordinary statements, interrogative for asking questions, imperative for

giving commands. But some languages can have other moods as well, for

example, expressing a speaker’s attitude about a statement, including

whether it is necessary, possible, or certain.

The now-extinct language Tonkawa (Coahuiltecan) had eight suffixes

signaling different moods/modalities (Mithun 1999 : 171). Mood suffixes

are shown in bold:

(23) Declarative naxadjganaw- o - ‘o · ’ ‘I married’

Assertive do · nan- a’a ‘He lies!’

Exclamatory ‘awac’a · la hedoxa- gwa ‘The meat is all gone!’

Interrogative yaxa- ’ -ga? ‘Did you eat?’

Intentive heul- a·ha’a ‘I shall catch him’

Imperative ‘andjo- u ‘Wake up!’

Potential ya · dj- ’a -n’ec ‘I might see him’

Exhortative hama’amdo · xa · dew- e·l ‘Let him be burned up’

Another interesting distinction in mood/modality is the realis/irrealis

distinction that is marked in some Native American languages. If the rea-

lis form is used, the speaker means to signal that the event is actual, that

it has happened or is happening, or is directly verifiable by perception.

The irrealis form, in contrast, signals something that can be imagined or

thought.

In English, we have no special inflection that signals mood; questions

are formed using syntactic means and intonation, imperatives deploy the

uninflected verb stem without any special endings. We have a remnant of

a subjunctive mood, which appears in counter-factual sentences (i.e., sen-

tences expressing something contrary to fact) like If I were an aardvark, I’d

eat insects ; in such sentences the subjunctive verb form is the same as the

plural form of the verb (so in the sentence just given, we have were rather

than was ) .

6.2.8 Evidentiality In languages that mark the inflectional category of evidentiality , speak-

ers must specify in every sentence what the source of information is,

whether firsthand knowledge because the speaker has seen or heard it,

whether it comes from someone else, whether it is assumed as common

knowledge. For example, in Wintu (Wintuan, an extinct language for-

merly spoken in Northern California) the suffix - nt�e · signals something

that the speaker knows through sensory experience, like seeing, hearing,

114 INTRODUCING MORPHOLOGY

touching or tasting. The suffix - ke· indicates that the source of information

is hearsay and - re· indicates something that the speaker has inferred from

various clues but does not know by direct personal experience (Mithun

1999 : 183–4). Central Pomo (a nearly extinct Pomoan language also spoken

in Northern California) has several clitics that express the source of infor-

mation; Mithun points out (1999: 181) that theoretically a Central Pomo

speaker could just say č�é mul to mean ‘it rained’, but it’s not likely that he

or she would say this. Instead, we might hear one of the five forms in (24) ,

where the circumstances in which any of the forms might be used is given

in parentheses:

(24) č�é mul = ʔma ‘it rained’ (it’s an established fact)

č�é mul = ya ‘it rained’ (I was there and saw it)

č�é mul = ʔdo· ‘it rained’ (I was told)

č�é mul = nme· ‘it rained’ (I heard the drops on the roof)

č�é mul = ʔka ‘it rained’ (Everything is wet)

Of course, in English we could say the same thing by adding an extra

clause (like ‘I heard that it rained’) or an adverb (‘supposedly it rained’),

but we do not have to express the source of our knowledge as a speaker

does in Wintu or Central Pomo. Aikhenvald ( 2004 : 17) estimates that only

about 25% of the world’s languages express evidentiality as an inflec-

tional category, so it is quite rare. It occurs in a number of Native

American languages as well as languages spoken in South America, the

Caucasus region, and in certain parts of the Tibeto-Burman language

family .

Challenge

Go to your university library and browse the shelves where you can

find grammars of unfamiliar languages. Find a grammar of a

language that you’ve never heard of before and see what kinds of

inflection (if any!) it has. Does it inflect nouns? Is there a case system?

If so, what kind? What kind of verbal inflections do you find? Do you

find any distinctions that we failed to cover in our survey, or other

kinds of inflection that strike you as interesting? Share your findings

with your classmates.

6.3 Inflection in English

6.3.1 What we have As we’ve seen in passing in the sections above, English is a language that

is quite poor in inflection. The distinction between singular and plural is

marked on nouns:

(25) Singular cat, mouse, ox, child

Plural cats, mice, oxen, children

Inflection 115

English has only a tiny bit of case marking on nouns : it uses the mor-

pheme -s (orthographically - ’s in the singular, - s’ in the plural) to signal

possession, the remnant of the genitive case. Pronouns, however, still

exhibit some case distinctions that are no longer marked in nouns:

(26) Nouns

singular non-possessive mother child

singular possessive mother’s child’s

plural non-possessive mothers children

plural possessive mothers’ childrens’

Pronouns

singular subject I you he/she/it

singular object me you him/her/it

singular possessive my your his/her/its

plural subject we you they

plural object us you them

plural possessive our your their

In verbs, number is only marked in the third person present tense,

where -s signals a singular subject. As we’ve seen, English verbs inflect for

past tense , but not for future, and there are two participles (present with

-ing and past with -ed ) that together with auxiliary verbs help to signal

various aspectual distinctions:

(27) Verbs

3rd person sg. present walks, runs

all other present tense forms walk, run

past tense walked, ran

progressive (be) walking, running

past participle (have) walked, run

Distinctions in aspect and voice are expressed in English through a

combination of auxiliary choice and choice of participle. The progressive,

which expresses, among other things, ongoing actions, is formed with the

auxiliary be plus the present participle:

(28) Present progressive I am mowing the lawn.

Past progressive I was mowing the lawn.

Future progressive I will be mowing the lawn.

The perfect (note that the perfect is not the same as the perfective,

which we discussed above) expresses something that happened in the past

but still has relevance to the present. This is signaled in English with the

past participle and a form of the auxiliary have:

(29) Perfect I have eaten the last piece of blueberry pie.

The passive voice in English is formed with the past participle as well,

but the auxiliary be is used instead of have :

(30) Passive I was followed by a voracious weasel.

116 INTRODUCING MORPHOLOGY

It is, of course, possible to combine various auxiliaries and participial

forms to express tense/aspect distinctions that are quite complex, as in,

for example, the past perfect progressive passive sentence I had been being

followed by a voracious weasel .

As you can see in (26) – (30) , English has both regular and irregular inflec-

tions. All of our regular inflections are suffixal, but irregular forms are

often formed by internal stem change (ablaut and umlaut) or by a combi-

nation of internal stem change and suffixation. Examples of irregular

forms are given in (31) :

(31) a. Irregular noun plurals

foot feet

mouse mice

ox oxen

child children

alumnus alumni

datum data

b. Irregular verb forms

sing sang sung

sit sat sat

swing swung swung

write wrote written

hold held held

tell told told

bring brought brought

We could no doubt think of more examples as well. It is often said that

the irregular plurals and past tenses in English form closed classes; that

is, they constitute a fixed list from which particular forms can be lost, but

to which no new forms can be added. The regular plural and past tense

endings are considered default endings . In other words, when a new

noun is added to English, its plural is formed with -s and when a new verb

is added, its past tense is formed with -ed . So if we borrow a noun from

another language or coin a completely new noun, their plurals will be

formed with the regular suffix ( fajitas, wugs ). Similarly for new verbs

( googled ).

Why do we have irregular forms? In some cases, they are the remnants

of ways of forming the plural or past tense that we no longer have today.

We saw in Chapter 5 that plurals like foot ∼ feet and mouse ∼ mice are the

remnants of a rule of umlaut that was lost at the earliest stages of English.

The irregular verbs are also a remnant of a way of inflecting verbs that

goes all the way back to the Germanic ancestor of English and even

beyond that to the way of inflecting verbs in proto-Indo-European. The

details of how that kind of inflection worked need not concern us here,

except to say that it involved internal vowel changes (ablaut). Suffice it to

say that that form of inflection is no longer used to inflect new verbs in

English.

Inflection 117

Challenge

Is it really true that we can never form new irregular verbs? Here’s an

experiment that you and your classmates can do. Ask five friends to

fill in the past tense of each nonsense verb in the following sentences:

Max likes to plite . Yesterday he _______ for two full hours.

Zelda glings very well. Years ago, her mother _______

professionally.

Sometimes we trell all night. Last weekend we ________ for two

whole days.

Compile your data with that of your classmates and try to see if

there are any patterns you can explain.

6.3.2 Why English has so little inflection You might wonder why English has so little inflection. In fact, if you study

the history of English, you’ll find that at one time English had quite a bit

more inflection than it now has. The earliest form of English, Old English,

was spoken from about 450 to 1100 CE . Old English had three genders –

masculine, feminine, and neuter – and a case system with four cases –

nominative, accusative, dative, and genitive; see (32) . Articles and adjec-

tives agreed with nouns in case and number, as (33) shows:

(32) Old English nouns

Masculine Feminine Neuter

Singular ‘stone’ ‘gift’ ‘ship’

Nom. stān giefu scip Acc. stān giefe scip Gen. stānes giefe scipes Dat. stāne giefe scipum

Plural

Nom. stānas giefa scipu Acc. stānas giefa scipu Gen. stāna giefa scipe Dat. stānum giefum scipum

(33) sē gōdan stān the. MASC.NOM good. MASC.NOM stone. NOM

sēo gōde giefu the. FEM.NOM good. FEM.NOM gift. NOM

ðæt gōde scip the. NEUT.NOM good. NEUT.NOM ship. NOM

Verb inflection was also more complex in Old English than in Modern

English. In Old English, verbs were inflected for person and number, and

were different for present tense and past tense in both the indicative and

118 INTRODUCING MORPHOLOGY

the subjunctive. In addition, some verbs were strong verbs , which means

that they show internal stem change – specifically ablaut – in some forms.

For example, in the strong verb drī fan ‘to drive’ the present tense always has a long [i] vowel, but the past has the vowel [a] in the first and third

person singular and a short [i] in the plural form and the second person

singular. Other verbs were called weak verbs ; these inflected using suf-

fixes rather than ablaut. As you can see with the weak verb d ē man ‘to judge’ in (34) , the vowel of the stem never changes, but in the past tense

there is always a suffix -d(e) :

(34) Strong Weak

Present drīfan ‘to drive’ dēman ‘to judge’ singular 1st drīfe dēme

2nd drīf(e)st dēm(e)st 3rd drīf(e)ð dēm(e)ð

plural 1st drīfað dēmað 2nd drīfað dēmað 3rd drīfað dēmað

Past

singular 1st drāf dēmde 2nd drife dēmdes(t) 3rd drāf dēmde

plural 1st drifon dēmdon 2nd drifon dēmdon 3rd drifon dēmdon

Why did English lose all this inflection? There are probably two reasons.

The first one has to do with the stress system of English: in Old English,

unlike modern English, stress was typically on the first syllable of the

word. Ends of words were less prominent, and therefore tended to be pro-

nounced less distinctly than beginnings of words, so inflectional suffixes

tended not to be emphasized. Over time this led to a weakening of the

inflectional system. But this alone probably wouldn’t have resulted in the

nearly complete loss of inflectional marking that is the situation in present

day English; after all, German – a language closely related to English –

also shows stress on the initial syllables of words, and nevertheless has not

lost most of its inflection over the centuries.

Some scholars attribute the loss of inflection to language contact in the

northern parts of Britain. For some centuries during the Old English

period, northern parts of Britain were occupied by the Danes, who were

speakers of Old Norse . Old Norse is closely related to Old English, with a

similar system of four cases, masculine, feminine, and neuter genders,

and so on. The actual inflectional endings, however, were different,

although the two languages shared a fair number of lexical stems. For

example, the stem b ō t meant ‘remedy’ in both languages, and the nomina- tive singular in both languages was the same. But the nominative plural

in Old English was b ō ta and in Old Norse b ó taR . 3 The form b ó ta happened

3. The R here is a runic character with a phonetic value close to [z].

Inflection 119

4. This is a slight simplifi cation, because some of the Latin infl ectional classes also have subclasses.

to be the genitive plural in Old Norse. Some scholars hypothesize that

speakers of Old English and Old Norse could communicate with each

other to some extent, but the inflectional endings caused confusion, and

therefore came to be de-emphasized or dropped. One piece of evidence for

this hypothesis is that inflection appears to have been lost much earlier

in the northern parts of Britain where Old Norse speakers cohabited with

Old English speakers, than in the southern parts of Britain, which were

not exposed to Old Norse. Inflectional loss spread from north to south,

until all parts of Britain were eventually equally poor in inflection (O’Neil

1980 ; Fennell 2001 : 128–9) .

6.4 Paradigms

If you’ve ever studied a foreign language – French, Latin, German, Russian

– you probably know at least intuitively what a paradigm is. A paradigm

consists of all of the different inflectional forms of a particular lexeme or

class of lexemes. Each distinct form of a lexeme exhibits a specific combi-

nation of the inflectional properties that are expressed in that language.

For convenience, you can think of a paradigm as a kind of table or grid

with cells, one for each inflected form for a given lexeme. For example, in

(32) and (34) above, I’ve shown you paradigms for the Old English nouns

‘stone’, ‘gift’, and ‘ship’, and for the verbs ‘to drive’ and ‘to judge’; these

paradigms show the various endings and stem changes that are exhibited

by nouns of different genders in the singular and plural in different cases,

and in the present and past of strong and weak verbs in different persons.

Traditionally the paradigm of a noun or adjective was called its declen-

sion and that of a verb its conjugation .

6.4.1 Inflectional classes Within a language, not all nouns or verbs may inflect in exactly the same

way; all members of a particular category will typically make the same

inflectional distinctions, for example, exhibiting case, number, or tense;

but the actual forms for particular cases, numbers, or tenses might differ

from one group of nouns or verbs to another. These different inflectional

subpatterns are called inflectional classes . In Latin , for example, nouns

generally belong to one of five inflectional classes that differ to some

extent in their inflectional suffixes: 4

(35) 1 2 3 4 5

‘star’ ‘servant’ ‘father’ ‘hand’ ‘thing’

Stem form stellā (F.) servo (M.) patr (M.) manu (F.) rē (F.) Singular

Nom. stella servus pater manus rēs Gen. stellae servī patris manūs rĕī Dat. stellae servō patrī manuī rĕī Acc. stellam servum patrem manum rem

Abl. stellā servō patre manū rē

120 INTRODUCING MORPHOLOGY

Plural

Nom. stellae servī patrēs manūs rēs Gen. stellārum servōrum patrum manuum rērum Dat. stellīs servīs patribus manibus rēbus Acc. stellās servōs patrēs manūs rēs Abl. stellīs servīs patribus manibus rēbus

Nouns that belong to the first inflectional class, traditionally called

first declension nouns, are usually feminine, and their stems always end

in a long - ā . Second declension nouns are typically masculine or neuter and have stems ending in -o . In the third declension, nouns may be of any

gender and stems typically end in a consonant. And so on.

Latin verbs fall into four inflectional classes or conjugations. Each con-

jugation is characterized by a particular vowel, called the theme vowel ,

which has no meaning, but is suffixed to the verb root to form a stem.

(36) 1 2 3 4

‘love’ ‘warn’ ‘say’ ‘hear’

Root am mon dic aud

Stem (root + am + ā mon + ē dic + i aud + ī theme vowel)

Present

singular 1st am+ō mon+e+ō dic+ō aud+i+ō 2nd am+ā+s mon+ē+s dic+i+s aud+ī+s 3rd am+a+t mon+e+t dic+i+t aud+i+t

plural 1st am+ā+mus mon+ē+mus dic+i+mus aud+ī+mus 2nd am+ā+tis mon+ē+tis dic+i+tis aud+ī+tis 3rd am+a+nt mon+e+nt dic+unt aud+i+unt

The person and number endings are attached directly to the root in the

first person singular of the first and third conjugations, and otherwise to

the stem, which consists of the root plus the theme vowel.

6.4.2 Suppletion and syncretism Suppletion and syncretism are terms that refer to relationships between

inflected forms in a paradigm. Suppletion occurs when one or more of

the inflected forms of a lexeme is built on a base that bears no relation-

ship to the base of other members of the paradigm. Consider, for example,

the verb go in English. In the present tense the base is go , of course: I, you,

we, they go; he/she/it goes . The progressive participle is going , and the past

participle gone , both built on the base go as well. The past tense of go ,

however, is a suppletive form went – that is, a base that is completely dif-

ferent from that of all the other forms. The Latin verb fer ō is notorious for its suppletive forms. Its present stem is fer (so, for example, the first person

plural form is ferimus ). However, its past tense forms are built on the stem

tul , and some of its participles are built on yet a third stem l ā t . Syncretism is another relationship we can find between the members

of a paradigm, specifically one in which two or more ‘cells’ in our inflec-

tional grid or table are filled with precisely the same form. Consider, for

Inflection 121

example, the Old English verb paradigm we looked at earlier (repeated

here for convenience):

(37)

Present drīfan ‘to drive’ dēman ‘to judge’ singular 1st drīfe dēme

2nd drīf(e)st dēm(e)st 3rd drīf(e)ð dēm(e)ð

plural 1st drīfað dēmað 2nd drīfað dēmað 3rd drīfað dēmað

Although there are distinct forms for first, second, and third person in

the singular, the same form is used for all three persons in the plural.

These forms display syncretism. Another, more complex example of syn-

cretism comes from the noun paradigms of a West Slavic language,

Sorbian , spoken in parts of Germany (Baerman 2007 ):

(38) Singular Dual Plural

Nominative žona žone žony

Accusative žonu žone žony

Genitive žony žonow žonow

Locative žonje žonomaj žonach

Dative žonje žonomaj žonam

Instrumental žonu žonomaj žonami

The chart (38) illustrates the paradigm for the noun ‘woman’. You can

see that in the singular the accusative and instrumental cases are syn-

cretic – they have exactly the same inflectional form – as are the dative

and locative cases. 5 In the dual, nominative and accusative are syncretic,

as are locative, dative, and instrumental. And in the plural, nominative

and accusative are syncretic as well. Morphologists are interested in see-

ing if there are any patterns to syncretism across languages – for example,

is it more typical in languages for plural forms of verbs to be syncretic

than singular forms? Or is it more common for nominative and accusative

forms of nouns to be syncretic than, say, nominative and dative? As yet

there is no definitive answer to these questions.

6.5 Inflection and productivity

It is often said that inflection differs from derivation in terms of produc-

tivity . We saw in Chapter 4 that some rules of word formation are more

productive than others. There are derivational affixes in English, as we

saw, that are quite dead or nearly so, others that are relatively productive,

and some that are fully productive. In contrast, rules of inflection are

almost always fully productive: every verb in English, for example, has a

5. The instrumental case is used to mark ‘instruments’; for example, in a sentence like I cut the bread with

a knife , the noun knife would be in the instrumental case. The locative case is used to mark locations; for

example, the phrase in the trees would be locative in The birds were singing in the trees.

122 INTRODUCING MORPHOLOGY

progressive form with the suffix -ing , and just about every verb can form

a past tense. I say “just about every verb” because there are occasional

verbs that native speakers of English (at least of American English) are

highly reluctant to use in the past tense: for example, I can use the verb

forgo/forego in the present tense ( I forego dessert most nights ), but for the past

tense forwent sounds too odd for most people to use either in spoken or

written form ( ??I forewent dessert last night ), and there’s no alternative.

Certainly not foregoed !

Challenge

Are there any other verbs in English that you can’t use in the past

tense? Note that what I mean here are verbs for which you might find

past tense forms in the dictionary, but would never actually use

yourself. Find a list of irregular verbs in English, either in a

dictionary, a grammar book, or online, and see if you can find some.

Then compare your list of “no past tense” verbs with your classmates’.

Is there any generalization you can make about the verbs for which

we have no past tenses?

6.6 Inherent versus contextual inflection

Consider the Latin phrases in (39) :

(39) bonus puer

good-MASC.SG.NOM boy-MASC.SG.NOM

‘good boy’

bona puella

good-FEM.SG.NOM girl-FEM.SG.NOM

‘good girl’

bonum dōnum good-NEUT.SG.NOM gift-NEUT.SG.NOM

‘good gift’

In each of these phrases the adjective and noun agree in number, gender,

and case. As you can see, the adjective ‘good’ can occur in any of the three

genders, depending on the context in which it occurs. The nouns ‘boy’,

‘girl’, and ‘gift’ are always, however, respectively masculine, feminine, and

neuter nouns. In other words, the inflectional category of gender is inher-

ent in nouns, whereas it is contextual in adjectives.

Put more generally, contextual inflection is inflection that is deter-

mined by the syntactic construction in which a word finds itself, whereas

inherent inflection is inflection that does not depend on the syntactic

context in which a word finds itself. Number is inherent in nouns and

pronouns, as is gender. But the case of a noun always depends on its syn-

tactic context. On the other hand, tense and aspect are inherent in verbs,

but person and number depend on the nouns or pronouns with which a

Inflection 123

verb occurs in a sentence. So number can be contextual for one category

(verbs) but inherent for another (nouns).

6.7 Inflection versus derivation revisited

We have now looked in some detail both at inflection and at derivation (or

lexeme formation, more generally). Remember that we distinguished the

two sorts of morphology in the following ways:

(40) Inf lection Derivation

never changes category sometimes changes category

adds grammatical meaning often adds lexical meaning

is important to syntax produces new lexemes

is usually fully productive can range from unproductive to

fully productive

One more thing we can add to these differences is that in words that

have both inflectional and derivational affixes, the derivational affixes

almost always occur inside the inflectional ones. Remember from Chapter

3 that words are formed like onions, with successive layers added one by

one. In these word structures, derivational affixes go closer to the base (or

root or stem) than inflectional affixes do. For example, in English we can

have words like kingdoms where the noun suffix -dom attaches to its base

before the plural suffix -s , or purified where the verb-forming suffix -ify

attaches to the adjective pure before the past tense suffix -ed is added:

(41)

king dom s

pure ify

It’s not possible in English to attach a plural or past tense suffix and then

a derivational suffix; we never form words like *kingsdom or *walkeder .

Challenge

We have seen that in words that have both derivational affixation and

inflection derivation occurs ‘inside’ inflection. Now consider

compounding. Is it possible for the first element in a compound in

English to bear an inflection, like a plural or a possessive suffix?

You’ll have to think carefully here. Collect relevant examples and see

if you can make up some yourselves. What do you think: can

inflection occur ‘inside’ compounds?

We’ve seen that the differences between inflection and derivation are usu-

ally quite clear, in terms of function, meaning, and position in word

124 INTRODUCING MORPHOLOGY

structures. In most cases, when we look at the morphology of languages

it’s not too hard to distinguish inflection from derivation. There are cases,

however, where the distinction doesn’t seem so clear-cut. One puzzling

case is that of nouns in the West Atlantic language Fula (Lieber 1987 : 74;

Arnott 1970 : 75). Each noun in Fula belongs to up to seven different noun

classes. One of those classes is a singular class, another a plural class, and

the remainder are classes for singular and plural diminutives, pejorative

diminutives (meaning something like ‘nasty little X’), augmentatives, and

pejorative augmentatives. The various noun class forms for ‘monkey’ are

shown in (42) :

(42) Fula waa ‘monkey’

11 waa-ndu singular

25 baa-�i plural

3 baa-ŋgel diminutive singular

5 baa-ŋgum pejorative diminutive singular (‘nasty little monkey’)

6 mbaa-kon diminutive plural

7 mbaa-ŋga augmentative singular 8 mbaa-ko augmentative plural

As (42) illustrates, noun class is marked in Fula not only by different

suffixes, but also by mutation of the initial consonant of the noun stem.

So in class 11, for example, the initial consonant of the stem is a continu-

ant [w], in classes 25, 3, and 5, the initial consonant is a stop [b] that cor-

responds in point of articulation to the continuant, and in classes 6, 7,

and 8 it is a prenasalized stop [mb], again corresponding in point of

articulation. Singular and plural, of course, are number distinctions, and

thus belong to the realm of inflection. Augmentatives and diminutives,

however, are expressive morphology (look back to Chapter 3 , section 2),

and are usually considered derivational. The whole list in (42) has the look

of a paradigm, though, and even more perplexing, adjectives and articles

agree with Fula nouns, even in the noun classes that form augmentatives

and diminutives. Agreement, of course, is a hallmark of inflection. The

point here is that it’s not at all clear in the case of Fula whether to count

augmentatives and diminutives as inflectional or derivational, or just to

concede that in this particular case the distinction just doesn’t make

sense.

An equally perplexing case can be found in the Bantu language Sesotho .

Like Fula, Sesotho has noun classes. Classes 1 and 2 are classes that con-

tain human nouns. Interestingly, agent nouns can be derived from verbs

by putting them into classes 1/2:

(43) Sesotho (Demuth 2000 : 278)

Infinitive of verb Corresponding agent noun (class 1)

ho-pheha ‘to cook’ mo-phehi ‘cook’

ho-ruta ‘to teach’ mo-ruti ‘teacher’

Formation of agent nouns from verbs certainly looks like derivation. Yet

in Sesotho, as in Fula, articles and adjectives must agree with the nouns

Inflection 125

they modify, in that they must bear the class prefixes corresponding to

those nouns. And agreement, of course, is part of inflection.

We leave this issue unresolved here – although it may not seem particu-

larly satisfying to a beginning student of morphology to know that the

distinction between inflection and derivation is not always crystal clear,

it is precisely cases like these that most intrigue morphologists!

6.8 How to: more morphological analysis

In this chapter, I have tried to expose you to some of the sorts of inflec-

tional distinctions that you might expect to find outside of English.

Reading about inflectional morphology is not the same as analyzing it

though. As a student morphologist you should be ready to figure out what

sorts of inflectional distinctions are expressed in an unfamiliar language

and how they are expressed. In this section, we will simulate such an

experience for you, and take you through the process of trying to analyze

the inflectional morphology of a language you otherwise know little or

nothing about. Our example comes from the Papuan language Yimas,

spoken in New Guinea (Foley 1991 : 217):

(44) Yimas verb forms

a. nakatay ‘I see him’

b. nantay ‘you see him’

c. nantay ‘he sees him’

d. impakatay ‘I see those two’

e. impantay ‘you see those two’

f. impantay ‘he sees those two’

g. pukatay ‘I see them (more than two)’

h. puntay ‘you see them’

i. puntay ‘he sees them’

j. naŋkratay ‘we two see him’ k. naŋkrantay ‘you two see him’ l. nampi-tay ‘those two see him’

m. impaŋkratay ‘we two see those two’ n. impaŋkrantay ‘you two see those two’ o. impampi-tay ‘those two see those (other) two’

p. puŋkratay ‘we two see them (more than two)’ q. puŋkrantay ‘you two see them’ r. pumpi-tay ‘those two see them’

s. nakaycay ‘we (more than two) see him’

t. nanantay ‘you (more than two) see him’

u. namputay ‘they (more than two) see him’

v. impakaycay ‘we (more than two) see those two’

w. impanantay ‘you (more than two) see those two’

x. impamputay ‘they (more than two) see those two’

y. pukaycay ‘we (more than two) see them (more than two)’

z. punantay ‘you (more than two) see them (more than two)’

aa. pumputay ‘they (more than two) see them (more than

two)’

126 INTRODUCING MORPHOLOGY

Analyzing the inflectional system of an unfamiliar language is not very

different from analyzing the sort of derivational data we looked at in

Chapter 3 . What we do to start off is to look at the translations of the

forms, and compare forms for areas of overlap.

If you glance even casually at the data in (44) , you’ll see that I’ve given

you part of a verb paradigm, namely the verb ‘to see’ in Yimas. You might

therefore expect to find something that occurs in every form that would

correspond to the root meaning ‘see’. Scanning the data, you will see that

almost every form except s, v, and y ends in the sequence tay . The three

forms that don’t end in tay end in cay . It would be a good initial guess,

then, that the morpheme for ‘see’ is tay . Since cay looks a lot like tay , we

might hypothesize that it also means ‘see’, although we don’t yet know

why those three forms are different. At this point, let’s set aside the three

forms with cay to think more about later.

The next thing we notice about the forms in (44) is that they vary in

the person and number of the subject: there are examples with first,

second, and third person subjects, not only in singular and plural, but

also apparently in the dual. So Yimas makes a three-way number dis-

tinction in its verb forms. We notice as well that the examples differ

from one another in the number of their object (again singular, dual,

and plural), although all of the object forms in these examples happen

to be in the third person. Given that we already suspect that the verb

root comes at the end, we would guess that subject and object are

marked on the verb with prefixes. So our next task is to start comparing

the various forms that have the same number object or the same person

and number subject to see if we can find separate prefixes for subject

and object.

Let’s start with the object forms. If you look at ( 44a–c ), you’ll notice

that all three forms begin with na- , but that (44a) has ka- after the na- ,

whereas in ( 44b, c ), there’s an n- after na- . We can hypothesize, then, that

na- must mean ‘he-object’, but we need to check ourselves to make sure

that other forms in (44) with third person singular objects begin with

na- . If we look at ( 44j, k, l ), you’ll see that our hypothesis is confirmed,

and if you look further down the data, you’ll also see na- at the begin-

ning of ( 44s, t, u ). Now, looking at ( 44d, e, f ), you’ll notice that all three

forms begin with impa- ; in (44d) this is followed by ka- and in ( 44e–f ) by

n- . We can guess then that impa- must mean third person dual object.

Finally ( 44g, h, i ) all begin with pu- ; again in (44g) pu- is followed by ka-

and in ( 44h–i ) by n- . It looks like pu- is used for a third person plural

object.

If na- is the third person object marker, then what’s left over between it

and the verb stem must be the subject marker: in (44a, b, c) , this would

leave ka- as the first person singular subject marker, and n- for both the

second and third person singular subject markers (syncretism!). Now, if

you peel off the other object markers impa- and pu- from the beginnings of

the words, and tay/cay from the end of the words, what you have left

should be all the subject markers. This is what we get:

Inflection 127

6. This is a rough statement of the rule.

(45) Yimas subject markers

1 sg. ka-

2 sg. n-

3 sg. n-

1 dual ŋkra- 2 dual ŋkran- 3 dual mpi- -

1 pl. kay-

2 pl. nan-

3 pl. mpu-

Beginning students might have the impulse to call the subject markers

‘infixes’, because they occur between the object markers and the verb

root, but they are not infixes. Remember that we only call an affix an infix

if it occurs within another morpheme. What we have here instead is a

sequence of two prefixes before the verb root. So the structure of the

Yimas verb is:

(46)

object prefix subject prefix verb root

A note about the Yimas data: we still haven’t explained why the verb

root appears to be tay in most verb forms but cay in ( 44s, v, and y ). With

the data I’ve given you, it’s actually impossible to say anything more

about these forms. This is a point at which you, as the morphologist

studying an unfamiliar language, must go looking for more data in the

hope that they will explain what’s going on. So here’s a cautionary

note: linguistic data can sometimes be a bit messy. Introductory text

books have a tendency to sanitize data so that the student linguist will

not be confronted with bits that can’t be explained. But keep in mind

that in the real world data are rarely perfectly sanitary. You often see a

pattern, but it’s not always perfect. This may be frustrating, but it isn’t

a bad thing – every bit of mess – or apparent mess – forces us, as lin-

guists, to keep looking for more data and to look more carefully at the

data we have.

A final note: if you look carefully in Foley’s ( 1991 ) grammar of Yimas, it

appears that there is a phonological rule that turns [t] to [c] if it is pre-

ceded by [y]. 6 So we were right to guess that tay and cay are the same mor-

pheme. In Chapter 9 , we’ll see more data like this, and learn how to han-

dle them.

128 INTRODUCING MORPHOLOGY

Summary In this chapter we’ve first surveyed quite a few sorts of inf lection that can be found in the languages of the world, looking at person,

number, gender and noun class, tense and aspect, voice, mood,

modality, and evidentiality. We’ve looked in some detail at the sorts of

inf lections that are found in English, and considered the historical

reasons why English has relatively little in the way of inf lection. We’ve

then looked at paradigms and important relationships between forms

in paradigms, such as suppletion and syncretism, and at the

distinction between inherent and contextual inf lection. We have

revisited the distinction between inf lection and derivation to see that

the line between them can be blurred. And finally, we’ve looked at a

set of data to see how to figure out how the inf lection in an

unfamiliar language works.

Exercises 1. Look at the following data. In each case, identify the form of the

morphological rule (i.e., prefixation, suffixation, infixation, reduplication, internal stem change, templatic) and its function (inflection or derivation):

a. Turkish (Kornfilt 1997 : 446) silâh ‘weapon’ silâhlı ‘armed person’ at ‘horse’ atlı ‘horseman’ yaş ‘age’ yaşlı ‘aged person’

Londra ‘London’ Londralı ‘person living in London’

b. Musqueam (Suttles 2004 : 139) p ’ ét ’θ ‘sew’ p ’ ép ’ ət ’θ ‘be sewing’ k ’w éc ‘look’ k ’w ék ’w əc ‘be looking’

�ás ‘fish with a net’ �á�əs ‘be fishing with a net’

c. Hausa (Newman 2000 : 454) ts ā ̀kō ‘chick’ ts ā ̀kī ‘chicks’ kw ā�̀ ō ‘frog’ kw ā�̀ ī ‘frogs’ z ā̀ bō ‘guinea-fowl’ z ā ̀bī ‘guinea-fowls’

2. In the two columns below, you find verb bases and imperfective forms for a number of verbs in Tagalog , an Austronesian language spoken in the Philippines (Schachter and Otanes 1972 : 365):

a. Verb base Imperfective lagyan ‘put in/on’ nilalagyan ‘is putting in/on’ regaluhan ‘give a present to’ nireregaluhan ‘is giving a present to’

walisan ‘sweep’ niwawalisan ‘is sweeping’

Write a rule that shows how the imperfective is formed in Tagalog. Now consider these data:

b. Verb base Imperfective lagyan linalagyan regaluhan rineregaluhan walisan winawalisan

Inflection 129

Apparently the imperfective forms in (b) are less preferred, but possible, imperfective forms for the same verbs. Write an alternative rule that accounts for these forms.

Finally, what are the two imperfective forms that you might expect from the verb stem yapakan ‘step on’?

3. Consider the following data from Swahili (Corbett 1991 : 43–4):

kikapu kikubwa kimoja kilianguka basket large one fell

vikapu vikubwa vitatu vilianguka baskets large three fell

Describe how number marking works in Swahili. On which categories is number marking inherent and on which is it contextual?

4. In Russian , both the noun student ‘student’ and the noun dub ‘oak’ are masculine, but there are slightly different declensions for animate and inanimate nouns. Discuss the paradigms below in terms of the patterns of syncretism they display (data from Corbett 1991 : 166):

Singular student ‘student’ dub ‘oak’ Nominative student dub Accusative studenta dub Genitive studenta duba Dative studentu dubu Instrumental studentom dubom Locative studente dube

Plural Nominative studenty duby Accusative studentov duby Genitive studentov dubov Dative studentam dubam Instrumental studentami dubami Locative studentax dubax

5. Consider the data below from Syrian Arabic (Cowell 1964 : 173–4). Segment the words into morphemes, and identify the meanings/ functions of the morphemes. What word formation processes are represented in these data? (Hint: Assume that the form for ‘he ate’ is á k a l , that this form has neither prefixes nor suffixes, and that a glottal stop always occurs before a vowel-initial stem.)

ʔákal ‘he ate’ byaakol ‘he eats’ ʔáklet ‘she ate’ btaakol ‘she eats’ ʔákalu ‘they ate’ byaaklu ‘they eat’ ʔakált ‘you (masc.) ate’ btaakol ‘you (masc.) eat’ ʔakálti ‘you (fem.) ate’ btaakli ‘you (fem.) eat’ ʔakáltu ‘you (pl.) ate’ btaaklu ‘you (pl.) eat’ kol ‘eat! (masc.)’

6. Below are the Latin verb paradigms for the verbs ‘love’ and ‘warn’ in the future and perfect tenses. Identify the morphemes in each form (root, stem, suffixes) and discuss how the future and perfect tenses differ from one another, and how the first and second conjugation verbs differ from one another.

130 INTRODUCING MORPHOLOGY

amābō ‘I will love’ monē bō ‘I will warn’ amābis ‘you will love’ monē bis ‘you will warn’ amābit ‘he/she will love’ monē bit ‘he/she will warn’ amābimus ‘we will love’ monē bimus ‘we will warn’ amābitis ‘you (pl.) will love’ monē bitis ‘you (pl.) will warn’ amābunt ‘They will love’ monē bunt ‘they will warn’ amāvı̄ ‘I loved’ monuı̄ ‘I warned’ amāvistı̄ ‘you loved’ monuistı̄ ‘you warned’ amāvit ‘he/she loved’ monuit ‘he/she warned’ amāvimus ‘we loved’ monuimus ‘we warned’ amāvistis ‘you (pl.) loved’ monuistis ‘you (pl.) warned’ amāvē runt ‘they loved’ monuē runt ‘they warned’

7. Dutch makes a distinction between weak and strong verbs. Below are the paradigms for the past tense of a weak verb ( werken ‘to work’) and a strong verb ( binden ‘to tie’). Discuss differences in the ways that weak and strong past tenses are formed in Dutch. Are the patterns of syncretism the same or different in the weak and strong forms?

ik werkte ‘I worked’ ik bond ‘I tied’ jij werkte ‘you worked’ jij bond ‘you tied’ hij, zij werkte ‘he/she worked’ hij/zij bond ‘he/she tied’ wij werkten ‘we worked’ wij bonden ‘we tied’ jullie werkten ‘you (pl.) worked’ jullie bonden ‘you (pl.) tied’ zij werkten ‘they worked’ zij bonden ‘they tied’

8. Consider the sets of verbs below from the Yuman language Diegueño (Langdon 1970 : 80–7).

a. a˙ap ‘to lay down a long object’ a˙kat. ‘to cut with a knife’ a˙ma� y ‘to sweep’ a˙nar. ‘to lower a long object, to drown’ a˙mar. ‘to cover over a long object, to bury someone’ a˙u� ‘to lay a long object on top of ’

b. cu˙kat. ‘to bite off ’ cu˙par. ‘to emit a victory yell’ cu˙kuw ‘to bite’ cu˙ya˙y ‘to hum’ cu˙sip ‘to smoke’ (e.g., a pipe) cu˙k w is ‘to chatter (like squirrel)’

c. tu˙kat. ‘to cut with scissors or ax, to cut in chunks’ tu˙mi� ‘to hang (small round object)’ tu˙pa˙ ‘to crack acorns’ tu˙u� ‘to put on’ (e.g., a hat)’ tu˙mar. ‘to cover over a small object’ tu˙yum ‘to put a round small object in sun’

a. Divide the words above into prefixes and roots and try to assign meanings to each morpheme.

b. Is the process you see illustrated here one of inflection or derivation? Give evidence to support your answer.

9. The following data are from Na’vi , a constructed language (conlang) created by the linguist Paul Frommer for the movie Avatar . Divide the words below into morphemes and discuss the position of the

Inflection 131

morphemes with respect to the various verb roots (data from http://files. learnnavi.org/docs/horen-lenavi.pdf ). Note that the verb ‘feed’ is a compound word.

eyk fpak taron yomtìng ‘lead’ ‘stop’ ‘hunt’ ‘feed’

near past ìmeyk fpìmak tìmaron yomtìmìng reflexive äpeyk fpäpak täparon yomtäping refl. near past äpìmeyk fpäpìmak täpìmaron yomtäpìmìng ceremonial uyeyk fpuyak taruyon yomtuyìng perf. cerem. oluyeyk fpoluyak tolaruyon yomtoluyìng refl. perf. cerem. äpoluyeyk fpäpoluyak täpolaruyon yomtäpoluyìng

10. Consider the data below from Central Guerrero Nahuatl , a Uto-Aztecan language spoken in Mexico (data from Amith and Smith-Stark 1994 : 349–51). Divide the words into morphemes and identify what each morpheme means. Discuss any difficulties you encounter.

a. tine:čλasoλa ‘you (sg.) love me’ b. ne:č λasoλa ‘he/she/it loves me’ c. nane:č λasoλan ‘you (pl.) love me’ d. ne:č λasoλan ‘they love me’ e. timi¢λasoλa ‘I love you (sg.)’ f. mi¢λasoλa ‘he/she/it loves you (sg.)’ g. timi¢λasoλan ‘we love you (sg.)’ h. mi¢λasoλan ‘they love you (sg.)’ i. nikλasoλa ‘I love him/her/it’ j. tikλasoλa ‘you (sg.) love him/her/it’ k. kiλasoλa ‘he/she/it loves him/her/it’ l. tikλasoλan ‘we love him/her/it’ m. nankiλasoλan ‘you (pl.) love him/her/it’ n. kiλasoλan ‘they love him/her/it’ o. tite:č λasoλa ‘you (sg.) love us’ p. te:č λasoλa ‘he/she/it loves us’ q. nante:č λasoλan ‘they love us’ r. nam:eč λasoλa ‘I love you (pl.)’ s. me:č λasoλa ‘he/she/it loves you (pl.)’ t. tame:č λasoλan ‘we love you (pl.)’ u. me:č λasoλan ‘they love you (pl.)’ v. nikinλasoλa ‘I love them’ w. tikinλasoλa ‘you (sg.) love them’ x. kinλasoλa ‘he/she/it loves them’ y. tikinλasoλan ‘we love them’ z. nankinλasoλan ‘you (pl.) love them’ aa. kinλasoλan ‘they love them’ bb. tine:č kone:w ‘I am your child’ cc. ne:č kone:w ‘I am his/her child’ dd. nane:č kone:w ‘I am your (pl.) child’ ee. ne:č kone:w ‘I am their child’

KEY TERMS isolating

analytic

agglutinative

fusional

polysynthetic

index of synthesis

index of fusion

index of exponence

head-marking

dependent- marking

implicational universal

CHAPTER OUTLINE

In this chapter you will learn about morphological typology:

how morphologists characterize the morphological systems

of languages.

♦ We will begin by describing the morphological systems

of five very different languages, looking at the kinds of

lexeme formation and inflection that they display.

♦ Then we will discuss both traditional ways of classifying

the morphology of languages and more contemporary

ways of doing so.

♦ Finally, we will look at how both the family a language

belongs to and the geographic area in which it is spoken

can influence its typological classification.

Typology7 C H A P T E R

134 INTRODUCING MORPHOLOGY

7.1 Introduction

In this book, we have focused so far on formal processes of morphology

that occur in many languages of the world; we have provided a sort of

toolkit for forming new words from which languages can pick and choose.

What we haven’t looked at yet is what we could call “the big picture”: how

does word formation work overall in specific languages and how can the

morphological systems of particular languages vary from one another? In

other words, rather than looking at specific processes and how they work,

we can look at how languages exploit different parts of our toolkit to con-

stitute their own unique systems of morphology. We can try to character-

ize the morphological systems of languages according to the sorts of

morphological processes that they exploit. We can compare languages to

each other to see if characteristics of their morphological systems corre-

late in any way with other parts of their grammars, say their syntax or

their phonology. And we can look at the distribution of morphological

patterns in terms of genetic relationships (language families) or areal

tendencies (where languages are spoken in the world). Studying language

from this sort of global perspective is the subject of linguistic typology .

7.2 Universals and particulars: a bit of linguistic history

The history of linguistics has for centuries seen a tug of war between

theories that emphasize universals – those things that are common to all

human languages, perhaps because they are part of our common biologi-

cal endowment – and particulars – those things that look unique and

appear to distinguish languages from one another.

At the turn of the twentieth century, partly as a legacy of colonialism,

linguists started studying indigenous languages of Africa, Asia, and the

Americas more seriously. In North America, the tradition of American

Structuralism stressed the uniqueness of languages, not surprising, con-

sidering the linguistic diversity of native North American languages and

their prodigious differences from one another and from more familiar

and better studied Indo-European languages. With the advent of Generative

Grammar in the middle of the twentieth century, the pendulum has

swung in the other direction. Chomskians stress what’s universal in lan-

guages, and search for ways to explain linguistic differences as the result

of small choices that languages make from a universal set of options that

our biological make-up, our hard-wiring for language as it were, makes

available.

Understanding this universal set of options is ever more important

today, with renewed efforts among linguists to study the many languages

that are endangered. Universals and particulars are both important: until

we have a sense of the full range of particulars, we can only begin to con-

front the issue of universals. That’s why studying the widest range of

languages possible is so important.

Typology 135

Do we know anything about morphological universals? Yes – and you’ve

gotten a taste of what we know here. We know, for example, that there is a

range of word formation strategies that appear in the languages of the

world. And there are some conceivable sorts of word formation strategies

that never occur. We know, for example, that there’s no language so far

that forms one sort of word from another – say nouns from verbs or verbs

from nouns – by reversing the sounds of the words, or by infixing [p] after

every third sound. But there are a lot of things we don’t know – what are

possible forms of reduplication or infixing, for example, and what is

impossible. So the search for particulars and universals goes on in tandem.

7.3 The genius of languages: what’s in your toolkit?

Students of linguistics often have a sense of excitement when they come

upon data from languages they’ve never heard of before and discover how

very different languages can look from one another. Although linguists in

the generative tradition are always quick to stress that languages are more

alike underlyingly than they seem superficially, what often strikes stu-

dents first are the wonderfully exuberant ways in which languages can do

things differently. Some of what gives this impression of difference is the

unique way in which the morphology of languages can package different

concepts in different forms.

The linguist Edward Sapir , writing at the turn of the twentieth century,

had a rather romantic name for the unique combination of processes that

characterize the grammar of each language – he called it the “genius” of

the language (Sapir 1921 : 120):

For it must be obvious to anyone who has thought about the question at

all or who has felt something of the spirit of a foreign language that there

is such a thing as a basic plan, a certain cut, to each language. This type or

plan or structural “genius” of the language is something much more

fundamental, much more pervasive than any single feature of it that we

can mention, nor can we gain an adequate idea of its nature by a mere

recital of the sundry facts that make up the grammar of the language.

These days, linguists might find quaint Sapir’s idea that each language is

imbued with something like a special spirit or soul that embodies its ‘basic

plan’. But Sapir’s idea of “genius” comes close to that feeling that students

have that the new languages they encounter are in some sense new creatures.

In this section we take a brief look at five very different languages –

Turkish , Mandarin Chinese , Samoan, Latin, and Nishnaabemwin – to try to

see something of this unique combination of morphological processes that

constitutes at least one part of the genius of each language. All of these

languages use morphology in one way or another, but each makes different

choices from the universal toolbag of rule types that we have surveyed so far

in this book. Some use predominantly one strategy, others many; some have

lots of inflection, others almost none. But each has its own unique pattern.

136 INTRODUCING MORPHOLOGY

1. Inkelas and Orgun ( 1998 ) give this word in phonetic transcription. Here it is rewritten in Turkish orthography,

as nothing hinges on the pronunciation.

7.3.1 Turkish (Altaic) Imagine one word that means ‘were you one of those whom we are not

going to be able to turn into Czechoslovakians?’ This may seem highly

unlikely, but it’s possible in Turkish: the word is çekoslovakyal � laş t � ramayac aklar � m � zdanm � yd � n � z , and it’s possible because Turkish is a language that delights in suffixation:

(1) Turkish (Inkelas and Orgun 1998 : 368) 1

çekoslovakya - lı - laş - tır - ama - yacak - lar - ımız -

Czechoslovakia - from - become - CAUSE – unable - FUT - PL - 1PL -

dan - mı - ydı - nız

ABL INTERR - PAST - 2PL

Let’s look in detail at the pieces that make up this word. Note first that

Turkish has a phonological rule called ‘vowel harmony ’ which makes the

vowels of suffixes agree with the preceding vowels in the base in backness

and sometimes roundness (we’ll look more closely at this rule in Chapter 9 ).

The first suffix that we encounter after the base is a suffix -li , which attach-

es to nouns to make personal nouns. With vowel harmony, the suffix -li will

show up as -lu after the front round vowel ö:

(2) -li (-lu) personal nouns (Lewis 1967 : 60)

şehir ‘city’ şehir-li ‘city dweller’

köy ‘village’ köy-lu ‘villager’

The next suffix -laş forms intransitive verbs from adjectives. Again,

taking vowel harmony into account, the suffix can appear as -leş if it

is preceded by a base with front vowels, and -laş if preceded by back

vowels:

(3) -laş (-leş) intransitive verbs (Lewis 1967 : 228–9)

ölmez ‘immortal’ ölmez-leş ‘become immortal’

garp-lı ‘West-from’ garp-lı-laş ‘become Westernized’

Next we have the suffix -dir , which forms causative verbs from intransi-

tive verbs. Note that the [d] in the suffix shows up as the corresponding

voiceless stop [t] if the preceding consonant is voiceless:

(4) -dir (-dur, -dür, -tir, etc.) causative (Lewis 1967 : 144–5)

don ‘freeze’ don-dur ‘cause to freeze’

öl ‘die’ öl-dür ‘kill’ ( = ‘cause to die’)

The following suffix is called the ‘impotential’, which essentially means

‘not able’:

(5) -eme (-ama) impotential (Lewis 1967 : 151)

gel-eme ‘unable to come’

anlı-ama ‘unable to understand’

Typology 137

Next is the future suffix - ecek :

(6) -ecek (-acak) future (Lewis 1967 : 114)

bul-acak ‘will find’

tanı-acak ‘will recognize’

Then the plural suffix -ler :

(7) -ler (-lar) plural (Lewis 1967 : 29)

kız ‘girl’ kız-lar ‘girls’

el ‘hand’ el-ler ‘hands’

And then, we get the first person possessive suffix -ımız , the ablative

marker -dan (which essentially means ‘from’), the question marker -mı , the

locative marker -dı (which means ‘at’), and the second person plural verb

marker -nız. As you can see, each of these suffixes can occur in simpler

words, but they can all be used together to form the enormously long and

complex word we started with. If we put it all back together again, we get

something like (8) :

(8) çekoslovakya-lı ‘someone from Czechoslovakia’ =

‘Czechoslovakian’

çekoslovakya-lı-laş ‘become Czech’

çekoslovakya-lı-laş-tır ‘cause to become Czech’

çekoslovakya-lı-laş-tır-ama ‘unable to cause to become Czech’

çekoslovakya-lı-laş-tır-

ama-(y)-acak

‘will be unable to cause to become

Czech’

çekoslovakya-lı-laş-tır ama-

(y)-acak-lar

‘will be unable to cause to become

Czech-pl.’

çekoslovakya-lı-laş-tır-ama

(y)-acak-lar-ımız

‘we will be unable to cause to

become Czech.-pl’

çekoslovakya-lı-laş-tır-ama-

(y)-acak-lar-ımız-dan

‘from (those) we will be unable to

cause to become Czech-pl’

çekoslovakya-lı-laş-tır-ama-

(y)-acak-lar-ımız-dan-mı-

ydı-nız

‘were you from (those) we will be

unable to cause to become

Czech-pl.’

Although the predominant way of forming words in Turkish is through

suffixation, it also has a process of compounding , as the examples in (9)

show:

(9) Turkish compounds (Lewis 1967 : 231–3)

Noun + noun baba + anne babaanne

father mother paternal

grandmother

baş + bakan başbakan

head minister prime minister

Adjective + noun kırk + ayak kırkayak

forty foot centipede

büyük + anne büyükanne

great mother grandmother

138 INTRODUCING MORPHOLOGY

2. Li and Thompson give a few examples of prefi xes, but it's not clear why they treat them as prefi xes rather

than as the fi rst elements of compounds.

As the examples in (9) show, Turkish compounds are right-headed.

We can see from our lengthy discussion of the example in (1) that Turkish

uses suffixation for both derivation and inflection. In addition to marking

number on nouns, Turkish also marks case, as the paradigm in (10) shows:

(10) ev ‘house’

evi Definite-accusative

evin Genitive

eve Dative

evde Locative

evden Ablative

Turkish verbs are inflected for person and number, and can appear in a

number of different tenses, including present, past, future, and condi-

tional. There are affixes to make verbs negative or interrogative, and verbs

can mark other distinctions as well. All of these inflections are suffixes;

verb forms can be quite long and complex.

What this brief description of Turkish morphology shows is that a lan-

guage can have wildly abundant morphology, and yet use no more than a

couple of tools from our universal toolkit. Turkish is an overwhelmingly,

exuberantly suffixing language, using suffixes for both lexeme formation

and inflection, but it has no processes of prefixation to speak of. What few

prefixes can be found in Turkish are always on borrowed words, and essen-

tially are not part of the native system of word formation of Turkish.

7.3.2 Mandarin Chinese (Sino-Tibetan) Where the genius of Turkish morphology lies in the exuberance of its pro-

cesses of suffixation, Mandarin Chinese makes entirely different choices

from our universal toolkit, although it too concentrates on just a few tools.

According to Li and Thompson ( 1971 ), Mandarin has no processes of

prefixation to speak of, and it has only a tiny handful of suffixes, among

them the three in (11) : 2

(11) Li and Thompson ( 1971 : 41–3)

-xué dòngwù-xué ‘animal-ology’ = ‘zoology’

shèhuì-xué ‘society-ology’ = ‘sociology’

zhé-xué ‘philosophy-ology’ = ‘study of philosophy’

-jiā kēxué-jiā ‘science -ist’ = ‘scientist’

yùndòng-jiā ‘athletics-ist’ = ‘athlete’

zhèngzhì-jiā ‘politics-ist’ = ‘politician’

-huà gōngyè-huà ‘industry-ize’ = ‘industrialize’

tóng-huà ‘similar-ize’ = ‘assimilate’

The suffix -xué attaches to nouns to make nouns meaning ‘the study of

X’, and -ji ā makes personal nouns, also from other nouns. Notice that in

the example kē xué-ji ā, both suffixes occur. The third suffix - hu à makes

verbs from nouns and adjectives. Suffixation is quite limited in Mandarin

though: there are relatively few suffixes, and we certainly do not find

words of the complexity of those in Turkish or even of words in English.

Typology 139

Mandarin also has full reduplication , which it uses for two purposes.

Verbs can be reduplicated to form derived verbs meaning ‘X a little’:

(12) Li and Thompson ( 1971 : 29)

jiāo ‘teach’ jiāo-jiāo ‘teach a little’

shuō ‘say’ shuō-shuō ‘say a little’

xiē ‘rest’ xiē-xiē ‘rest a little’

And some adjectives can be reduplicated to make intensive adjectives,

that is, adjectives that mean ‘very X’:

(13) Li and Thompson ( 1971 : 33)

pàng ‘fat’ pàng-pàng ‘very fat’

hóng ‘red’ hóng-hóng ‘very red’

yuán ‘round’ yuán-yuán ‘very round’

Though Mandarin is relatively poor in affixation and reduplication, it

is incredibly rich in compounding . Mandarin has not only compound

nouns and compound adjectives, as English does, but also all sorts of com-

pound verbs, as the examples in (14) show:

(14) Examples from Li and Thompson ( 1971 : 49–55), Ceccagno (undated ms. 3–8),

and Li ( 1995 : 256)

a. [N + N] N hè-mă ‘river-horse’ = ‘hippopotamus’

hăi-gŏu ‘sea-dog’ = ‘seal’

chún-gāo ‘lip-ointment’ = ‘lipstick’

b. [N + N] N shū-guŏ ‘vegetable-fruit’ = ‘vegetables and fruit’

c. [A + A] A liàng-lì ‘bright-beautiful’ = ‘bright and beautiful’

d. [A + N] N hēi-chē ‘black-vehicle’ = ‘illegal vehicle’

zhŭ-yè ‘main-page’ = ‘home page’

e. [V + N] N jiān-shì ‘supervise-matter’ = ‘supervisor’

wén-xiōng ‘hide-breast’ = ‘bra’

f. [V + N] V dài-găng ‘wait for-post’ = ‘wait for a job’

jìn-dú ‘prohibit-poison’ = ‘ban (sale/use of) drugs’

g. [A + V] V gōng-shı̄ ‘public-show’ = ‘make public’

h. [V + V] V dă-pò ‘hit-broken’

lā-kāi ‘pull-open’

zhui-lei ‘chase-tired’

As the examples in (14) show, Mandarin has many different types of

compound, indeed, many more types than English, which as we’ve seen

has very productive compounding processes. In Mandarin, some

compounds are attributive, for example those in (14a) and (14d) . The

examples in ( 14b, c ) are coordinative compounds. And some are subordina-

tive, for example those in ( 14e, f, g ). Most of these compounds are endocen-

tric (14a, b, c, d, f, g, h) , but those in (14e) are exocentric. Some are right-

headed (14a, c, d, f, g) , some have two heads (the coordinative compounds

in (14b, c) ), and some are left-headed (14h) .

The examples we have given above all concern lexeme formation in

Mandarin because Mandarin has rather little in the way of inflection.

Nouns are not inflected for number, nor do verbs inflect for person or

140 INTRODUCING MORPHOLOGY

3. In Chapter 1 we translated guo as ‘past’, but this was not quite right. The morpheme guo marks what Li and

Thompson ( 1971 : 226) call “experiential aspect,” which signals that the speaker has experienced the event.

It is, however, often used in the context of an event that has already happened.

number. Tense is not marked morphologically and aspect is marked only

by separate particles that appear after the verb. Mandarin does have a

system of noun classifiers that are used when counting or otherwise quan-

tifying nouns, but again separate particles rather than affixes are used to

mark the class of particular nouns. So we can return to the sentences we

looked at in Chapter 1 and see them in the wider context of Mandarin

morphology:

(15) Wo jian guo yi zhi chang jing lu.

I see EXP 3 one CLASSIFIER giraffe

(16) Wo jian guo liang zhi chang jing lu

I see EXP two CLASSIFIER giraffe

The verb jian ‘see’ doesn’t change its form from one aspect to another,

nor does the noun chang jing lu ‘giraffe’ change from singular to plural. In

order to signal more than one giraffe, one needs to signify a specific num-

ber or quantity.

What this shows us is that Mandarin is just as able as Turkish to come

up with new words and to express grammatical distinctions, but its strat-

egy for doing so makes use of different means .

7.3.3 Samoan (Austronesian) What makes Samoan an interesting contrast to Turkish and Mandarin is

that it uses a wide variety of word formation processes without seeming

to favor one over another. To pursue our mechanical metaphor, its toolbag

is chock-full of different tools. In this language we can find prefixation,

suffixation, and circumfixation, both partial and full reduplication, and

also to some extent compounding. There’s also even a bit of internal stem

change in the form of a morphological process of vowel lengthening. Here

are some examples:

(17) Prefixation: fa’a ‘causative’ (Mosel and Hovdhaugen 1992 : 175–6)

alu ‘go’ fa’aalu ‘make go’

goto ‘sink’ fa’agoto ‘make sink’

ga’o ‘fat’ fa’aga’o ‘apply grease to’

māsima ‘salt’ fa’amāsima ‘salt’ = ‘put salt on’

The prefix fa’a can be put on either verbs or nouns to make verbs mean-

ing ‘cause X’ or ‘make X’ or ‘put X on’.

(18) Circumfixation: fe- -a’i ‘reciprocal’ (Mosel and Hovdhaugen 1992 : 182)

f ı̄ nau ‘quarrel’ fef ı̄ naua’i ‘quarrel with one another’

logo ‘inform’ felogoa’i ‘consult with one another’

mata ‘look’ femātaa’i ‘look at one another’

Although Mosel and Hovdhaugen say that prefixes are usually mutually

exclusive – that is, there can only be one in a word – the circumfix fe- -a’i

can occur outside the prefix fa’a-, as you see in the word in (19) :

Typology 141

(19) fe - fa’a - māfanafaa - a’i

RECIP - CAUSE - warm - RECIP

‘be of comfort to one another’

In addition to prefixes and circumfixes, Samoan can also form words by

suffixation:

(20) Suffixation: -ga forms abstract nouns from verbs (Mosel and

Hovdhaugen 1992 : 195)

amo ‘carry’ amoga ‘carrying’

a'o ‘learn’ a’oga ‘education’

savali ‘walk’ savaliga ‘a walk’

Suffixing -ga to a verb and lengthening the first vowel of the verb stem

forms another kind of derived noun which can be concrete and often,

according to Mosel and Hovdhaugen ( 1992 : 195) has a flavor of plurality:

(21) Suffixation of -ga and vowel lengthening

amo ‘carry’ āmoga ‘person(s) carrying loads on yokes’

a’o ‘learn’ ā’oga ‘school’

savali ‘walk’ sāvaliga ‘people on march’

The vowel lengthening that occurs in this process can be considered a

form of internal stem change.

Samoan is also rich in processes of reduplication , as we already saw in

section 5.4 . As we saw there, Samoan has a process of partial reduplication

that forms verbs from nouns. To repeat example (10) from Chapter 5 :

(22) Partial reduplication: N → V

lafo ‘plot of land’ lalafo ‘clear land’

lago ‘pillow, bolster’ lalago ‘rest, keep steady’

pine ‘pin, peg’ pipine ‘secure with pegs’

Partial reduplication can also be used to make ergative verbs from

nonergative verbs:

(23) Partial reduplication: V nonergative → Vergative (Mosel and Hovdhaugen

1992 : 222–3)

lo’u ‘bent’ lolo’u ‘bend’

motu ‘break (non-erg.)’ momotu ‘break (erg.)’

sa’e ‘overturn’ sasa’e ‘cause to capsize’

In both cases, partial reduplication copies the first consonant and vowel

of the base.

New words are also formed in Samoan by full reduplication. We saw one

example (9b) in section 5.4 , which is repeated in (24) , and another example

is given in (25) :

(24) Full reduplication: V → N (Mosel and Hovdhaugen 1992 : 229)

‘apa ‘beat, lash’ ‘apa‘apa ‘wing, fin’

au ‘flow on, roll on’ auau ‘current’

solo ‘wipe, dry’ solosolo ‘handkerchief ’

142 INTRODUCING MORPHOLOGY

(25) Full reduplication: frequentative/intensive

a’a ‘kick’ a’aa’a ‘kick repeatedly’

’emo ‘blink, flash’ ’emo’emo ‘twinkle’

fo’i ‘return’ fo’ifo’i ‘keep going back’

In (25) we see a process of full reduplication that takes verbs and makes

them into frequentatives , that is, forms that mean ‘X repeatedly’, or

intensives , forms that mean ‘X a lot’.

Finally, Samoan also has compounding , as the examples in (26) show:

(26) Compounding (Mosel and Hovdhaugen 1992 : 241–3)

alagā ‘source’ ‘oa ‘valuable goods’ alagā’oa ‘source of wealth’

‘aliti ‘bed’ tai ‘sea’ ‘alititai ‘seabed’

fāsi ‘piece of ’ moli ‘soap’ fāsimoli ‘piece of soap’

suā ‘liquid’ esi ‘paw paw’ suāesi ‘paw paw soup’

These examples are all left-headed endocentric attributive compounds.

As we can see, although compounding in Samoan is possible, this lan-

guage has nowhere near the richness of compound types that can be

found in Mandarin.

Interestingly, although Samoan sentences express case relations (ergative/

absolutive) and clauses are marked for tense, aspect, and mood, Samoan has

no inflectional paradigms (Mosel and Hovdhaugen 1992 : 169). In fact, rela-

tions like case, tense, aspect, and mood are expressed by independent parti-

cles, rather than by prefixes, suffixes, or reduplication, in this language;

hence most of our examples here have been derivational.

7.3.4 Latin (Indo-European) Like Turkish, and unlike Mandarin and Samoan, Latin is a heavily

inflected language. And like Turkish, its inflections are almost entirely

suffixal. 4 However, its inflection looks rather different from Turkish

inflection in that often several meanings are combined into a single

inflectional morpheme in Latin. Its toolbag is somewhat larger for deriva-

tion than for inflection, with some prefixation and compounding in addi-

tion to suffixation. Indeed, you will probably recognize elements of Latin

derivational morphology, as many of them have been borrowed into

English. We will look at inflection first, then derivation.

Latin nouns are inflected for case, number, and gender, and adjectives

are inflected to agree with them. (27) shows the paradigm for the femi-

nine noun puella ‘girl’, and (28) a noun phrase with an agreeing adjective:

(27) ‘girl’ Singular Plural

Nom. puella puellae

Gen. puellae puellārum

Dat. puellae puellı̄s

Acc. puellam puellās

Abl. puellā puellı̄s

4. Latin has a small amount of reduplication and infi xation in its verbal paradigms, but both processes occur

only in a small set of verbs, so we will not go into them here.

Typology 143

(28) bona puella good- NOM.SG . girl- NOM.SG ‘good girl’

bonae puellae good- NOM.PL . girls- NOM.PL ‘good girls’

Each inflection carries a combination of meanings that includes case,

number, and gender. For example, the morpheme - ā rum is used in the

genitive plural, and in addition, signals that this noun belongs to the first

Latin declension, almost all of whose members are feminine in gender.

Verbs have a number of different stems which form the basis of inflec-

tional paradigms that show aspect (imperfect vs. perfect) and voice (active

vs. passive), as well as person and number. A portion of the paradigms for

the verbs ‘love’ and ‘warn’ are shown in (29) (these are the same examples

you looked at in exercise 6 of Chapter 6 ):

(29) amābō ‘I will love’ monēbō ‘I will warn’

amābis ‘you will love’ monēbis ‘you will warn’

amābit ‘he/she will love’ monēbit ‘he/she will warn’

amābimus ‘we will love’ monēbimus ‘we will warn’

amābitis ‘you (pl.) will love’ monēbitis ‘you (pl.) will warn’

amābunt ‘They will love’ monēbunt ‘they will warn’

amāvı̄ ‘I loved’ monuı̄ ‘I warned’

amāvistı̄ ‘you loved’ monuistı̄ ‘you warned’

amāvit ‘he/she loved’ monuit ‘he/she warned’

amāvimus ‘we loved’ monuimus ‘we warned’

amāvistis ‘you (pl.) loved’ monuistis ‘you (pl.) warned’

amāvērunt ‘they loved’ monuērunt ‘they warned’

These verb forms are built on one of the stem forms, called the Theme

Vowel stem (a m ā -, mon ē - ) to which a future suffix -bi- or a perfect suffix -v- is

attached. Then person and number suffixes are attached. Interestingly, dif-

ferent person and number affixes are used in the past than in other tenses:

(30) Person and number suffixes in Latin verbs

Non-past

singular 1 -ō plural 1 -imus

2 -s 2 -itis

3 -t 3 -unt

Past

singular 1 -ı̄ plural 1 -imus

2 -istı̄ 2 -istis

3 -it 3 -ērunt

In the non-past, the suffixes combine person and number; in some

sense, however, the second set of suffixes also signals past tense in addi-

tion to person and number, since they are only used in the past tense.

Latin has both derivational suffixes and prefixes. For example, it forms

abstract nouns from verb roots by adding the suffix -or :

(31) timor ‘fear’ timēre ‘to fear’

amor ‘love’ amāre ‘to love’

The suffix - men attaches to either roots or theme vowel stems to form

nouns that denote the result of an action:

144 INTRODUCING MORPHOLOGY

(32) agmen ‘line of march, band’ agere ‘to lead’

certāmen ‘contest, battle’ certāre ‘to contend’

The prefix amb- attaches to verbs and means ‘around’ and the prefix

in- attaches to adjectives to form negative adjectives:

(33) ı̄ re ‘to go’ amb-ı̄ re ‘to go about’

sānus ‘sane’ in-sānus ‘insane’

Latin does not use compounding as much as English and Germanic

languages do, but it does have some compounds:

(34) flōs, flōris ‘flower’ coma ‘hair’ flōri-comus ‘ flower-crowned’

āla ‘wing’ pēs ‘foot’ āli-pēs ‘ wing-footed’

When two roots are put together into a compound, the linking vowel

-i- is used between them.

7.3.5 Nishnaabemwin (Algonquian) The final language we will look at is Nishnaabemwin, an Algonquian lan-

guage spoken in southern Ontario, Canada (Valentine 2001 ). It makes

heavy use of affixation, especially suffixation, and has an extremely rich

system of inflection. What’s most interesting about Nishnaabemwin, how-

ever, is the way that it combines bound morphemes to form new lexemes.

Let’s look first at inflection. Nouns have either animate or inanimate

gender. They can be inflected for number, and have different forms for

diminutive, pejorative, and what is called ‘contemptive’, a suffix that adds

a negative meaning, but one that is less strongly negative than the pejora-

tive suffix. Nouns can also occur in a locative form, and what is called the

‘obviative’ form, which serves to distance a noun from the speaker in a

narrative. Finally, there are prefixes and suffixes that indicate possession

of a noun. Some of the relevant forms for the noun zhiishiib ‘duck’ are

given in (35) :

(35) zhiishiib ‘duck’

zhiishiib-ag Plural

zhiishiib-an Obviative

zhiishiib-enh Contemptive 5

zhiishiib-ens Diminutive

zhiishiib-ish Pejorative

zhiishiib-ing Locative

n-zhiishiib-im ‘my duck’

These inflections are not mutually exclusive. If they occur in combina-

tion, the diminutive, for example, must precede the pejorative, which in

turn precedes the suffix that occurs in the possessive. Nor are these the

only inflections that can appear on nouns: this is just a small selection!

Verb inflection is even more complex than noun inflection, and we can-

not really do justice to it here. But just to give you a taste of how very

intricate the inflection of verbs is in Nishnaabemwin, consider Valentine’s

5. I have constructed this form and the next one. Valentine ( 2001 ) does not cite these forms himself.

Typology 145

( 2001 : 219) description of the inflectional possibilities of intransitive verbs

that take animate subjects:

For the INDEPENDENT and CONJUNCT ORDERS , there are theoretically nine

person/number/obviation categories, four MODES (and ITERATIVES in the

conjunct), two POLARITIES , three TENSES , and two functions ( VERBAL and

PARTICIPIAL in the conjunct) creating 882 inflectional combinations. For the

IMPERATIVE ORDER , there are three person/number combinations, and three

modes, creating theoretically nine inflectional combinations, making a

total of roughly 890 forms for the animate intransitive verb.

The independent form of the verb is used in main clauses, and differ-

ently inflected forms are used in subordinate clauses (the conjunct form)

and in imperatives. By polarities, Valentine means that verbs have differ-

ent forms for positive and negative. Compare this to the four forms of the

verb walk in English ( walk, walks, walked, walking ); English uses the same

forms of the verb in main and subordinate clauses, and uses a separate

word for negation.

Derivation is no less complex than inflection in this language: words

rarely consist of a single root morpheme. Instead, various bound mor-

phemes are joined together to form words. Intransitive verbs, for exam-

ple, frequently consist of two or three pieces. The last piece, called the

‘final’, expresses a verbal concept. The first piece, called the ‘initial’,

expresses something that modifies the verbal concept; in English we

would express similar concepts with separate adjectives, adverbs, or

prepositions:

(36) Examples from Valentine ( 2001 : 327) with initial giin- ‘sharp’

giin’zi ‘be sharp (of tongue)’- ANIM. SUBJ . giinaa ‘be sharp’- INANIM.SUBJ .

Verbs may also have a third piece that occurs between the initial and

the final; this is called the ‘medial’ and corresponds to what in English

would usually be a nominal concept (Valentine 2001 : 330, 334):

(37) dewnike ‘have an ache in one’s arm’ dew ‘sore’ + nik ‘arm’ + e

‘have’

bookjaane ‘have a broken nose’ book ‘broken’ + jaan ‘nose’ +

e ‘have’

gaagiijndbe ‘have a sore head’ gaagiij ‘sore’ + ndib ‘head’ + e

‘have’

Such forms can undergo further derivation by adding another final ele-

ment, so from gaagiijndbe ‘have a sore head’ we can get gaagiijndbekaazo ,

which means ‘pretend to have a sore head’ by adding the final -kaazo ,

which means ‘pretend to’. And of course each of these verbs can then take

various inflectional elements.

Nouns can be made up of several bound morphemes as well. For exam-

ple, the nouns in (38a) are made up of an initial bound element that has

an adjectival sort of meaning, and a second bound element that means

‘thing’. Those in (38b) have an initial element that is a free form, and a

bound final element that means ‘building, habitation’:

146 INTRODUCING MORPHOLOGY

(38) a. From Valentine ( 2001 : 484)

gete hii ‘old thing’ gete- ‘old’ -hii ‘thing’

shkihii ‘new thing’ oshk- ‘new’ -hii ‘thing’

b. bzhikiiwgamig ‘cowshed’ bizhikiw ‘cow’ -gamigw ‘building’

gookooshgamig ‘pig sty’ gookoosh ‘pig’ -gamigw ‘building’

Valentine uses terms like ‘initial’, ‘medial’, and ‘final’, rather than calling

the bound morphemes that make up these forms ‘prefixes’ or ‘suffixes’,

probably because these morphemes are much more like roots than like

affixes in their meanings. We might therefore think of them as bound bases.

There are some derivational suffixes in Nishnaabemwin, however. For

example the suffix -aagan creates nouns from one class of transitive verbs:

(39) naabkawaagan ‘scarf, necklace’ naabkaw ‘wear anim. noun around

neck’

noodaagan ‘employee’ noodaw ‘hire anim. noun’

And Nishnaabemwin also has compounds, which differ from the forms

in (38) by being composed of two independent stems (Valentine 2001 :

516–17):

(40) jiibaakwe-kik ‘cooking pot’ (cook + pot)

shkode-daaban ‘train’ (fire + vehicle)

waasgamg-kosmaan ‘bell pepper’ (pepper + squash)

One thing that is striking about the morphological system of

Nishnaabemwin is the overall complexity of words: words rarely consist of

a single morpheme, and frequently consist of many morphemes.

7.3.6 Summary Comparing these languages, we can see a bit of what Sapir means about

the “genius” of a language. Although we don’t need to romanticize the

unique character of each language, studying morphology opens our eyes

to the different mixture and balance of word formation processes to be

found in individual languages. Each language has a different combination

of word formation processes that gives the language its unique character.

But we should keep in mind as we wonder at all this diversity that we

should always be on the lookout for the commonalities or universals that

mark all these languages as human languages.

Challenge

Go to your university library and look for a grammar of a language

you know nothing about. Make sure the grammar you choose has a

section on morphology. Write a two- or three-page description of the

sorts of inflection and derivation that your chosen language displays,

thinking about both inflection and derivation, and about the

different kinds of word formation rules your language displays. We

will look at these again shortly.

Typology 147

7.4 Ways of characterizing languages

Up to this point we have viewed the morphological systems of languages in

an impressionistic way, looking at the combination of inflectional and deri-

vational processes that give the language its overall morphological pattern.

Morphologists continually seek to go beyond simple descriptions, however.

In looking at the morphological systems of individual languages, we are

always looking for patterns. We are interested in what sorts of morphologi-

cal rules we might expect to find in languages and what they tell us about

the general faculty of human language. We are also interested in classifying

languages by looking at whether particular sorts of traits cluster together,

and whether those clusters tell us something deeper about the nature of

language. We will return to the first of these considerations in the last chap-

ter of this book. Here, we will look more closely at different ways we may

classify the morphology of languages, and how various classifications illumi-

nate the nature of language. This latter enterprise is called typology . We

start this section by looking at a very traditional way of classifying languag-

es, and then look at more contemporary schemes of morphological typology.

WALS: The World Atlas of Linguistic Structures

A great resource for student linguists is the World Atlas of Linguistic

Structures, or WALS, which can be found online at http://wals.info . WALS

is a website that provides information on all sorts of phonological,

syntactic, and morphological characteristics of the languages of the

world, surveying everything from phonological inventories to types of

clauses and illustrating the cross-linguistic results in easily accessible

maps. Each of its 150 or so chapters covers a particular characteristic of

language and provides an explanation of that phenomenon along with

examples. For morphology, we can find lots of information on inflection,

especially on the expression of person, number, gender, case, tense,

aspect, and number in the languages of the world.

For example, chapter 26 surveys the extent to which inflectional

morphology is expressed by prefixes or suffixes (or by both or

neither). In the vast majority of languages that exhibit inflection, it is

expressed by suffixes rather than prefixes. However, there is a

concentration of strongly prefixing languages in sub-Saharan Africa

and a cluster of languages that use neither prefixes nor suffixes for

inflection in Southeast Asia (these languages display little or no

inflection at all) but even a quick glance at the map accompanying

chapter 26 shows the prevalence of suffixing.

Chapter 27 looks at the extent to which reduplication – either partial

or full – can be found in the languages of the world. Of the 368 languages

surveyed for this chapter, 278 of them are said to have both full and

partial reduplication, 35 to have only full reduplication, and 55 to have

no productive reduplication at all. English is one of this last group, so

WALS allows us to see one way in which our language is quite atypical!

148 INTRODUCING MORPHOLOGY

7.4.1 The fourfold classification Morphological typology has a long history, going back at least to the early

nineteenth century in the work of Wilhelm von Humboldt ( 1836 ; refer-

ence in Comrie 1981/1989). In this tradition, also developed by the linguist

Edward Sapir , who we mentioned above, it was common to divide lan-

guages into four morphological types: isolating (or analytic), agglutina-

tive, fusional, and polysynthetic.

An isolating or analytic language is one in which each word consists of

one and only one morpheme. Vietnamese is often cited as an example of

an isolating language. For example, nouns do not inflect for plurality. The

noun dô , ng hô

, means ‘watch’ or ‘watches’. If one wants to be specific about

how many watches are in question, it is possible to use a numeral and

then a noun classifier before the noun (Nguyen and Jorden 1969 : 119), as

(41a) shows:

(41) a. hai cái dô , ng hô

two CL watch

‘two watches’

b. Mai tô i làm cái d-ó Tôi làm cái d-ó hô m qua.

tomorrow I do CL that I do CL that yesterday

‘I will do that tomorrow’ ‘I did that yesterday’

Similarly, as (41b) illustrates, verbs do not inflect for tense in Vietnamese.

Instead, if one wants to be specific about the time of an event, it is neces-

sary to use specific adverbs like ‘tomorrow’ or ‘yesterday’.

Of the languages we have profiled in section 7.3 , Mandarin comes clos-

est to being an isolating language. Although Mandarin has abundant

compounding, it has little that would count as morphological inflection.

Like Vietnamese, plurality, tense, and aspect are all expressed by separate

words.

Unlike isolating languages, agglutinative languages have complex

words. Furthermore, those words are easily segmented into separate mor-

phemes and each morpheme carries a single chunk of meaning. In our

grammatical sketches in 7.3, Turkish is the language that comes closest to

an agglutinative ideal. For example, we gave the various case forms of the

Turkish noun ‘house’ in (10) , repeated here (42) :

(42) ev ‘house’

evi Definite-accusative

evin Genitive

eve Dative

evde Locative

evden Ablative

To form the plurals of these nouns, all one needs to do is add the mor-

pheme -ler , which goes after the root and before the case endings:

(43) evler ‘house- PLURAL ’

evleri Plural definite-accusative

Typology 149

evlerin Plural genitive

evlere Plural dative

evlerde Plural locative

evlerden Plural ablative

The two sorts of morphemes are easily separated in terms of both form

and meaning.

A fusional language, like an agglutinative language, allows complex

words, but its morphemes are not necessarily easily segmentable: several

meanings may be packed into each morpheme, and sometimes it may be

hard to decide where one morpheme ends and another one starts. Latin is

a good example of a fusional language . We can, for example, compare the

noun paradigm in Latin with that in Turkish. Whereas it is easy in Turkish

to separate off one morpheme that means ‘plural’ and another that

means ‘genitive’, it is not possible find separate morphemes that go with

those concepts in Latin. Let’s look again at the paradigm for ‘girl’ in Latin

in example (27) , repeated here:

(44) ‘girl’ Singular Plural

Nom. puella puellae

Gen. puellae puellārum

Dat. puellae puellı̄s

Acc. puellam puellās

Abl. puellā puellı̄s

If we assume that the root for the noun ‘girl’ is puell , the best we can do

is to say that the morpheme that means genitive singular is -ae and the

one that means genitive plural is -ārum . Each of these also carries gender

information (remember that nouns with these endings are most often

feminine) as well. But there is no way of separating part of these mor-

phemes into smaller pieces that mean ‘genitive’ or ‘singular’ or ‘plural’ or

‘feminine’). This is the hallmark of fusional morphology.

The final morphological type is called polysynthetic . In a polysynthetic

language words are frequently extremely complex, consisting of many

morphemes, some of which have meanings that are typically expressed by

separate lexemes in other languages. In our grammatical sketches in sec-

tion 7.3 , Nishnaabemwin is a language that could easily be characterized

as polysynthetic. Remember that it not only has an intricate inflectional

system, but forms complex words out of two or more bound bases. We

repeat the examples from (37) here as an illustration of polysynthesis

(Valentine 2001 : 330, 334):

(45) dewnike ‘have an ache dew ‘sore’ + nik ‘arm’ + e ‘have’

in one’s arm’

bookjaane ‘have a broken nose’ book ‘broken’ + jaan ‘nose’ + e ‘have’

gaagiijndbe ‘have a sore head’ gaagiij ‘sore’ + ndib ‘head’ + e ‘have’

These forms can then act as bases for all the inflectional affixes that

attach to intransitive verbs in Nishnaabemwin.

150 INTRODUCING MORPHOLOGY

7.4.2 The indexes of synthesis, fusion, and exponence The problem with the traditional fourfold classification is that languages

rarely fall neatly into one of the four classes. For example, English is not

quite an isolating language – it has some inflection – but it is certainly

not an agglutinating or a fusional language (Old English was much closer

to being a fusional language, though). Another problem is that sometimes

the inflectional system of a language falls into one category, but the deri-

vational system fits better into another. English again can serve as an

illustration: English derivational morphology is actually not that far from

being agglutinating, as an example like operationalizability ( operat-ion-al-iz-

able-ity ) suggests.

One way of dealing with these problems is to give up the fourfold clas-

sification in favor of three different scales which can be used to gauge

different aspects of the overall morphological behavior of a language: the

index of synthesis, the index of fusion, and the index of exponence.

The index of synthesis looks at how many morphemes there are per

word in a language. Isolating languages will have few morphemes per

word – in the most extreme cases, only one morpheme per word.

Agglutinative or polysynthetic languages, however, will typically have

many morphemes per word. And because this is a scale, languages like

Samoan or English can fall somewhere in-between the extremes.

The index of fusion is a measure of the degree to which morphemes

are phonologically separable from their bases. In languages with a low

degree of fusion, the phonological boundaries between morphemes are

clear either because the languages are isolating in the classical sense or

because they are agglutinative. In languages higher on the index of

fusion, the phonological boundaries become increasingly less clear.

Languages like Mandarin or Turkish would therefore be low on this

index. Languages like Arabic or Hebrew with templatic (root and pattern)

morphology would be high on the index of fusion, and languages like

Latin would be somewhere in-between.

The index of exponence measures how many meanings or inflectional

features can be expressed simultaneously in a single morpheme in a lan-

guage. We might expect, typically, for each morpheme to express a single

meaning or feature – past tense, say, or plural. But in many languages a

single morpheme might combine several meanings. Consider the example

in (46) from the Chapacura-Wanham language Wari’ , spoken in Brazil

(WALS, chapter 21, Everett and Kern 1997 : 339):

(46) Toc na com

drink. SG 3SG.REAL.NONFUT.ACTIVE water

‘He is drinking water’

In the Wari’ example, the first morpheme toc combines the meaning

drink with singular. The second na combines third person with three

other inf lectional meanings. This language, then, allows what is known

as cumulative exponence . Turkish, however, has a tendency to be low

on the index of exponence, since typically each morpheme conveys only

one meaning or inf lectional feature. WALS uses the three indexes to

Typology 151

look at inf lectional morphology in the languages of the world. But

there is no reason why we could not also look at the derivational and

inf lectional morphologies of a language separately and see where they

fit on the scales. In terms of inf lection, English would be low on the

index of synthesis, but we might place it higher on that scale if we’re

looking at English derivation, since many words in the language are

formed by compounding, prefixation, or suffixation. Similarly, we

might class English higher on the index of exponence if we’re looking

at verbal inf lection (the suffix -s carries the meanings ‘third person’,

‘present’, and ‘singular’ packed together in a form like walks ) than if

we’re looking at derivation, where each morpheme typically has one

distinct meaning.

Challenge

Take a look at the grammatical sketch of an unfamiliar language that

you made earlier in this chapter. How would you characterize your

language in terms of the fourfold classification? Where would you

place it in terms of the index of synthesis, the index of exponence,

and the index of fusion? Does your language pose any special

problems for these means of classification?

7.4.3 Head- versus dependent-marking Above, we have looked at the morphologies of languages in terms of the

ease with which words can be segmented and the relationship between

meaning and form in morphemes. There are other things we can look at,

however, in classifying and comparing languages.

One thing we can look at is the way that morphology signals the rela-

tionship between words in phrases. The main element in each syntactic

phrase is called its head ; the head of a noun phrase (NP) is the noun, the

head of a verb phrase (VP) is the verb, and so on. The other elements that

combine with the head to become a phrase might be called the depen-

dents of the head. Dependents of a noun can be adjectives, determiners,

or possessives, and dependents of a verb can be its subject or object.

Languages can choose to mark relationships between the head and its

dependents in different ways: the relationship can be marked exclusively

on the head, or exclusively on the dependent, or on both or neither. If the

relationship is marked by some morpheme on the dependent, this is

called dependent-marking , and if it is marked on the head, it is called

head-marking .

As illustrated in (47a) , the relationship between the head noun and its

possessor is marked on the possessor in English , but on the head in

Hungarian (47b) (examples from Nichols 1986 : 57):

(47) a. English

the man ’s house

dependent head

152 INTRODUCING MORPHOLOGY

b. Hungarian

as ember ház- a

the man house-3 SG .

dependent head

‘the man’s house’

The NP in English therefore shows dependent-marking between a posses-

sor and the head noun, whereas the NP in Hungarian shows head-marking.

Whole clauses may also exhibit either dependent-marking or head-

marking. In Dyirbal , NPs are case-marked to show their relationship to the

verb: 6

(48) From Nichols ( 1986 : 61)

balan d�ugumbil baŋul ya�angu baŋgu yuguŋu balgan ART.NOM woman. NOM ART.ERG man. ERG ART.INSTR stick. INS hit

dependent dependent dependent head

‘The man is hitting the woman with a stick’

Dyirbal would therefore be considered a dependent-marking language

within clauses.

In contrast, the Mayan language Tzutujil shows head-marking within

clauses: the verb is marked for the person and number of its subject and

object, but there is no marking on the subject and object themselves to

show their function in the clause (Nichols 1986 : 61):

(49) x- ∅- kee- tij tzyaq ch’ooyaa7

ASP -3SG. -3PL. -ate clothes rats

Head dependent dependent

‘Rats ate the clothes’

As I mentioned above, it is also possible for languages to show neither

head-marking nor dependent-marking. For example, a language that is

isolating and has no inflection would have neither head- nor dependent-

marking. On the other hand, it is possible for a language to have inflec-

tional markings on both the head and its dependents. Turkish , for exam-

ple, marks both the possessor and the possessed noun in an NP:

(50) ev-in kapι-sι

house- GEN door-3 SG

dependent head

‘the door of the house’

When the relationship between the dependent and the head is marked on

both constituents, we have what is called double-marking .

7.4.4 Correlations In the last sections we have seen various ways in which the morphologies

of languages can be classified. Typologists are interested in more than

classification, however. They are also interested in seeing whether there

6. Nichols ( 1986 ) considers the verb to be the head of a clause, but not all linguists agree with her on this.

Typology 153

are any predictable correlations between particular morphological char-

acteristics or between morphological characteristics and other (syntactic

or phonological) aspects of grammar. In other words, they seek to find

whether there are any patterns to the kinds of morphology one finds in a

language, and if there are, why those patterns might exist.

For example, as Whaley ( 1997 : 131) points out, isolating languages usu-

ally have rigidly fixed word orders. This correlation makes perfect sense:

if a language has no morphological way of marking the function of noun

phrases in a sentence, those functions must be signaled by the position of

a noun phrase in a sentence.

Linguists such as Joseph Greenberg ( 1963 ) and Joan Bybee ( 1985 ) have

looked at many languages and observed that there are several other correla-

tions to be found. For example, Greenberg noted the two patterns in (51) :

(51) If a language has inflection, it will also have derivation.

If a language has separate morphemes for number and case, and if

both are either prefixes or suffixes, the number morpheme almost

always occurs closer to the base than the case morpheme.

Observations such as these are called implicational universals. Based

on observations of lots of languages, they are not true in every language,

but they are true in a statistically significant number of languages.

Bybee ( 1985 ) also observed a statistically significant trend in the order-

ing of inflectional affixes in the languages of the world. What she noticed

is that in languages with a number of different inflectional affixes on

verbs, those affixes tended to come in a particular order. For example, if a

language exhibits both tense and person/number affixes, the tense affix

usually comes closer to the verb stem than the person/number affix. And

if there are aspectual affixes, these tend to precede tense affixes.

We must still ask, however, why these particular correlations should

exist. Bybee, for example, claims that the order of inflectional morphemes

on verbs has something to do with their relevance to the verb itself. We

might think of this in terms of the concepts of inherent and contextual

inflection that we discussed in section 6.6 . For example, tense and aspect

are inherent categories for verbs, but person and number are contextual:

they signal agreement with one or more of a verb’s arguments (its subject

or object, for example). So inherent inflection comes closer to the verb stem

than contextual inflection. The second of Greenberg’s implicational univer-

sals might be explained in the same way. Number is an inherent inflection

on nouns, but case is contextual, and inherent marking comes closer to the

noun stem than contextual marking. Whether this is generally the case,

however, is something that will require looking at many more languages.

7.5 Genetic and areal tendencies

In addition to classifying languages on the basis of specific structural

characteristics that they display in their morphologies, we can look at

typological patterns in a more global way. There are two ways to do this.

154 INTRODUCING MORPHOLOGY

We can look at whether there are sorts of morphology that tend to be

prevalent in particular language families or sub-families. And we may

look at whether there are specific sorts of morphology that tend to be

found in certain geographic areas even among languages that belong to

different language families.

We can give several examples of genetic tendencies. If we look, for

example, at compounding in two different branches of the Indo-European

family, Italic (Romance) and Germanic, we can see an interesting pattern:

although both branches make use of compounds, the sorts of compounds

they favor are quite different. Germanic languages like English tend to

favor endocentric attributive and subordinate compounds like those in

(52a) , whereas Italic languages seem to prefer exocentric subordinate com-

pounds, and have few attributive compounds (see section 3.4 for these terms):

(52) a English

endocentric attributive: dog bed, windmill

endocentric subordinate: dishwasher, hand made

b. Italian

exocentric subordinate: lavapiatti ‘wash-dishes’ = ‘dishwasher’

Another example comes from the Bantu sub-family of languages. It is rela-

tively rare in the languages of the world for inflectional morphology to be

accomplished predominantly by prefixing. Nevertheless, as we mentioned

above, there is a large concentration of such languages in the central and

southern parts of Africa. This is the area of Africa in which we find the

Bantu languages which are in turn part of the larger Niger-Congo family.

Bantu languages frequently inflect nouns and verbs by adding prefixes.

Similarly, although the vast majority of the languages of the world display

some sort of inflection, many languages in the Sino-Tibetan language

family are isolating. A nice illustration of this can be found in the WALS

map that accompanies chapter 26.

As for areal tendencies , Whaley ( 1997 : 13) gives a fascinating example. He

points out that three languages spoken in close proximity in the Balkan

region of Europe all mark the definiteness of nouns by adding a suffix:

(53) Albanian mik-u ‘friend-the’

Bulgarian trup-at ‘body-the’

Rumanian om-ul ‘man-the’

What makes this example so interesting is that these three languages

belong to completely different sub-families of Indo-European – Albanian

forms its own branch; Bulgarian is Balto-Slavic; and Rumanian is Italic –

and none of the other languages in these three branches show definite-

ness with suffixes! Geographic proximity can be the only explanation for

the distribution of this morphological trait.

Another example of a morphological pattern that is especially preva-

lent in a particular geographic region is verbal compounding. We saw in

Chapter 3 that English rarely compounds two verbs (although there are a

few examples like stir-fry or slam-dunk ). In contrast, verbal compounds are

not at all unusual in Asia, even in genetically unrelated languages. For

Typology 155

example, although Japanese is thought to be a language isolate (it is not

related to any language family), and Mandarin Chinese is a member of the

Sino-Tibetan family, both display verbal compounds, as the examples in

(54) show:

(54) Japanese (Fukushima 2005 : 570–85)

naki-saken cry-scream ‘cry and scream’

Mandarin (Li and Thompson 1971 : 55)

mai-dao buy-arrive ‘manage to buy’

It would be interesting to explore the historical and social forces that

lead languages in the same geographic area to develop similar morpho-

logical patterns, but we will not do so here (see Heine 2014 , though, for a

brief treatment).

7.6 Typological change

Although this is not a text about historical linguistics, in the context of

typology it is important to point out that as languages change over time,

their typological characteristics can change as well. I’ll use English as an

example here, as a fair amount is known about the evolution of the Indo-

European language family in general and about the development and

change of English morphology in particular. In Chapter 6 we looked at

changes to the inflectional system of English, specifically about the loss

of inflection. To reprise, Old English had a far more complex inflectional

system than modern English does, with grammatical gender, four cases,

agreement between determiners, adjectives, and nouns, and a far more

extensive system of verb inflection than we have now. By Middle English,

much of this system had been lost. In typological terms, English moved

farther away from being fusional and synthetic and closer to being isolat-

ing or analytic.

Haselow ( 2011 : 216) shows the same trend over a much longer time span

in looking at the development of Indo-European to the Germanic branch

and ultimately to English. Historical linguists have hypothesized that

Indo-European was highly fusional. Indo-European morphology frequent-

ly involved productive processes of internal vowel change (ablaut) in roots

for purposes of both inflection and derivation. To these roots stem forma-

tives and then inflections could be added. Some of these vowel changes

survived into Germanic and ultimately into Old English, but over the

course of time these ablaut processes had become unproductive: the only

remnants we have are the strong verbs in OE from which modern English’s

few ‘irregular’ verbs evolved. In terms of derivational morphology, a few

verb and noun pairs like sing and song or ride and road survive, but it’s not

clear that speakers of modern English even see these as morphologically

related. As the Indo-European system of vowel changes evolved and

became unproductive, affixation, conversion, and compounding became

more prominent derivational processes. Modern English, then, is typo-

logically quite far from its Indo-European ancestor.

156 INTRODUCING MORPHOLOGY

In this chapter we have surveyed five languages to see what

morphological resources they make use of. Turkish is an agglutinative

language that largely relies on suffixing. Mandarin Chinese is an

inflectionally isolating language that makes heavy use of compounding

to form new lexemes. Samoan makes use of a wide range of different

word formation processes to derive new lexemes, but is rather poor in

inflection. Latin has heavily fusional inflection, and primarily derives

new lexemes using prefixes and suffixes. And Nishnaabemwin is a

polysynthetic language. We also looked at the traditional fourfold

morphological classification of languages into isolating, agglutinative,

fusional, and polysynthetic types, and at more useful typological tools

such as the indexes of synthesis, of fusion, and of exponence, and at

the distinction between head- and dependent-marking. Finally, we

looked briefly at genetic and areal tendencies in morphological

patterning.

Summary

7. The superscript numerals on the original Yay sentence in (b) above are tone markings.

8. TR means ‘transitive suffi x’; 3 tr means 3rd person transitive subject.

Exercises 1. On the basis of the data below, try to classify these languages as isolat-

ing, agglutinative, fusional, or polysynthetic.

a. Swahili (Vitale 1981 : 18)

Juma a-li-wa-piga watoto Juma 3. SG.SUBJ-PST-3.PL.OBJ -hit children ‘Juma hit the children’

Watoto wa-li-m-piga Juma Children 3. PL.SUBJ-PST-3.SG.OBJ -hit Juma ‘The children hit Juma’

b. Yay (Whaley 1997 : 127) mi 4 ran 1 tua 4 ŋwa 1 lew 6

not see CLASS snake CMPLT ‘He did not see the snake’ 7

c. Musqueam (Suttles 2004 : 28) x w -q w é-nəc-t-əs inward-penetrate-bottom- TR -3 TR 8

‘[She] punches holes in the bottom of it’

d. Old English (Baugh and Cable 2013 : 58)

On þyss-um ē aland-e cō m ūp sē

on this- NEUT.DAT.SG island- NEUT.DAT.SG came up the. MASC.NOM.SG

God-es þē ow Augustinus

God- MASC.GEN.SG. servant. MASC.NOM.SG Augustine

and his gefē r-an.

and his companion- MASC.NOM.PL

‘God’s servant Augustine and his companions came up on this island’

Typology 157

e. Náhuatl Puebla Sierra (Nida 1946 : 171 – called Zacapoaxtla Aztec there) nan-č oka-to-skih-h 2P L- cry-DU R-CON D-P L ‘you all would keep crying’

2. Consider the following paradigms from the Mayan language Tzutujil (Dayley 1985 : 87):

waraam ‘to sleep’

Perfect 1sg. in warnaq

2sg. at warnaq

3sg. warnaq

1pl. oq warnaq

2pl. ix warnaq

3pl. ee warnaq

Completive 1sg. xinwari

2sg. xatwari

3sg. (x)wari

1pl. xoqwari

2pl. xixwari

3pl. xeewari

Incompletive 1sg. ninwari

2sg. natwari

3sg. nwari

1pl. noqwari

2pl. nixwari

3pl. neewari

Identify all the morphemes in the paradigms above. On the basis of your analysis, how would you classify Tzutujil using the traditional fourfold clas- sification system?

3. On the basis of the data below, try to classify these languages as head- marking, dependent-marking, or double-marking.

a. Chechen (Nichols 1986 : 60)

de:-n a:xč a

father- GEN money

‘father’s money’

b. Huallaga Quechua (Nichols 1986 : 72)

hwan-pa wasi-n

John- GEN house-3

‘John’s house’

c. Abkhaz (Nichols 1986 : 60, from Hewitt 1979 : 116)

à-č ’k˚ən yə-y˚nə́ the-boy his-house

‘the-boy’s house’

4. Consider the forms below from the Yuman language Hualapai (data from Watahomigie et al. 1982 : 234). Divide the words into morphemes, gloss them, and then discuss whether possession in Hualapai is expressed by head-marking, dependent-marking, or double-marking. If you have trouble making a decision, discuss what further information about Hualapai would

158 INTRODUCING MORPHOLOGY

help you to decide.

nya hú ’va ‘my head’

ma mhú ’ny ‘your head’

nyihá hu'h ‘his/her head’

nya qwá wva ‘my hair’

ma mqwá wny ‘your hair’

nyihá qwá wh ‘his/her hair’

5. Review the sections of Chapter 5 where we discussed morphological pro- cesses like reduplication, infixation, internal stem change, and templatic morphology. Do they present any problems for the traditional fourfold classification? Choose two examples from Chapter 5 and discuss whether they are easily classified or not.

6. Look at the World Atlas of Language Structures Online ( http://wals.info ). At the WALS website click on “Features,” and then click on article 24 “Locus of Marking in Possessive Noun Phrases.” Which is more prevalent in the languages of the world, head-marking or dependent-marking? Now click on the accompanying map. Can you notice any areal tendencies in head-marking and dependent-marking in possessive noun phrases?

7. Look at the Universals Archive website ( http://typo.uni-konstanz.de/ archive/intro/index.php ). Click on “Search” and at the search screen, type “morphology” in the box next to “Original,” and then click on “Submit Query.” You should get about 25 hits. From these hits, find five implica- tional universals.

8. Consider the following examples from the Otomanguean language Mixtec (Macaulay 1996 : 79): a. a-ni-ka-žesámá-rí

TEMP-CP-PL-eat-1 ‘We already ate’

b. a-ni-ka-kã ́ʔã -ró xĩ ́maestro TEMP-CP-PL -talk-2 with maestro ‘You ( PL ) already talked with the teacher’

TEMP stands for temporal, CP for completive, and PL for plural. 1 and 2 stand for first person and second person respectively.

What problem does this example raise for Bybee’s implicational universal?

KEY TERMS valency

argument

active

passive

anti-passive

causative

applicative

noun- incorporation

clitic

phrasal verb

phrasal compound

CHAPTER OUTLINE

In this chapter you will learn about the intersection

between morphology and syntax, which is the study of

sentence structure.

♦ We will examine morphological processes like

passivization and causativization that change the number

of arguments of a verb.

♦ We will look at phenomena that sit on the borderline

between word formation and syntax: clitics, phrasal

verbs like call up in which the two parts can sometimes

be separated in sentences, and compounds that contain

whole phrases or even whole sentences.

♦ We will look at the competition between syntactic and

morphological expression of grammatical concepts.

8 C H A P T E R

Words and sentences: the interface between morphology and syntax

160 INTRODUCING MORPHOLOGY

8.1 Introduction

As you've learned so far in this book, morphology is concerned with the ways in which words are formed in the languages of the world. Syntax, in contrast, is concerned with identifying the rules that allow us to com- bine words into phrases and phrases into sentences. Morphology and syntax, then, are generally concerned with different levels of linguistic organization. Morphologists look at processes of lexeme formation and inflection such as affixation, compounding, reduplication, and the like. Syntacticians are concerned, among other things, with phrase structure and movement rules, and rules concerning the interpretation of anaphors and pronouns. Nevertheless, there are many ways in which morphology and syntax interact, and indeed where the line between syntax and mor- phology is blurred.

We saw in Chapter 6 that inflectional morphology is defined as mor- phology that carries grammatical meaning; as such it is relevant to syn- tactic processes. Case-marking, for example, serves to identify the syntac- tic function of an NP in a sentence. Inflectional markers like tense and aspect affixes identify clauses of certain types, for example, finite or infinitive, conditional or subjunctive. Person and number markers often figure in agreement between adjectives and the nouns they modify, or between verbs and their subjects or objects. In some sense, inflection can be viewed as part of the glue that holds sentences together.

In this chapter we will first look in more detail at several types of ver- bal morphology that affect sentence structure by changing what is called the valency of verbs. Valency concerns the number of arguments in a sentence, where arguments are noun phrases like the subject and object selected by the verb of the sentence. In section 8.3 we will look at the borderline between morphology and syntax. While it is usually clear to linguists which phenomena belong to which level of organization, the boundary between the two levels is not always crystal clear. There are cases in which derivational morphemes appear to attach to whole phras- es, for example, or elements that seem not quite bound enough to be affixes, but not quite free enough to be viewed as independent words. Finally, in section 8.4 we will consider the formation of comparative and superlative adjectives in English as a case where both morphological and syntactic means of expression exist and where a complex set of con- ditions influences which type of expression is favored in any given environment.

8.2 Argument structure and morphology

Above, we defined the valency of a verb as the number of arguments it takes. Arguments , in turn, are defined as those phrases that are semanti- cally necessary for a verb or are implied by the meaning of the verb. Generally, arguments occur obligatorily with a verb, as the examples in (1) show:

Words and sentences 161

(1) a. Fenster snores. *Snores. 1

b. Fenster devoured the pizza. *Fenster devoured. *Devoured the pizza.

c. Fenster put the wombat in the bathtub. *Fenster put the wombat. *Fenster put in the bathtub. *Fenster put.

The verb snore has only one argument, its subject noun phrase. Verbs that have only one argument are traditionally called intransitive . The verb devour requires two arguments, its subject and object noun phrases. Two- argument verbs are transitive . And the verb put requires a subject, an object, and another phrase that expresses location. If a verb requires three arguments, it is traditionally called ditransitive .

The arguments of a verb are often, but not always, obligatory. For example, the verb eat must have a subject, and it can have an object, but the object is not necessary.

(2) a. My goat eats tin cans. b. My goat eats.

Although it is optional we still consider the object tin cans to be an argu- ment of the verb because the verb eat implies something eaten, even if that something is not overtly stated; notice that in (2b) , we assume that the goat eats something (more specifically, we assume that the goat eats something foodlike, if we don’t explicitly say otherwise, as we do in (2a)) .

Of course, in English it is always possible to add prepositional phrases to a sentence that express the time or location of an action, the instru- ment with which it is done, or the manner in which it is done ( Fenster put the wombat in the bathtub with great care on Thursday. . . ). These extra phrases are not necessary to the meaning of the verb, however, and are not argu- ments. Instead, they are called adjuncts . We will not need to talk about adjuncts here.

Valency-changing morphology alters the number of arguments that occur with a verb, either adding or subtracting an argument, making an intransitive verb transitive or a transitive verb intransitive, for example. English has some morphology that changes argument structure, but other languages, as we will see, have far more morphology of this sort.

8.2.1 Passive and anti-passive The most obvious example of valency-changing morphology in English is the passive voice. Example (3) shows a pair of active and passive sentences in English:

(3) a. Fenster bathed the wombat. b. The wombat was bathed (by Fenster).

1. Here we use the convention of marking an unacceptable sentence with an asterisk.

162 INTRODUCING MORPHOLOGY

2. The relative case is the case of the transitive subject in West Greenlandic.

3. T/A means ‘tense/aspect’; DO means ‘direct object’.

4. Bauer ( 1993 : 396–7) actually shows that there are several suffi xes that make verbs passive in Maori.

In the active sentence, the verb has two arguments, its agent (the one who does the action) and the patient or theme (what gets affected or moved by the action); the agent functions as the subject of the action, and the patient as the object. In the passive sentence, an agent is unnecessary. If it occurs, it appears in a prepositional phrase with the preposition by . The patient is the subject of the passive sentence. In effect there is no longer any object, and the passive form of the verb therefore has one fewer argu- ment than the active form.

Part of what signals the passive voice in English is passive morphology on the verb. English passives are formed with the auxiliary verb be and a past participle, which is signaled for regular verbs by adding -ed to the verb base. Irregular verbs can form the past participle in a number of ways: by adding -en ( write ∼ written ), by internal vowel change ( sing ∼ sung ), or by internal vowel change and addition of -t ( keep ∼ kept ).

Other languages also have morphological means to signal the change in argument structure in passive sentences. Example (4) shows an active and a passive sentence from West Greenlandic , and (5) an active–passive pair from Maori :

(4) West Greenlandic (Fortescue 1984 : 265) a. inuit nanuq taku-aat

people. REL 2 polar.bear see-3 PL .3 SG . INDIC . ‘The people saw the polar bear’

b. nanuq (inun-nit) taku- niqar -puq polar.bear (people- ABL ) see- PASSIVE-3SG.INDIC . ‘The polar bear was seen (by the people)’

(5) Maori (Bauer 1993 : 396) a. E koohete ana a Huia i a Pani

T/A scold T/A pers Huia DO pers Pani 3 ‘Huia is scolding Pani’

b. I koohete- tia a Pani e Huia T/A scold-pass pers Pani by Huia ‘Pani was scolded by Huia’

In (4b) , you can see that the morpheme niqar makes a verb passive in West Greenlandic, and (5b) shows that a verb in Maori can be made passive by add- ing the suffix -tia . 4 As was the case in English, the addition of these mor- phemes goes along with passive syntax, that is, making the patient/theme into the subject of the sentence, and making the agent optional. If the agent appears, it is marked with the ablative case in West Greenlandic, and by the preposition e in Maori.

Passive sentences are relatively familiar to speakers of English, but English has nothing like what is called the anti-passive . Like the passive, the anti-passive takes a transitive verb and makes it intransitive by reduc- ing the number of its arguments. What’s different, though, is which

Words and sentences 163

argument gets eliminated. For the passive, it’s the transitive subject that disappears (or is relegated to a prepositional phrase or a case form other than that typical for subject), whereas for the anti-passive, it’s the transi- tive object that disappears, as the example in (6) from Yidi� shows:

(6) Yidi� (Dixon 1977 : 279) yiŋu bu�a buga-:d�iŋ this. ABSOLUTIVE woman. ABSOLUTIVE eat- ANTIPASSIVE ‘This woman is eating’

In Yidi�, the anti-passive is marked on the verb by adding the suffix -:d�iŋ . Since Yidi� is an ergative case-marking language (see section 6.2 ), the subject of a transitive verb is in the ergative case. The subject of an intran- sitive verb is in the absolutive case, as you see in example (6) . So while ‘ eat’ is normally transitive in Yidi�, you can see that it has become intransitive here.

As with the passive, it is also possible to express the ‘missing’ argument overtly, but in a case form other than that usually used for the direct object. Whereas in an active sentence, the direct object of a transitive verb is marked with the absolutive case, in an anti-passive sentence, the subject is absolutive and the object, if it appears, is either in the dative or the locative case:

(7) Dixon ( 1977 : 277) wagu:d�a wawa-:d�i�u gudaga-nda man. ABSOLUTIVE saw- ANTIPASSIVE dog- DATIVE

‘The man saw the dog’

While the translation of (7) makes it look like this sentence means exactly the same thing as an active sentence, this is only because there is no real way of capturing the nuance of this sentence in English, a lan- guage that lacks the anti-passive.

8.2.2 Causative and applicative Passive and anti-passive morphology signal a reduction in the number of arguments that a verb has. There are other sorts of morphology that sig- nal that arguments have been added to a verb.

Causatives signal the addition of a new subject argument, which semantically is the causer of the action. If the verb has only one argument to begin with, the causative sentence has two, and if it has two to begin with, the causative sentence has three arguments. Compare the Swahili sentences in (8) and (9) :

(8) Vitale ( 1981 : 158) a. maji ya-me-chemka

water it- PER -boil ‘The water boiled’

b. Badru a-li-chem-sh-a maji Badru he- PST -boil- CAUSE water ‘Badru boiled the water’ (lit. ‘caused the water to boil’)

164 INTRODUCING MORPHOLOGY

(9) Vitale ( 1981 : 156) a. Halima a-li-ki-pika chakula

Halima she- PST -it-cook food ‘Halima cooked the food’

b. Juma a-li-m-pik-ish-a Halima chakula Juma he- PST -her-cook- CAUSE Halima food

‘Juma caused Halima to cook food’

In (8a) , the verb ‘boil ’ has only one argument, its patient/theme. In (8b) , along with the causative morpheme -(i)sh , an external causer argument is added as the subject of the sentence. Similarly, in (9a) , the verb ‘cook’ has two arguments, an agent ( Halima ) and a patient (‘food’); the agent is the subject of the sentence. When the causative suffix -(i)sh is added, a third argument ( Juma ) is added and it becomes the subject.

Applicative morphology, like causative morphology, signals the addi- tion of an argument to the valency of a verb. But the added argument is an object, rather than a subject. We can again use Swahili for our exam- ple:

(10) Vitale ( 1981 : 44); Baker ( 1988 : 393) a. ni-li-pika chakula

I- PST -cook food ‘I cooked some food’

b. ni-li-m-pik-i-a chakula Juma I-PST-for him-cook- APPL food Juma ‘I cooked some food for Juma’

The suffix - i signals that a second object ( Juma ) has been added to the verb.

Challenge

Aside from the passive, English is usually said to have little in the way of valency-changing morphology. Consider the following data, though:

i. Fenster ate pickles. ii. Fenster over-ate. iii. *Fenster over-ate pickles.

What is the effect of the prefix over- on the valency of the verb eat ? Now consider sentences (iv–vi):

iv. The plane flew. v. *The plane flew the field.

vi. The plane over-flew the field.

Think about these examples, and try to think of other verbs formed with the prefix over- in English. Or better yet, try searching for them in COCA or the BNC. Does over- work like any of the valency-changing morphemes we have looked at in this chapter?

Words and sentences 165

8.2.3 Noun incorporation There is one more way in which morphology interacts with the argument structure of verbs. Consider the data in (11) from the Araucanian language Mapudungun :

(11) Baker and Fasola ( 2009 : 595) a. Ñi chao kintu-le-y ta chi pu waka

my father seek- PROG-IND.3SG.SBJ the COLL cow ‘My father is looking for the cows’

b. Ñi chao kintu-waka-le-y. my father seek- cow- PROG-IND.3SG.SBJ ‘My father is looking for the cows’

Sentences (11a) and (11b) mean precisely the same thing in Mapudungun. In (11a) , the direct object ‘the cows’ is an independent noun phrase in the sentence, but in (11b) , it forms a single compound-like word with the verb root ‘seek’. This sort of structure – where the object or another argument of the verb forms a single complex word with the verb – is called noun incorporation . Noun incorporation tends to occur in languages with poly- synthetic morphology (see section 7.4 ). In Mapudungun, the object noun follows the verb root in both the incorporated and the unincorporated forms. But this need not be the case, as the example in (12) from the Iroquoian language Mohawk shows:

(12) Baker ( 1988 : 20) a. Ka-rakv ne sawatis hrao-nuhs-aʔ

3N- be.white DET John 3 M -house- SUF ‘John’s house is white’

b. Hrao-nuhs- rakv ne sawatis 3 M - house-be.white DET John

‘John’s house is white’

As (12a) shows, the direct object follows the verb when it occurs inde- pendently in Mohawk, but it precedes the verb when it is incorporated.

There is much discussion among morphologists and syntacticians whether noun incorporation should be explained as a result of morpho- logical rules or syntactic rules. We will return to this question in Chapter 10 .

8.3 On the borders

As we saw in the last section, one point of tangency between morphology and syntax occurs where morphology has an effect on the argument structure of verbs. There, it was clear that affixes – clearly morphological elements – can reduce or increase the number of arguments that a verb takes – clearly a matter of syntax. What we will look at in this section, however, are cases where it is not so clear what belongs to morphology and what belongs to syntax – cases, in other words, that inhabit a sort of borderland between the two levels of organization.

166 INTRODUCING MORPHOLOGY

8.3.1 Clitics One of these borderland creatures is something that linguists call a clitic . Clitics are small grammatical elements that cannot occur independently and therefore cannot really be called free morphemes. But they are not exactly like affixes either. In terms of their phonology, they do not bear stress, and they form a single phonological word with a neighboring word, which we will call the host of the clitic. However, they are not as closely bound to their host as inflectional affixes are; frequently they are not very selective about the category of their hosts. Those clitics that come before their hosts are called proclitics , those that come after their hosts enclitics .

Two types of clitics are often distinguished: simple clitics and special clitics . Anderson ( 2005 : 10) defines simple clitics as “unaccented variants of free morphemes, which may be phonologically reduced and subordi- nated to a neighboring word. In terms of their syntax, though, they appear in the same position as one that can be occupied by the corre- sponding free word.” In English, forms like -ll or -d , as in the sentences in (13) , are simple clitics:

(13) a. I’ll take the pastrami, please. b. I’d like the pastrami, please.

In these sentences, -ll and -d are contracted forms of the auxiliaries will and would , and they occur just where the independent words would occur – following the subject I and before the main verb. Like affixes, they are pronounced as part of the preceding word. Unlike affixes, they do not select a specific category of base and change its category or add grammati- cal information to it. Contracted forms like -ll or -d in English will attach to any sort of word that precedes them, regardless of category:

(14) a. The kid over there’ll take a pastrami sandwich. b. No one I know’d want a pastrami sandwich.

In (14a) -ll is cliticized to the adverb there , and in (14b) -d is cliticized to the verb know .

Special clitics are phonologically dependent on a host, as simple clitics are, but they are not reduced forms of independent words. Compare the example in (15) from French :

(15) a. Je vois Pierre. I see Pierre.

b. Je le vois. I him see.

c. *Je vois le. I see him.

Although the object pronoun le in French is written as a separate word, it is phonologically dependent on the verb to its right; in other words, the object pronoun and the verb are pronounced together as a single phono- logical word. There is no independent word that means ‘him’ in French. So le and the other object pronoun forms in French are special clitics.

Words and sentences 167

Clitics are of interest both to syntacticians and to morphologists pre- cisely because they have characteristics both of bound morphemes and of syntactic units. Like bound morphemes, they cannot stand on their own. But unlike morphemes, they are typically unselective of their hosts and have their own independent functions in syntactic phrases.

8.3.2 Phrasal verbs and verbs with separable prefixes Also inhabiting the borderland between morphology and syntax are phrasal verbs in English and verbs with separable prefixes in German and Dutch . Phrasal verbs are verbs like those in (16) that consist of a verb and a preposition or particle:

(16) call up ‘telephone’ chew out ‘scold’ put down ‘insult’ run up ‘accumulate’

Frequently, phrasal verbs have idiomatic meanings, as the glosses in (16) show, and in that sense they are like words. In terms of structure, the combination of verb and particle/preposition might seem like another sort of compound in English. Remember, however, that one of the criteria for distinguishing a compound from a phrase in English (see section 3.4 ) was that the two elements making up compounds could not be separated from one another. We cannot take a compound like dog bed and insert a word to modify bed (e.g., * dog comfortable bed ). In contrast, however, the two parts of the phrasal verb can be, and sometimes must be, separated:

(17) a. I called up a friend. b. I called a friend up. c. I called her up. d. *I called up her.

When the object of the verb is a full noun phrase, the particle can precede or follow it. In the former case it is adjacent to its verb, but in the latter case it is separated from the verb. And when the object is a pronoun, the particle must be separated from the verb. So do we consider phrasal verbs to be a matter of study for morphologists, or do we leave them to syntacti- cians? There is no set answer to this question.

A similar issue arises with what are called separable prefix verbs in Dutch . Consider the examples in (18) :

(18) Booij ( 2002 : 205) a. . . . dat Hans zijn moeder opbelde ∼ Hans belde zijn moeder op

that Hans his mother up.called ∼ Hans called his mother up

b. . . . dat Jan het huis schoonmaakte ∼ Jan maakte het huis schoon that Jan the house clean.made ∼ Jan made the house clean

c. . . . dat Rebecca pianospeelde ∼ Rebecca speelde piano that Rebecca piano.played ∼ Rebecca played piano

Like phrasal verbs in English, separable prefix verbs in Dutch often have idiomatic meanings. For example, opbellen , like English ‘call up’, means ‘to

168 INTRODUCING MORPHOLOGY

telephone’. Each separable prefix verb in Dutch consists of a verb preceded by a word of another category; in (18) , op is a preposition, schoon is an adjective, and piano is a noun. These words therefore look a bit like pre- fixed words or perhaps compounds. But there’s a difference.

To understand the examples in (18) you need to know that Dutch exhibits different word orders in main clauses than in subordinate clauses. In main clauses the main verb is always the second constituent in the clause. If the subject is first, the main verb comes right after it. But in subordinate clauses, the main verb always comes last. The exam- ples in (18) show that when the verbs opbellen ‘call up’, schoonmaken ‘make clean’, and piano-spelen ‘play piano’ occur in a subordinate clause, those complex verbs come at the end of the clause. The first elements op, schoon , and piano occur attached to the verb, almost like prefixes. However, when the verb is used in a main clause, the verb itself occurs after the subject, but its first element appears separated from it at the end of the sentence. So unlike normal prefixes, these elements some- times are not attached to the verbs with which they normally form a unit. Are separable prefix verbs a matter for morphologists or for syntac- ticians? Again, there is no easy answer to this question, as they lie on the border between the two.

8.3.3 Phrasal compounds Our final example of a phenomenon that is neither clearly syntactic nor clearly morphological is called a phrasal compound . A phrasal compound is a word that is made up of a phrase as its first element, and a noun as its second element. Phrasal compounds can be found in many of the Germanic languages, including English , Dutch, and German :

(19) a. English (Harley 2009 ) stuff-blowing-up effects bikini-girls-in-trouble genre comic-book-and-science-fiction fans

b. Dutch (Hoeksema 1988 ) lach of ik schiet humor ‘laugh or I shoot humor’

c. German (Toman 1983 : 47) die Wer war das Frage ‘the who was that question’

On the one hand, phrasal compounds pass one of the acid tests for com- pounding: it is impossible to insert a modifying word in-between the phrase and the head of the compound:

(20) a. *stuff-blowing-up exciting effects b. exciting stuff-blowing-up effects

On the other hand, the first elements are clearly phrases, or even whole sentences, as the example in (21) shows:

Words and sentences 169

(21) God-is-dead theology

And phrases and sentences are the subject matter of syntax. Again, it is no easy question to decide whether phrasal compounds are the subject of morphology or of syntax. Indeed, it would be reasonable to conclude that they should be of interest to both morphologists and syntacticians.

Challenge

Consider the forms below:

wife-and-motherdom do-it-yourself-er old-fartery one-step-behindhood dog-in-the-mangerish get-even-with-themism I-can-do-it-too-ness anti-business-as-usual ex-friends-with-benefits post-fall-of-Baghdad pre-shock-and-awe

All of these are words that were found in COCA (taken from Bauer, Lieber, and Plag 2013 : 513–14). Discuss the internal structure of these words and then consider what sorts of issues they raise for the interaction of morphology and syntax.

8.4 Morphological versus syntactic expression

The final topic we will consider in this chapter is that of morphological versus syntactic expression of particular grammatical distinctions. I’ll illustrate what I mean with a comparison of the future tense in English and French . In English, the future is signaled syntactically, using a combi- nation of the modal auxiliary will plus the uninflected form of the verb, as we see in (22a) . In French, however, the future tense is expressed using an inflectional ending, as in (22b) :

(22) a. English I will speak.

He will arrive.

b. French je parl-er-ai.

I speak- FUT-1.SG Il arriv-er-a. he arrive- FUT-3.SG

One way of describing the difference between English and French in this regard is that the future is expressed periphrastically in English but morphologically in French. We can refer to a way of expressing a

170 INTRODUCING MORPHOLOGY

concept as periphrastic when we use a syntactic construction consist- ing of two or more separate words to express that concept.

Another example can be illustrated with causatives . We saw earlier that in Swahili verbs are inflected with a specific morpheme to mean ‘cause to verb’:

(23) Badru a-li-chem-sh-a maji Badru he- PST -boil- CAUSE water ‘Badru boiled the water’ (lit. ‘caused the water to boil’)

In English we do not have a morphological causative, but we can express causativity periphastically:

(24) I made the water boil.

Note that for this particular verb, we can also use the verb alone with the causative meaning, as in the sentence I boiled the water . But we have no specific morphological means of making a causative verb.

Interestingly, English displays at least one example where a particu- lar grammatical concept can be expressed either morphologically or periphrastically, namely the inf lection of adjectives for the compara- tive and superlative . When comparative and superlative are expressed morphologically, we use the suffixes - er and - est . When we express them periphrastically, we use more for the comparative and most for the superlative.

(25) adjective comparative superlative red redder reddest pure purer purest active more active most active honest more honest most honest

Challenge

Consider the list of adjectives below and try to decide whether you prefer to form the comparative and superlative morphologically or periphrastically:

wide intelligent horrible gentle direct defunct serene stunning ill bumpy official spartan beautiful difficult famous fake meager bold

What factors seem to influence your choice between the morpho- logical and the periphrastic comparative and superlative? Are there any adjectives for which both morphological and periphrastic forms seem possible? Are there any for which neither the morphological nor the periphrastic forms sounds OK?

Words and sentences 171

Morphologists have found that there are a number of factors that influ- ence whether speakers prefer the morphological or the periphrastic compara- tive and superlative. Phonological factors that influence the choice have to do with the number of syllables in the base adjective and its stress pattern. One- syllable adjectives favor the morphological realization and adjectives of three or more syllables prefer the periphrastic form (e.g., red, redder, reddest as opposed to difficult, more difficult, most difficult ). Two syllable adjectives can usually take the morphological comparative and superlative if their second syllable is unstressed, but if the second syllable is stressed, the periphrastic form is preferred (e.g., happy, happier, happiest versus direct, more direct, most direct ). Morphological factors that influence the choice include whether or not the adjective is simple or complex. For complex adjectives, the periphras- tic form is usually preferred (e.g., active, more active, most active ). And syntactic factors include whether the adjective is followed by a preposition and a com- parative clause or not in its syntactic context: consider, for example, whether you prefer more proud or prouder in a context like I am of Fenster than of Letitia . What is most interesting about the comparative and superlative in English is that there are no hard and fast rules. Even if speakers tend to prefer one or the other comparative and superlative forms for particular adjectives, many adjectives can form their comparatives and superlatives either way. So if you check COCA, you might find more red along with redder , for example.

Summary In this chapter we have investigated the relationship between morphology and syntax. We have seen that there are ways in which

morphology affects the syntax of sentences, by either reducing or

increasing the number of arguments a verb may appear with. We

have also looked at cases where it is not entirely clear whether a

phenomenon is a matter of morphology or of syntax or of both. Among

these phenomena, we find clitics, phrasal verbs, separable prefix verbs,

and phrasal derived words and compounds. And we have looked briefly

at an example where morphological and syntactic expression are both

possible. What such phenomena really show us is that morphology and

syntax are often intimately intertwined, and often morphologists must

investigate both levels of organization to really understand how a

language works. We will see in Chapter 10 that phenomena like the

ones we've looked at in this chapter raise serious questions for

theorists, and have been the matter of much discussion.

Exercises 1. Consider the sentences below and discuss how the passive is formed in

Swahili . Use sentences (8) – (10) to help you gloss these sentences. Vitale ( 1981 : 23, 31)

a. Juma a-li-fungua mlango

‘Juma opened the door’

172 INTRODUCING MORPHOLOGY

b. mlango u-li-fungu-liwa ‘The door was opened’

2. Consider the prefix out- in English :

a. Fenster ran. b. *Fenster ran Letitia. c. Fenster outran Letitia. d. *Fenster outran.

Describe the effect that the prefix out- has in sentences (a)–(d). Now, think of other verbs formed with out- in English. Does out- have a consistent effect on the argument structure of verbs it attaches to?

3. Although phrasal compounds may seem somewhat exotic to you, they appear not infrequently in journalistic writing, especially in headlines, and in more informal writing, for example, on the sports pages or in feature writing. Choose your favorite newspaper and try to find two examples of phrasal compounds. Share your examples with classmates and try to analyze what sorts of phrases can occur as the first elements of your compounds.

4. Discuss the difference in argument structure and in verbal morphology between the pairs of sentences below:

a. Malagasy (Keenan and Polinsky 1999 : 604)

i. mijaly Rabe

suffers Rabe

‘Rabe suffers’

ii. mampijaly an-dRabe Rasoa

makes-suffer acc-Rabe Rasoa

‘Rasoa makes Rabe suffer’

b. Chichewa (Mchombo 1999 : 506)

i. Kalúlú akuphíká maûngu

Hare is cooking pumpkins

‘The hare is cooking pumpkins’

ii. Kalúlú akuphíkíra mkángó maûngo

Hare is cooking lion pumpkins

‘The hare is cooking pumpkins for the lion’

5. Consider the English sentences below: a. The water boiled. b. Fenster boiled the water. a. The tomatoes grew. b. Letitia grew the tomatoes. a. The door opened. b. Roddy opened the door.

First discuss the difference in argument structure between the (a) sentences and the (b) sentences. Then compare these sentences to the Swahili sentences in (8) (see p. 165). Discuss the differences between the Swahili sentences and the ones here.

6. Musqueam exhibits an interesting set of morphologically related verbs that differ in valency. Analyze the following forms and describe both the affixes used and their effects on argument structure (Suttles 2004 : 235) (note that the first two examples are a bit different from the second two):

Words and sentences 173

háy ‘finish’ shá·ỷ ‘finished’ shá·ỷ stəx� ‘have it finished’ łə́lqt ‘dip it’ słélq ‘in the

water’ słélqstəx� ‘keep it in the

water’

x̆é·’t θ t ‘measure it’

sx̆ eʔé � t θ ‘measured,

marked’ sx̆eʔé

� t θ stəx� ‘blaze (as a trail),

designate (as a time)’

θə́yt ‘fix it’ sθəθəỷ ‘right’ sθəθə́ỷstəx� ‘keep it on course’

7. For each of adjectives below, check in COCA whether you find the morphological comparative and superlative, the periphrastic comparative and superlative, or both. For example, for the adjective pure , you'd look for purer , more pure , purest , and most pure . See what you find and discuss the factors that seem to influence the choice of the morphological and periphrastic forms.

weird winning (as in a winning team) happy famous

KEY TERMS allomorph

assimilation

epenthesis

underlying representation

vowel harmony

onset

nucleus

coda

rhyme

palatalization

lexical strata

CHAPTER OUTLINE

In this chapter we will learn about the intersection between

morphology and phonology, which is the study of the

sound structure of languages.

♦ We will learn that some morphemes exhibit allomorphs,

that is, phonologically distinct variants.

♦ We will learn how to analyze both phonologically

predictable and unpredictable allomorphy.

♦ We will learn about how syllable structure and

morphology can interact.

♦ And we will consider the nature of lexical stratification,

where different types of phonological behavior

characterize different parts of the morphology of a

language.

Sounds and shapes: the interface between morphology and phonology

9 C H A P T E R

176 INTRODUCING MORPHOLOGY

9.1 Introduction

Phonology is the area of linguistics that is concerned with sound regu-

larities in languages: what sounds exist in a language, how those sounds

combine with each other into syllables and words, and how the prosody

(stress, accent, tone, and so on) of a language works. Phonology interacts

with morphology in a number of ways: morphemes may have two or more

different phonological forms whose appearance may be completely or at

least partly predictable. Some phonological rules apply when two or more

morphemes are joined together. In some languages morphemes display

different phonological behavior depending on whether they are native to

the language or borrowed into it from some other language. Some mor-

phological rules may depend on the syllable structure of a language. In

this chapter we will explore the various ways in which phonology inter-

acts with morphology.

In this chapter we will frequently make use of phonetic transcriptions,

so you may want to review the IPA before you begin reading it. We will

also make use of terminology which classifies sounds by their point of

articulation (labial, dental, alveolar, and so on) and by their manner of

articulation (voiced vs. voiceless, stop, fricative, liquid, and so on). You

can find summaries of this terminology in the charts at the beginning

of the book.

9.2 Allomorphs and morphophonological rules

Allomorphs are phonologically distinct variants of the same morpheme.

By phonologically distinct, we mean that they have similar but not iden-

tical sounds. And when we say that they are variants of the same mor-

pheme, we mean that these slightly different-sounding sets of forms

share the same meaning or function. For example, the negative prefix

in- in English is often pronounced in- (as in intolerable ), but it is also some-

times pronounced im- or il- ( impossible, illegal ), as English spelling shows.

Since all of these forms still mean ‘negative’, and they all attach to adjec-

tives in the same way, we say that they are allomorphs of the negative

prefix. Another example you’ve already seen is the regular past tense in

English. Although the regular past tense in English is always spelled -ed ,

it is sometimes pronounced [t] ( packed ), sometimes [d] ( bagged ), some-

times [əd] ( waited ). 1 Still all three phonological variants designate the

past tense. Similarly, the plural morpheme in Turkish sometimes appears

as -lar and sometimes as - ler , so Turkish has two allomorphs of the plural

morpheme.

As we will see below, in many cases, it is phonologically predictable

which allomorph appears where; sometimes, however, which allomorph

appears with a particular base is unpredictable. For example, we will see

that it is usually possible to predict the form of the regular allomorphs of

1. Or [id] in some dialects.

Sounds and shapes 177

the English past tense morpheme, but there are quite a few verbs whose

past tenses are irregular (e.g., sang, f lew, bought ).

9.2.1 Predictable allomorphy Let’s look more closely at the prefix in- in English . As the examples in (1a)

show, it frequently has the form in- . However, sometimes it appears as im-, il- ,

or ir- , as the examples in (b) and (c) show. And if you think about sound rather

than spelling, it can also be pronounced [iŋ-], as the examples in (1d) show:

(1) a. inalienable

intolerable

indecent

b. impossible

c. illegal

irregular

d. incongruous [iŋkaŋgɹuəs]

incoherent [iŋkohiɹənt]

The various allomorphs of the negative prefix in- in English are quite regu-

lar, in the sense that we can predict exactly where each variant will occur.

Which allomorph occurs depends on the initial sound of the base word.

For vowel-initial words, like alienable , the [in-] variant appears. It appears

as well on words that begin with the alveolar consonants [t, d, s, z, n]. On

words that begin with a labial consonant like [p], we find [im-]. Words that

begin with [l] or [ɹ] are prefixed with the [il-] or [iɹ-] allomorphs respec-

tively, and words that begin with a velar consonant [k], are prefixed with

the [iŋ-] variant. What you should notice is that this makes perfect sense

phonetically: the nasal consonant of the prefix matches at least the point

of articulation of the consonant beginning its base, and if that consonant

is a liquid [l,ɹ] it matches that consonant exactly. This allomorphy is the

result of a process called assimilation . Generally speaking, assimilation

occurs when sounds come to be more like each other in terms of some

aspect of their pronunciation.

If you have studied a bit of phonology, you know that regularities in the

phonology of a language can be stated in terms of phonological rules.

Phonologists assume that native speakers of a language have a single basic

mental representation for each morpheme that we call the underlying

representation . Regular allomorphs are derived from the underlying rep-

resentation using phonological rules. For example, since the English nega-

tive prefix in- is pronounded [in] both before alveolar-initial bases ( tolerable,

decent ) and before vowel-initial bases ( alienable ), whereas the other allo-

morphs are only pronounced before specific consonant-initial bases, pho-

nologists assume that our mental representation of in- is [in] rather than

[iɹ], [il], or [iŋ]; often (but not always, as we will see below) the underlying

form of a morpheme is the form that has the widest surface distribution. 2

When the underlying form is prefixed to a base beginning with anything

2. If you study phonology further, you will fi nd that this is somewhat of a simplifi cation, but for our purposes,

it is good enough.

178 INTRODUCING MORPHOLOGY

other than a vowel or alveolar consonant, the following phonological rule

derives the correct allomorph:

(2) Nasal assimilation : a nasal consonant assimilates to the point of

articulation of a following consonant, and to the point and manner

of articulation of the consonant if it is a liquid.

Phonologists use different forms of notation to express the above rule in

a more succinct fashion, but we’ll restrict ourselves to informal state-

ments of rules here.

This sort of assimilation – called nasal assimilation – is not unusual in

the languages of the world. We find something similar in the language

Zoque (Nida 1946 / 1976 : 21):

(3) pama ‘clothes’ ʔəs mpama ‘my clothes’

kayu ‘horse’ ʔəs ŋkayu ‘my horse’

tuwi ‘dog’ ʔəs ntuwi ‘my dog’

As the examples in (3) show, the possessive prefix is a nasal consonant

that has three different allomorphs. Which allomorph is prefixed depends

on the point of articulation of the noun it attaches to. In Zoque, we might

say that part of forming the possessive of a noun involves prefixing an

underlying nasal consonant which undergoes a phonological rule that

assimilates it to a following consonant.

Another example of a predictable form of allomorphy is the formation

of the regular past tense in English. In Chapter 2 , we looked at the past

tense in English in the context of figuring out what the mental lexicon

looks like. We can now go into its formation in somewhat more detail.

Consider the data in (4) , which shows two of the three allomorphs of the

regular past tense:

(4) a. Verbs whose past tense is pronounced [t]

slap, laugh, unearth, kiss, wish, watch, walk

b. Verbs whose past tense is pronounced [d]

rub, weave, bathe, buzz, judge, snag, frame, can, bang, lasso,

shimmy

The regular past tense in English illustrates a different sort of assimila-

tion, called voicing assimilation , where sounds become voiced or voiceless

depending on the voicing of neighboring sounds. The verbs that take the

past tense allomorph [t] all end in voiceless consonants: [p, f, θ, s, ʃ, tʃ, k].

Those that take the [d] allomorph, all end either in a voiced consonant [b,

v, ð, z, dȢ, g, m, n, ŋ, etc.] or in a vowel (and all vowels are voiced, of course).

Why just this distribution? Clearly, the past tense morpheme has come to

match the voicing of the final segment of the verb base: verbs whose last

segment is voiceless take the voiceless variant.

There is one allomorph of the past tense we haven’t covered yet.

Consider what happens if the verb base ends in either [t] or [d]:

(5) Verbs whose past tense is pronounced [əd]

defeat, bond

Sounds and shapes 179

Here, a process of epenthesis is at work. Epenthesis is a process that

inserts a segment into a word. Here, a schwa is inserted to separate [d]

of the past tense from the alveolar stop at the end of the verb. Again,

this makes perfect sense phonetically; if the [d] allomorph were used,

it would be hard to distinguish from the final consonant of the verb

root.

What is the underlying form of the past tense morpheme in English?

As I indicated before, it is often a good strategy to assume that the allo-

morph with the widest distribution is the underlying form. But there is

something else to consider as well. Phonologists typically assume that

the underlying form of a morpheme must be something from which all

of the other allomorphs can be derived using the simplest possible set

of rules. In this case, the allomorph [d] has the widest distribution,

because it occurs with all voiced consonants except [d], and with all

vowel-final verb stems. And if we assume that the underlying form of

the regular past tense is [d], we need only two simple rules to derive the

other allomorphs:

(6) The Past Tense Rule

a. If the verb stem ends in [t] or [d] (the alveolar stops), insert [ə]

before the past tense morpheme (e.g., defeated [dəfit + d] →

[dəfit + əd]).

b. Assimilate [d] to the voicing of an immediately preceding

consonant (e.g., licked [lik + d] → [lik + t]).

Challenge

Rather than take my word for it that choosing the allomorph [d] as

the underlying representation of the past morpheme yields the

simplest set of rules, construct an argument that the set of rules in

(6) really is simpler than alternatives. To do so, first suppose that we

had chosen [t] instead of [d] as the underlying morpheme. Try to state

informally what the rule(s) would have to be to derive the other

allomorphs of the past tense. Then suppose that we’d chosen [əd] as

the underlying representation. What would the rules have looked like

then? Now compare the rules to each other and discuss which set is

simplest.

A third example of regular and predictable allomorphy comes from

Turkish . As we’ve seen, in Turkish, virtually every morpheme, derivational

and inflectional alike, has more than one allomorph. For example, the

plural morpheme has the allomorphs -ler and -lar , and the genitive suffix

has the allomorphs - in, -un, -ιn , and - ün . The reason for this is that Turkish

displays a process of vowel harmony whereby all non-high vowels in a

word have to agree in backness, and all high vowels in both backness and

roundness. When suffixes are added to a base, they must agree in the

relevant vowel characteristics with the preceding vowels of the base:

180 INTRODUCING MORPHOLOGY

(7) From Lewis ( 1967 : 29ff)

‘hand’ ‘measure’ ‘evening’ ‘fear’

Abs. pl. el-ler ölçü-ler akşam-lar korku-lar Gen. sg. el-in ölçü-n-ün akşam-ιn korku-n-un

Since the roots of the nouns el ‘hand’ and ölçü ‘measure’ have front vowels,

the plural suffix must agree with them in frontness, so the - ler allomorph

appears. On the other hand, ak ş am ‘evening’ and korku ‘fear’ have back vowels, and the -lar allomorph appears. Since the genitive ending has a

high vowel, the vowel harmony is more complicated. If the noun root

consists of vowels that are front and non-round, we find the genitive allo-

morph with a front, non-round vowel, that is, -in . Similarly, if the root

contains front, round vowels, so does the suffix; so ölçü gets the front

round allomorph - ün . Roots with back non-round vowels like akşam take the - n allomorph, and roots with back round vowels like korku take the - un

allomorph.

We might ask in this case what the underlying form of these affixes is,

and here it’s a bit difficult to pick one of the existing allomorphs as our

choice. For example, neither plural allomorph - ler nor - lar has a wider

distribution than the other. One possibility that we might consider, then,

is that the mental representation of the plural morpheme in Turkish is

something like what we find in (8) :

(8) Turkish plural morpheme =

non-high

non-round

-1 r

Part of the rule of vowel harmony in Turkish might then say that a non-

high vowel in a suffix comes to match the backness of the vowels in a root

that precedes it.

Challenge

We have proposed an underlying form for the plural morpheme in

Turkish. Now propose one for the genitive suffix and try to give an

informal statement of the rule of vowel harmony that gives rise to the

different allomorphs.

We can give one more example of predictable allomorphy from Turkish,

which affects consonants rather than vowels. Let’s look at a bit more data

from Turkish (Myers and Crowhurst 2006 ( www.laits.utexas.edu/phonolo-

gy/turkish/index.html )):

(9) Nom.Sg Dat.Sg Nom.Pl

‘stalk’ sap sapa saplar

‘container’ kap kaba kaplar

The examples in (9) are interesting because the forms look straightfor-

ward until we get to the Dat.Sg. of the word ‘container’. If we look only at

Sounds and shapes 181

the nominative forms of these words we might think that the underlying

forms of the roots are sap and kap respectively, that is, that the underlying

roots are identical to the nominative singular forms. However, when we

look at the dative form, we find that ‘container’ has a voiced consonant

[b] preceding the dative suffix, where ‘stalk’ maintains its voiceless [p].

How can we explain why the two words behave differently in the dative

form? If we assume that the root of ‘container’ underlyingly ends in [b]

rather than [p] (so the underlying roots are sap and kab respectively), we

can say that voiced consonants become voiceless either word finally or

before consonants, but remain voiced when they occur between vowels.

Note that in this case, the underlying form of the word ‘container’ is not

equivalent to any of the surface forms that we see in the data in (9) .

So far, we have looked at relatively simple examples of allomorphy, but

more complex allomorphy can and does occur in the languages of the

world. Consider the data below from the Austronesian language Balantak ,

spoken in Indonesia (data from Busenitz and van den Berg 2012 : 14–15).

The forms illustrate the forms of the actor voice irrealis prefix:

(10) a. mang-ala ‘to take, get’

meng-keke ‘to dig’

ming-ili ‘to buy’

mong-gopot ‘to sort, arrange’

mung-kukudi ‘to peel’

b. mam-bala ‘to fence in’

mem-pepel ‘to hammer’

mim-pipir ‘to splatter’

mom-popok ‘to cut’

mum-buani ‘to fish with a net’

c. man-taring ‘to cook’

men-deer ‘to spread’

min-sikoop ‘to shovel’

mon-sosop ‘to suck’

mun-tunu ‘to burn’

d. manga-wawau ‘to make’

menge-leelo’ ‘to call, invite’

mingi-nika’ ‘to marry’

mongo-yoong ‘to shake’

mungu-rudus ‘to pick’

The examples in (10) show that there appear to be many allomorphs for

the irrealis actor focus morpheme in Balantak, or to put it differently, that

several phonological processes can affect that morpheme. First, it is appar-

ent that the vowel in the prefix must be the same as the first vowel in the

verb base. This is, of course, another case of vowel harmony. The examples in

(10b) and (10c) show us that the final consonant of the prefix assimilates to

the initial consonant of the base if it is an obstruent (i.e., a stop or a fricative).

Finally, the examples in (10d) show us that an epenthetic vowel is inserted at

the end of the prefix if the base begins with a sonorant and that this vowel

182 INTRODUCING MORPHOLOGY

harmonizes with the first vowel in the base. What, then, would the underly-

ing form of all these allomorphs be? The data in (10a) suggest that the under-

lying form should be consonant final rather than vowel final, as the conso-

nant final allomorphs have the widest distribution. Further, the data suggest

that the final consonant of the prefix should be ng [ŋ] because this conso-

nant occurs both before velar-initial bases and vowel-initial bases (again, that

means that it has the widest distribution). The consonant [m] occurs only

before labials, and [n] only before alveolars. As for the vowel, there seems to

be no argument for preferring one surface vowel over the other, so van den

Berg and Busenitz simply show the underlying form with an unspecified

vowel, symbolized by V. The underlying form, then, is mVng -.

9.2.2 Unpredictable or partially predictable allomorphy As we’ve seen above, some allomorphy is regular enough to be captured by

phonological rules. But not all allomorphy is regular. Take, for example,

past tenses of verbs in English . We have already looked at the regular past

tense. Every native speaker or student of English knows that there are also

quite a few verbs that don’t form the past tense by adding - ed . Consider

Table 9.1 , which gives a selection of examples.

Table 9.1. (based on classes in Huddleston and Pullum 2002 )

Infinitive Irregular past Pattern

1 burn burnt devoicing of suffix

2 keep kept vowel shortening

3 hit hit no change

4 feel felt vowel shortening with devoicing of suffix

5 bleed bled vowel shortening and no suffix

6 leave left devoicing of stem consonant

7 sing sang vowel ablaut (i ~ æ)

8 win won vowel ablaut (i ~ )

9 fight fought vowel ablaut (ai ~ ɔ)

10 come came vowel ablaut ( ~ e)

If you think back to Chapter 2 , when we discussed the mental lexicon,

we suggested that irregular past tense allomorphs are simply stored in the

mental lexicon, and not derived by rules. So speakers of English have a

lexical entry for the verb root sing , and along with it an associated entry

for past tense sang . It is possible, though, that things are a bit more com-

plicated. Think back to the experiment in section 6.3 where you asked a

number of friends to make the past tense of the hypothetical verb gling .

Probably a significant number of them offered either glang or glung . Since

this is not a real verb, clearly they didn’t have a past tense stored for it.

Sounds and shapes 183

Rather, they must have been making use of some sort of pattern to create

these forms. In English there happen to be quite a few verbs whose present

and past tenses show the same i ∼ æ alternation as sing or the i ∼ alterna-

tion of win . There appears to be an abstract pattern that speakers are tap-

ping into here that relates a present tense with [i] to a past tense with [æ]

or [ ] if the verb ends in a nasal or a nasal plus some other consonant (e.g.,

like swim, ring, sting, win, stink ). Psycholinguists continue to work towards

figuring out the exact nature of such patterns.

The example of unpredictable allomorphy we looked at above concerns

English inflection. Let’s look at another example that has to do with deri-

vation. Consider the forms in (11) :

(11) a. designate [�dε.zig.neit] designation [dε.zig.�nei.ʃ + n]

b. unionize [�ju.njə.naiz] unionization [ju.njə.nai.�z + ei.ʃ n]

c. prosecute [�pɹa.sə.kjut] prosecution [pɹa.sə.�kju.ʃ + n]

d. resolve [ɹə.�zalv] resolution [ɹε.zə.�l + u. ʃ n]

e. expedite [�εk.spə.dait] expedition [εk.spə.�di. + ʃ n]

f. define [də.�fɑin] definition [dε.fə.�n + i.ʃ n]

g. absorb [əb.�zɔrb] absorption [æb.�zɔrp. + ʃ n]

h. circumcise [�səɹ.k m.saiz] circumcision [səɹ.kəm.�si.Ȣ + n] i. decide [də.�said] decision [də.�si.Ȣ + n]

All of the verbs in the lefthand column have noun forms with the suffix

- tion . But if you compare the transcriptions of the verbs and nouns care-

fully, you will see that both the verb bases and the derivational affix have

various allomorphs. For example, the suffix seems to be - n in (11a and c)

but - e iʃ n in (11b) . It looks like -u ʃ n in (11d) , but -iʃ n in (11f) , and -ʃ n in

(11g) . In ( 11a, c, and e ) the [t] at the end of designate, prosecute , and expedite

seem to have changed to [ʃ], the [v] at the end of resolve seems to have disap-

peared, and the [b] at the end of absorb has changed to [p]. And if you look

carefully at many of these forms, the stress pattern on the derived noun

is different from that of its verb base. In other words, there is quite a com-

plicated pattern of allomorphy associated with this suffix.

Is it predictable? Parts of it are. For example, if a verb ends in [v] and has

a derived noun with the - tion suffix, it will always lose its [v] and the suffix

will be pronounced - ution (think about the derived nouns for dissolve,

absolve, revolve , etc.). Similarly, if a verb ends in [t] and takes the - ion suffix,

the [t] will become [ʃ]. And if a verb ends in [z] or [d] and takes the - tion

suffix, those consonants will become [Ȣ]. Since the sounds [ʃ] and [Ȣ] are

palatal sounds, this process is called palatalization . You can also observe

that regardless of where primary stress falls in the base, the derived form

has stress on the penultimate (next to the last) syllable.

But the choice of allomorphs is not entirely predictable. For example, it’s

not clear if we can predict when we will get -ation , say, as opposed to -ion on

a particular verb base: we find -ation on the verbs accuse and refute, but not

in transfuse and prosecute; those have the -ion allomorph. The derived noun

form from combust is combustion, but that of infest is infestation. Why not

combustation and infestion instead? The verb base propose yields proposition,

but accuse yields accusation. Why not proposation , or accusition ? To some

184 INTRODUCING MORPHOLOGY

extent the choice of allomorphs seems to be quite arbitrary. We will leave

this affix here, but return to it in section 9.5 , where we will consider why

this affix (and a number of others) display such pervasive and unpredict-

able allomorphy .

9.3 Other morphology–phonology interactions

Not all morphology–phonology interactions have to do with allomorphy.

Here we will look at two cases of reduplication in which phonological

structure is important to morphological analysis. In Chapter 5 we learned

about full reduplication and partial reduplication. Here, we will look

more closely at partial reduplication, as in some languages the descrip-

tion of just what in a word gets reduplicated (which we’ll refer to as the

‘copy’) depends on understanding the syllable structure of words.

Before we look at reduplication, though, we need to look briefly at how

the sound segments of words are organized into larger units of structure.

Briefly, phonologists argue that phonemes are grouped into syllables and

syllables into larger units of structure. Syllables are composed of three

parts, called the onset , the nucleus , and the coda , as you see illustrated

in (12) . The only obligatory part of a syllable is the nucleus, which typi-

cally consists of a vowel. Any consonants that go before the nucleus are

called the onset, and any that go after the onset are called the coda. The

nucleus and the coda form a grouping that we call the rhyme . The symbol

that we use for ‘syllable’ is the Greek letter sigma.

(12)

onset rhyme

nucleus coda

How do we know how to divide words into syllables? This is a difficult

question, but here we will use a rule-of-thumb called “maximize the

onset.” In order to find the boundaries between syllables, we can follow

the simple procedure illustrated in (13) . Consider, for example, the words

opera and compensate in English: 3

(13) a. Mark the nucleus of each syllable

b. Make any consonants before the first vowel an onset. Then look

at the consonants between the first and second nucleii. If there’s

only one, make it the onset of the following syllable. If there’s

more than one, put as many in the onset of the second syllable

3. In the examples in (13) I use English orthography rather than phonetic transcriptions, but for the beginner

it sometimes helps to transcribe words.

Sounds and shapes 185

as could begin a word. Repeat the same process between the

second and third, and so on. For example, the p and r in opera

can go with the vowels that follow them, but in compensate, only

the p and s can go with the vowels that follow them, as words in

English cannot begin with the sequences mp or ns.

c. Any consonants that have not been put in the onset of a syllable

go into the coda of the preceding syllable.

nu o nu o nu o nu co o nu co o nu co | | | | | | | | | | || | | o p e r a c o m p e n s a t e

d. Join the nucleus and the coda into a rhyme. If there’s no coda,

just write rhyme above the nucleus.

r r r r r r

nu o nu o nu o nu co o nu co o nu co

o p e r a c o m p e n s a t e

e. Join the onset and the following rhyme.

σ σ σ σ σ σ

r r r r r r

nu o nu o nu o nu co o nu co o nu co

o p e r a c o m p e n s a t e

This procedure can be used for English, but it can also be used for other

languages as well as long as we have some idea of what consonants can

begin words in that language.

Once we have an idea of how we can divide words into syllables, we can

distinguish different types of syllables. For our purposes, it will be enough

to know the difference between heavy syllables and light syllables . A

heavy syllable is one that either has a long vowel or ends in a consonant

(in which case it has a coda). A light syllable has a short vowel and no coda.

What does this have to do with reduplication? Interestingly, there are

some languages in which partial reduplication rules require the copy to

be a particular type of syllable. Take, for example, the Austronesian lan-

guage Mokilese , where the progressive form of the verb is formed by redu-

plicating a heavy syllable (data from Inkelas 2014 : 170):

(14) Mokilese verb progressive form

‘plant’ pɔdok pɔd-pɔdok

‘tear’ soorɔk soo-soorɔk

‘find’ diar dii-diar

In the first example, the first syllable is just pɔ , but this is a light sylla-

ble, so the reduplication rule takes the consonant of the second syllable

186 INTRODUCING MORPHOLOGY

onset and adds it on to pɔ to make the heavy syllable pɔd . In the second

example, the first syllable of the verb is already heavy (it has a long vowel),

so there’s no need to add anything from the second syllable to make up

the copy. The third example requires another kind of adjustment to the

copy for it to be heavy. Since the following syllable begins with a vowel, we

can’t use an onset consonant to fill out the coda of the reduplicating syl-

lable. So the reduplication rule has to lengthen the first vowel to make a

heavy syllable for the copy.

In the Pama-Nyungan language Diyari , the partial reduplication rule,

which is used for a number of different purposes, copies what’s called a

‘minimal word’, which for our purposes can be defined roughly as a unit

consisting of two syllables (data from Inkelas 2014 : 170):

(15) Diyari w�ila w�ila-w�ila ‘woman’

kanku kanku-kanku ‘boy’

ku�kuŋa ku�ku-ku�kuŋa ‘to jump’

t�ilparku t�ilpa-t�ilparku ‘bird species’

If we were to look only at the first two examples in (15) , we might think

that Diyari has a rule of full reduplication, but the third and fourth

examples show that this is not the case; if the base word consists only of

two syllables, that is what gets reduplicated, but if the base is longer, the

first two syllables (the minimal word) gets reduplicated.

The reduplication rules of Mokilese and Diyari show us that knowing

about the syllable structure of words can help us to understand how mor-

phological rules work. This is true not only in the case of reduplication, but

also in examples of morphology that can be found in English itself. What I

have in mind here is the formation of hypocoristics. We briefly mentioned

hypocoristics, the formation of nicknames, in Chapter 5 , noting that it was

a subtractive process in English. But what we did not look at was the pho-

nological basis of the process, which also has to do with syllable structure.

Just to give you a taste of how it works, consider the challenge below.

Challenge

Consider the names and corresponding nicknames below. Can you

figure out what sort of syllable the subtractive process of hypocoristic

formation needs to leave behind to have a well-formed nickname in

English?

Patricia Pat or Trish

Michael Mike

Jennifer Jen

David Dave

Elizabeth Liz

Douglas Doug

You might want to transcribe these names into IPA and divide them

into syllables using the procedure in (13) to help with your analysis.

Sounds and shapes 187

4. The data in (16) represent two different types of verbs in Tagalog that are formed with the same prefi x. The

‘actor focus’ verbs are roughly like active (as opposed to passive) verb forms in English, and what Schachter

and Otanes call “derived” verb forms are ones that denote destructive activity or activity directed at several

objects or people. For the purposes of this problem, it doesn’t matter that the prefi x has several different uses.

5. Although the spelling suggests that this form begins with a vowel, it is pronounced with a glottal stop before

the vowel, so phonetically it is actually [ʔanak].

9.4 How to: morphophonological analysis

So far we’ve looked at morphophonological processes in a number of dif-

ferent languages. In this section, we’ll take a close look at how morpholo-

gists go about analyzing allomorphy in a language that’s unfamiliar. Take

a look at the data in (16) from the Austronesian language Tagalog

(Schachter and Otanes 1972 : 290–1):

(16) Verb root Actor focus or derived verb form 4

anak manganak ‘give birth (to)’

bakya mambakya ‘hit with a wooden shoe’

dukot mandukot ‘steal’

gulo manggulo ‘create disorder’

hiwa manghiwa ‘cut (sthg. intentionally)’

kailangan mangailangan ‘need’

ligaw manligaw ‘pay court to’

manhid mamanhid ‘get numb’

nood manood ‘watch’

pili mamili ‘choose (several things)’

sakit manakit ‘cause pain’

takot manakot ‘frighten (several people)’

walis mangwalis ‘hit with a broom’

The first thing that should leap out at you when you see these data is

that the forms in the right-hand column (let’s refer to them as the derived

forms) seem to have some sort of prefix. But it’s not always exactly the

same prefix. The prefix looks like mang- in the first form, mam- in the

second, and man- in the third. There’s clearly some allomorphy displayed

in this set of data. To see better what’s going on, it is often a good strategy

to rearrange the data so that similar forms are put together. This allows

you to begin to see patterns. There are a number of ways of doing this, but

in (17) , I’ve rearranged them into four groups. In the first, we can clearly

segment off the prefix mang- . In the second group, it’s still possible to seg-

ment off a prefix and leave behind something that looks exactly like the

verb root. What’s left over is either mam- or man- . The examples in (17c)

look like something is missing, though. If we were just putting together a

prefix with the stem, we might expect mammanhid and mannood , rather

than the forms we actually find. And finally in (17d) , we have forms in

which the initial consonant of the verb root clearly seems to be absent.

(17) a. Verb root Actor focus verb

anak 5 manganak ‘give birth (to)’

188 INTRODUCING MORPHOLOGY

gulo manggulo ‘create disorder’

hiwa manghiwa ‘cut (sthg. intentionally)’

walis mangwalis ‘hit with a broom’

b. bakya mambakya ‘hit with a wooden shoe’

dukot mandukot ‘steal’

ligaw manligaw ‘pay court to’

c. manhid mamanhid ‘get numb’

nood manood ‘watch’

d. pili mamili ‘choose (several things)’

takot manakot ‘frighten (several people)’

sakit manakit ‘cause pain’

kailangan mangailangan ‘need’

Let’s start with the forms in (17a) and (17b) that are easily segmentable.

Segmenting the derived verbs, we find the allomorphs mam -, man- , and

mang- . The first occurs on a root beginning with [b], the second with roots

beginning with [d] and [l], and the third with roots beginning with [ʔ], [g],

[h], or [w]. This far, the data should not surprise you: Tagalog seems to have

a process of nasal assimilation, just as English and Zoque do. The labial-

final allomorph mam- occurs with a labial consonant, the alveolar-final

allomorph man- occurs with alveolar initial roots, and the velar-final allo-

morph mang- occurs with roots that begin with either velar or glottal

consonants. So far, things seem fairly neat.

When we turn our attention to the data set in (17c) , however, we will

see that there’s more to be said about the derived forms of the verb. It’s

not so clear how to segment the forms in this set. Suppose we assume

that the prefix is mam- or man- , as it was in (17a, b) ; we are then left with

anhid or ood as allomorphs of the roots. Alternatively, we might assume

that the prefix in these cases is just ma- . If we do so, then the bases

would be exactly the same as the roots, namely manhid and nood . This

might seem like the best solution at the moment – just adding another

allomorph to the set we already have of mang-, man-, and mam-, but let’s

keep our minds open to both solutions until we’ve finished looking at

all the data.

The data in (17d) present us with a new problem. If we assume that

Tagalog has nasal assimilation, we would expect that we would put

together mam- with pili to get mampili and man- with takot to form manta-

kot , and so on. But instead we get mamili and manakot . The initial conso-

nant of the root seems to disappear when the prefix is attached. Now, if

we look back and compare the verb roots that we find in (17b) and com-

pare them to those we find in (17d) , we will see that the former begin with

voiced labial or alveolar consonants, whereas those in (17d) begin with

voiceless consonants. It looks like when the prefix attaches to a base that

begins with a voiceless consonant, the prefix first assimilates to the point

of articulation of the following voiceless consonant, and then that conso-

nant disappears.

Sounds and shapes 189

Here, we need to stop and think more about the nasal assimilation rule.

In section 9.2 I suggested that each set of allomorphs has a single underly-

ing form, from which the others are derived by phonological rule. Since

nasal assimilation in Tagalog seems to be a predictable process, we would

assume this to be the case here as well. So we need to decide at this point

what the underlying form should be. Remember that it’s often a good

strategy to pick as the underlying form the allomorph that has the widest

distribution, in other words the one that occurs with the most classes of

sounds. Here, as the data in (17) show, the allomorph mang- occurs with

glottal initial roots ([h] and [ʔ]), as well as with velars ([g], [k], [w]). The allo-

morph mam- occurs only with labial-initial roots ([p], [b]), and the man- allo-

morph only with alveolar-initial roots ([d], [t], [l]). We can reasonably make

the hypothesis then that the underlying form of the prefix is mang- , since

it occurs with two different classes of sounds. If so, then the forms in (17)

have the following underlying representations:

(18) a. mang + anak

mang + gulo

mang + hiwa

mang + walis

b. mang + bakya

mang + dukot

mang + ligaw

c. mang + manhid

mang + nood

d. mang + pili

mang + takot

mang + sakit

mang + kailangan

Now we can give an informal statement of two phonological rules that

will derive the allomorphs from the underlying forms:

(19) a. The nasal of mang- assimilates to the point of articulation of a

following consonant.

b. A voiceless consonant is deleted when preceded by a nasal

consonant.

For the forms in (17a) , neither rule applies. For those in (17b) , only the first

applies, since the verb roots don’t begin with voiceless consonants. But in

(17d) both rules apply.

(20) No rules: mang + gulo → mang + gulo

Rule (19a) only: mang + bakya → mam + bakya

Both rules: mang + pili → mam + pili → mam + ili

What about the examples in (17c) , however? Remember that we were

undecided as to whether the allomorph of the prefix should end in a nasal

at all, and we were leaning towards the solution in which the allomorph

was ma- , as it would allow us to say that the verb roots always had the

190 INTRODUCING MORPHOLOGY

same form. We can see now, however, that that might not be the right

solution. Suppose that we were to assume that the correct allomorph is

ma- ; since we have postulated that the underlying version of the prefix is

mang- , we would have to derive ma- from underlying mang-. This in turn

would require us to add a third rule that would delete -ng before a nasal-

initial verb root:

(21) mang + manhid → ma + manhid

Before we add this third rule, however, let’s see what happens if we

assume that for these bases the allomorphs for the prefix are mam- or

man- . Note that we already have an assimilation rule that accounts for

which of the allomorphs shows up – we get mam- before an m- initial

root, and man- before an n - initial root. And we already have a rule that

deletes consonants after a suffix that ends in a nasal. If we tweak that

rule slightly, we can derive the forms in (17c) without adding a third

rule:

(22) A nasal or voiceless consonant is deleted when preceded by a nasal

consonant.

The forms in (17c) can then be derived as follows:

(23) mang + manhid → mam + manhid → mam + anhid

mang + nood → man + nood → man + ood

So although it seemed at first that assuming the allomorph of the prefix

in (17c) to be ma- made more sense, looking at the bigger picture, making

the other choice allows us to derive all the allomorphs using a simpler set

of rules. We assume then that this is the right solution. We must keep in

mind, though, that we’ve only looked at a tiny set of data. If we were to

continue looking at the morphology and phonology of Tagalog, we might

decide that the analysis we’ve decided upon here needs to be revised again.

9.5 Lexical strata

What we have seen in this chapter is that building complex words is

frequently accompanied by phonological effects such as assimilation

or vowel harmony. In this section we will see that in some languages

such phonological effects do not apply uniformly across the entire

lexicon of the language, but instead are confined to a subset of the

lexicon. Indeed some languages have two or more different layers to

their lexicons which behave differently in terms of phonological

effects. In this section we will look at three such languages, English,

Dutch, and French.

9.5.1 English As we saw in section 9.2.2 , the suffix - tion is associated with complex and

partially unpredictable allomorphy, both of the suffix itself and of the

bases it attaches to. It turns out that it’s not the only suffix in English that

acts that way. Consider the examples in Table 9.2 .

Sounds and shapes 191

All seven of the suffixes in Table 9.2 are non-native to English. Specifically,

they were borrowed from Latin either directly or by way of French. All of

them are like -tion in showing complex patterns of allomorphy. When they

are added to bases, the final consonants of those bases sometimes change:

(24) sacrifice [s] sacrific-ial [ʃ]

Christ [t] Christ-ian [tʃ]

dialogue [g] dialog-ic [dȢ]

allude [d] allus-ive [s]

historic [k] historic-ity [s]

delude [d] delus-ory [s]

decide [d] decis-ion [Ȣ]

The stress pattern on the base often changes as well:

(25) �ar.chi.tec.ture ar.chi.�tec.tu.ral

�con.tra.ry con.�tra.ri.an

�Ger.man Ger.�ma.nic

�al.ter.nate al.�ter.na.tive

hi.�sto.ric hi.sto.�ri.ci.ty

ex.�crete �ex.cre.to.ry 6

Furthermore, all of these suffixes can attach either to bound bases or to

full words. And all of them prefer to attach to bases that are themselves

non-native to English. The items in the last column in the table are in

parentheses because they are among the few native bases (sometimes the

only one) on which these affixes can be found.

If we now look at suffixes that are native to English – that is, suffixes

that were present in Old English, rather than borrowed from some other

language – we find quite a different pattern: consider Table 9.3 .

Table 9.2. Some non-native suffixes in Engish

Affix Rule

Stem

change

Stress

change

Attaches to

bound

bases

Attaches to

words

Attaches to

non-native

bases

Attaches to

native bases

-al N→A sacrificial architectural minimal architectural yes (tidal)

-ian N→N,A Christian contrarian pedestrian Bostonian yes (earthian)

-ic N→A dialogic Germanic geographic problematic yes no

-ive V→A allusive alternative nutritive impressive yes (talkative)

-ity A→ N historicity historicity atrocity similarity yes (oddity)

-ory V→A delusory excretory perfunctory contradictory yes no

-tion V→N decision revelation perception restoration yes (starvation)

6. Note that the stressing in the last pair in (25) is American English. Speakers of other dialects of English might

stress these words differently.

192 INTRODUCING MORPHOLOGY

When these suffixes attach to bases, they change neither the sounds of

those bases nor their stress pattern:

(26) �pood.le �pood.le.dom

�so.rrow �so.rrow.ful

�neigh.bor �neigh.bor.hood

�her.mit �her.mit.ish

�bo.ttom �bo.ttom.less

�ha.ppy �ha.ppi.ness

Typically they attach freely to either native or non-native bases, but

they do not attach to bound bases; the word vengeful is in parentheses

because it seems to be the only example where one of these suffixes might

be said to be attached to a bound base, but it’s a questionable example,

since venge , according to the OED , is an obsolete word in English.

In fact, the different behavior of the two sets of affixes can be nicely

illustrated by comparing the suffixes - ic and - ish , both of which can take

nouns and make adjectives from them. Compare the adjectives they form

from the non-native base dialogue :

(27) dialogue dialogic dialoguish

[�dɑi.ə.lag] [dɑi.ə.�lɑ.dȢik] [�dɑi.ə.la.giʃ]

The suffixes themselves differ only in their final sound, but - ic both

changes the final consonant of its base and causes its stress pattern to

change, whereas - ish has neither effect. What this illustrates is that

English derivational morphology exhibits two different lexical strata , lay-

ers of lexeme formation that display different phonological behavior.

We can make one more interesting observation about the lexical strata

of English. Consider the derived words in (28) :

(28) a. Two non-native suffixes : -al + -ity sequentiality

-ian + -ity Christianity

-tion + -al organizational

-ive + -ity productivity

Table 9.3. Some suffixes native to English

Affix Rule

Stem

change

Stress

change

Attaches to

bound bases

Attaches to

words

Attaches to

non-native

bases

Attaches to

native bases

-dom N→N none none no kingdom yes yes

-er V→N none none no writer yes yes

-ful N→A none none (vengeful?) sorrowful yes yes

-hood N→N none none no knighthood yes yes

-ish N,A→A none none no mulish yes yes

-less N→A none none no shoeless yes yes

-ness A→N none none no happiness yes yes

Sounds and shapes 193

b. Two native suffixes : -ful + -ness sorrowfulness

-less + -ness hopelessness

-er + -hood riderhood

-er + -less printerless

c. Native outside non-native -al + -ness sequentialness

-ian + -ness Christianness

-tion + -less organizationless

-ive + -ness productiveness

d. Non-native outside native -hood + -al *knighthoodal

-ish + -ity *mulishity

-less + -ity *shoelessity

-ness + -ic *happinessic

Not every suffix can attach to other suffixed words in English, but some-

times we can get complex words with two or more layers of suffixes. As

(28) shows, we can often affix a non-native suffix to a base that already has

a non-native suffix, and similarly put a native suffix on a base that already

has a native suffix. Further, we can often stack up two suffixes if the first

(the innermost in terms of structure) is non-native and the second native.

What’s much more difficult – although not absolutely impossible, as we

will see in Chapter 10 – is to first affix a native suffix and then put a non-

native suffix outside it. This makes perfect sense: non-native suffixes pre-

fer to attach to non-native bases. Once a native suffix has been added to a

base, regardless of whether that base was native or non-native to begin

with, the derived word counts as a native word as far as further affixation

is concerned.

The affixes we’ve looked at here show very clear and very different

behavior, which justifies our saying that English derivational morphol-

ogy displays two different lexical strata. To be honest, not all suffixes in

English are as easily classified as the ones we’ve looked at in this sec-

tion. While the other affixes that are native to English behave much as

those discussed here, this is not the case with all non-native affixes.

Some affixes that are borrowed, and therefore should be part of the

non-native stratum of English, behave more like native affixes in that

they have no phonological effects on their bases and attach indiscrimi-

nately to both native and non-native bases. We will not go further into

the intricacies of English derivation here, but merely point out that

while the outlines of the two strata are quite clear, there is some blur-

ring between them.

9.5.2 Dutch and French Dutch and English are closely related languages, and they share a history

of contact with French and Latin. It is therefore not surprising that the

morphology of Dutch exhibits two lexical strata, just as English does.

We’ll give just a brief illustration here. In Dutch, the suffix - heid ‘-ness’ is

of native origin, and - iteit ‘-ity’ is non-native. As we saw in English, the

native suffix attaches easily either to native or non-native bases, but the

non-native one can only occur on non-native bases (Booij 2002 : 95):

194 INTRODUCING MORPHOLOGY

(29) - heid blindheid ‘blindness’ (native base)

diversheid ‘diverseness’ (non-native base)

-iteit *blinditeit (native base)

diversiteit ‘diversity’ (non-native base)

As in English, non-native affixes can occur on either bound bases or free

words, whereas native affixes only occur on free words. And as in English,

if a word contains both native and non-native affixes, the native ones

must occur outside the non-native ones.

What may be somewhat more surprising is that French , a language

itself descended directly from Latin, also shows signs of lexical strata

(Huot 2005 ). French suffixes can be divided into those that are called

‘popular’ (in French ‘populaire’) and those that are called ‘learned’ (in

French ‘savant’). The former have descended from Latin undergoing all

the sound changes that the vocabulary of French has been subject to. The

latter come from scholarly Latin by borrowing later in the history of

French. Popular suffixes typically attach to popular roots, and learned suf-

fixes to learned roots (Huot 2005 : 65):

(30) Popular suffixes that prefer popular roots

-age doublage ‘doubling’, grattage ‘scratching’

-ier jardinière ‘gardener’, pétrolier ‘oil-tanker’

-eux chanceux ‘lucky’, venteux ‘windy’

Learned suffixes that prefer learned roots

-ion inscription ‘inscription’, punition ‘punishment’

-if actif ‘active’, duratif ‘durative’

-aire articulaire ‘articular’, réfractaire ‘refractory’

Popular suffixes sometimes do attach to learned roots, but learned suf-

fixes do not attach to popular roots:

(31) Popular suffix on learned root

infectieux ‘infectious’, torrentueux ‘torrential’

Popular suffixes tend not to attach to already suffixed words, but learned

suffixes can sometimes attach to other learned suffixes:

(32) démiss + ion + aire démissionnaire ‘one who has resigned’

And finally, popular roots sometimes have corresponding learned

allomorphs:

(33) Popular Learned

angle ‘angle’ angul + aire ‘angular’

cercle ‘circle’ circul + aire ‘circular’

peuple ‘people’ popul + aire ‘popular’

So what we see here is that there are two sets of affixes that display

somewhat different patterns of behavior. The lexicon of French thus gives

us another example where morphology is not neat and homogeneous, but

instead seems to be organized into two relatively discrete layers.

Sounds and shapes 195

Exercises 1. The following forms are from the now-extinct language Wappo , until

recently spoken in California (Thompson, Park, and Li 2006 : 125–7). The first set of examples is glossed for you. Using them as a model, first analyze the next two sets of data and then answer the questions below.

a. olol – asaʔ ‘is making X dance’ dance – CAUS:DUR

olol – is – taʔ ‘made X dance’ dance – CAUS – PST

olol – is – ya:miʔ ‘will make X dance’ dance – CAUS – FUT1

olol – asiʔ ‘make X dance!’ dance – CAUS: IMP

olol – asa – lahkhiʔ ‘isn’t making X dance’ dance – C AU S: DU R– N EG

olol – is – ta – lahkhiʔ ‘wasn’t making X dance’ dance – C AU S – P ST – N EG

olol – is – lahkhiʔ ‘don’t make X dance!’ dance – C AU S : I M P – N EG

b. hicasaʔ ‘is making X pound Y’ hicistaʔ ‘made X pound Y’ hicisya:miʔ ‘will make X pound Y’ hicasiʔ ‘make X pound Y!’ hicasalahkhiʔ ‘isn’t making X pound Y’ hicistalahkhiʔ ‘wasn’t making X pound Y’ hicislahkhiʔ ‘don’t make X pound Y’

Summary In this chapter we have looked at the connection between phonology

and morphology. Morphemes frequently have allomorphs,

phonologically distinct variants that occur in different environments.

Sometimes, as we saw, those environments are predictable, and we can

postulate phonological rules that explain the distribution of the

allomorphs. Indeed, we can often postulate a single underlying

phonological form from which all the allomorphs can be derived. We

have looked at a number of typical kinds of phonological rules that

explain allomorphy in various languages: assimilation of various sorts,

epenthesis, vowel harmony, and intervocalic voicing . We have also seen

that not all allomorphy is entirely predictable; as the morphology of

English shows, it can be quite unpredictable where one allomorph or

another shows up. And we have looked at morphophonological

processes that depend on syllable structure. Finally, we have looked at

three cases in which different segments of the lexicon constitute

different lexical strata displaying different phonological behavior or

different patterns of allomorphy.

196 INTRODUCING MORPHOLOGY

c. hintoʔasaʔ ‘is making X sleep’ hintoʔistaʔ ‘made X sleep’ hintoʔisya:miʔ ‘will make X sleep’ hintoʔasiʔ ‘make X sleep!’ hintoʔistalahkhiʔ ‘wasn’t making X sleep’ hintoʔislahkhiʔ ‘don’t make X sleep!’

Once you have segmented and glossed the (b) and (c) sets of data, list all the allomorphs of all morphemes. Is all of the allomorphy predictable? Where you can, explain informally what seems to determine the distribution of the allomorphs.

2. In Tagalog , the circumfix ka … an creates nouns designating the class or group of whatever the base denotes. Identify all allomorphs in the forms below and explain their distribution using informal phonological rules (Schachter and Otanes 1972 : 101):

banal ‘devout’ kabanalan ‘devoutness’ bukid ‘field’ kabukiran ‘fields’ bundok ‘mountain’ kabundukan ‘mountains’ lungkot ‘sadness’ kalungkutan ‘sadness’ pangit ‘ugly’ kapangitan ‘ugliness’ pulo ‘island’ kapuluan ‘archipelago’ dagat ‘sea’ karagatan ‘seas’ dalita ‘poverty’ karalitaan ‘poverty’ Tagalog ‘a Tagalog’ katagalugan ‘the Tagalogs’

3. The Dutch diminutive has several allomorphs. Determine what they are, and explain their distribution (De Haas and Trommelen 1993 : 279):

a. gum ‘eraser’ gumetje b. roman ‘novel’ romanetje c. parasol ‘parasol’ parasoletje d. kar ‘cart’ karetje e. lichaam ‘body’ lichaampje f. pruim ‘plum’ pruimpje g. bezem ‘broom’ bezempje h. koning ‘king’ koningkje i. haring ‘herring’ haringkje j. streep ‘stripe’ streepje k. kabinet ‘cabinet’ kabinetje l. almanak ‘almanac’ almanakje m. wereld ‘world’ wereldje n. banaan ‘banana’ banaantje o. tuin ‘garden’ tuintje p. kuil ‘hole’ kuiltje q. altaar ‘altar’ altaartje

HINT: Examples a–d have short vowels in their final syllables. Examples n–q have long vowels or diphthongs in their last syllables.

4. Form the plurals of the following words in English , and transcribe them in the IPA:

lip lathe pot kiss

Sounds and shapes 197

tack buzz club church thud garage thug judge cliff arena path hero stove

a. How many allomorphs are there for the plural morpheme in English? b. Which of the allomorphs makes the best candidate for the underlying

form of the plural morpheme? c. Formulate a phonological rule that derives the various allomorphs of

the plural morpheme from the underlying form.

5. Thinking about the pattern you discovered in exercise 4, now consider

the plurals of the following words:

wolf calf house mouth elf knife

How do these differ from the plurals you discussed above?

6. In exercise 1 of Chapter 5 you looked at a process of infixation in the Austronesian language Leti . Some of the data you looked at there are given again in (a) (Blevins 1999 ):

(a) kakri ‘cry’ kniakri ‘the act of crying’ pali ‘float’ pniali ‘the act of floating’ sai ‘climb’ sniai ‘the act of climbing’ teti ‘chop’ tnieti ‘the act of chopping’ vaka ‘ask’ vniaka ‘the act of asking’

vanunsu ‘knead’ vnianunsu ‘massage’ = ‘the act of kneading’

Now compare those data to the ones in (b): (b) kili ‘look’ knili ‘the act of looking’

kini ‘kiss’ knini ‘the act of kissing, kiss’ surta ‘write’ snurta ‘the act of writing, memory’ tutu ‘support’ tnutu ‘the act of supporting, support’ virna ‘peel’ vnirna ‘the act of peeling’

What are the two allomorphs of the nominalizing affix in Leti? What determines which allomorph goes with which bases?

7. Consider the data below from the Mayan language Tzutujil (Dayley 1985 : 206–7):

k’uluuj ‘to meet, encounter’ k’ulaani ‘married’

jaqooj ‘to open’ jaqali ‘open’ d’eb’ooj ‘to stain with thick liquid’ d’eb’eli ‘thick (of liquid)’ b’olooj ‘to twine, boil meat’ b’olaani ‘cylindrical’ d’oyooj ‘to cut with an axe’ d’oyoli ‘cuttable’ wonooj ‘to push with the head’ wonoli ‘bent over’ ketooj ‘to cut with a very sharp

machete’ keteli ‘discoid, wheel-shaped’

198 INTRODUCING MORPHOLOGY

ch’ikooj ‘to clean land for tilling’ ch'ikili ‘stuck in’ jotooj ‘to raise’ jotoli ‘be above’ ch’anooj ‘to spank a naked person’ ch'anali ‘naked’

Identify all allomorphs and try to state the conditions under which each occurs. What morphophonological process is illustrated by the data in the second column?

8. The following data are from the constructed language (conlang) Dothraki that figures in the HBO TV series Game of Thrones . Consider the forms below. Segment the forms in the second column into a base and a suffix and give a meaning for the suffix. What are the allomorphs of this suffix? Propose an underlying form and phonological rules to derive the surface forms from the underlying form (data from http://wiki.dothraki.org/ Derivational_morphology ).

fonak ‘hunter’ fonakasar ‘hunting party’ oqet ‘sheep’ oqeteser ‘flock of sheep’ zir ‘bird’ zirisir ‘flock of birds’ jano ‘dog’ janosor ‘pack of dogs’

9. Consider the data below from Hopi , a Uto-Aztecan language spoken in the southwestern part of the United States (data from Jeanne 1982 : 246–8). Describe the allomorphy that is exhibited in the future tense in Hopi. Propose an underlying form for each verb and state the phonological rules that can be used to derive the surface forms from the underlying forms.

non-future future ʔ��ya ʔ�yni ‘plant’ nöösa nösni ‘eat’ w��vi w�pni ‘climb’ p��wi p�wni ‘sleep’ qaaci qacni ‘be in position (inanimate)’ s��wi s�wni ‘dizzy’ caama camni ‘take out’ heeva hepni ‘seek’

10. Pame is an Otomanguean language spoken in Mexico. Identify the singular/dual and plural affixes in Pame. If you observe allomorphy of either affix or base, describe the allomorphy, and propose underlying forms and phonological rules deriving the surface form from the underlying form (data from Gibson and Bartholemew 1979 : 310).

sg or dual pl ŋgobέʔεt mbέʔεt ‘flag’ ŋgodὲocʔ ndὲocʔ ‘bridge’ ŋgokhwè ʔ ŋkhwè ʔ ‘bean’ ŋgosą́oŋ nsą́oŋ ‘night’ ŋgolhwą́ nlhwą́ ‘ear of corn’ ŋgopʔó ho mbʔo ́ho ‘seat’ ŋgokwą́ŋ ŋgwą́ŋ ‘tree, piece of wood’ ŋgohwε�́ʔ hwε�́ʔ ‘thorn’ ŋgomhε�́ mhε�́ ‘tortilla’

KEY TERMS Item and Arrangement

Item and Process

Word and Paradigm

realizational model

multiple exponence

Lexical Integrity Hypothesis

blocking

rival affixes

competition

bracketing paradoxes

CHAPTER OUTLINE In this chapter you will get your first taste of the theoretical

challenges that morphologists face.

♦ We will consider what the best way is to characterize

morphological rules.

♦ And we will examine several theoretical controversies

that have occupied morphologists in recent years: the

extent to which rules of morphology and rules of syntax

can interact, the nature of blocking, the best way to

analzye so-called bracketing paradoxes, and the

characterization of affixal polysemy.

Theoretical challenges

10 C H A P T E R

200 INTRODUCING MORPHOLOGY

10.1 Introduction

Up to this point, we’ve spent a lot of time looking at the way morphology

works in languages – what kinds of morphemes there are, what to call

them, how to analyze data, and so on. This is an important and necessary

first step to becoming a morphologist, but there’s more to morphology

than just being able to analyze data. When we study morphology, or

indeed any of the other subfields of linguistics, we have a much larger

goal in mind, which is to characterize and understand the human lan-

guage faculty. Put a bit differently, our ultimate goal as linguists is to

figure out the way in which language is encoded in the human mind. For

morphologists, our specific goal is to figure out how the mental lexicon is

encoded in the mind. Doing so requires us to model the mental represen-

tation of language, to make claims about exactly what morphological

rules look like, and to propose hypotheses about what is possible in

human languages and what is impossible.

A good hypothesis about language is one which is empirically testable:

it should be clear what sort of data to look for that would disprove the

hypothesis. To illustrate this, let’s look first at two examples of theoretical

hypotheses that have been proposed by morphologists. I start with these

two precisely because they make clear claims about what sorts of mor-

phology we should expect to find in languages and because these claims

have subsequently been disproven.

The first hypothesis we’ll look at is called the Righthand Head Rule:

(1) The Righthand Head Rule (Williams 1981 : 248)

“In morphology, we define the head of a morphologically complex

word to be the righthand member of that word.”

You’ll recall from Chapter 3 that the head of a compound was the mor-

pheme that determined the syntactic and semantic category of the com-

pound. Clearly, in English, it’s the righthand element in the compound

that’s the head (so sky blue is syntactically an adjective like blue and semanti-

cally a type of blue as well). More broadly, the head of a word is that mor-

pheme that determines the category of the word, and in languages that

have gender in nouns, or inflectional classes in nouns and verbs, the head

determines the gender or class of the word as well. For example, in German,

the suffix - heit attaches to adjectives to form nouns, specifically feminine

nouns. Since - heit determines the category and gender of the derived noun,

it is the head of the word. The Righthand Head Rule is a theoretical hypoth-

esis that basically says that all compounds should be right-headed, and only

suffixes (and not prefixes) can be the heads of words.

At first glance, this hypothesis is plausible enough when we look at

English . Compounds are indeed right-headed, and for the most part in

English it’s the suffix that determines the category of a complex word.

However, the Righthand Head Rule can easily be disproven by looking at

data from other languages. For example, in Chapter 3 we saw that both

Vietnamese and French have left-headed compounds:

Theoretical challenges 201

(2) French timbre poste ‘stamp-post’ = ‘postage stamp’

Vietnamese nhá thuong ‘establishment be-wounded’ = ‘hospital’

And it is not hard to find languages in which prefixes change the cate-

gory of words and determine their gender or class. Even English has at

least one prefix that changes category, and therefore would have to be

recognized as the head of the derived word:

(3) de- debug

delouse

de-ice

The prefix de- attaches to nouns and makes verbs. Similarly, Swahili has

a prefix ku- that forms nouns from verbs:

(4) From Vitale ( 1981 : 10)

ku-tafutwa kwa Juma

-ing-search for Juma

‘the searching for Juma’

Since this prefix determines the category of the derived word, we would

have to consider it to be the head of the word. These examples show, then,

that the Righthand Head Rule cannot be correct.

A second theoretical proposal that turns out not to be correct is the

Unitary Base Hypothesis :

(5) The Unitary Base Hypothesis (Aronoff 1976 : 48)

“We will assume that the syntacticosemantic specification of the

base, though it may be more or less complex, is always unique. A

WFR [word formation rule] will never operate on either this or

that.”

The Unitary Base Hypothesis in effect says that we should never expect to

find in a language a morpheme that attaches to bases of two different

categories, say adjective and noun, or noun and verb. We have seen, how-

ever, that there are many affixes that can attach to more than one base:

-ize in English attaches to both adjectives ( legalize ) and nouns ( unionize ) to

form verbs, and - er attaches to both verbs ( writer ) and nouns ( villager ) to

form nouns. 1 It seems that affixes sometimes (in fact frequently!) do

attach to “either this or that.” The Unitary Base Hypothesis makes a clear

claim about what we should expect to find in the languages of the world,

but that is not in fact what we find.

Why start out a chapter on theory with two incorrect hypotheses?

What is important is that these hypotheses are testable: we know what

sort of data to look for, and having looked for those data can deter-

mine that these hypotheses cannot correctly characterize our theo-

retical model of the mental lexicon. Notice that I haven’t talked about

1. Aronoff is aware of examples like these, and is forced to argue that there are two different - ize suffi xes

and two different -er suffi xes that are homophonous, that is, that sound identical. It is generally accepted,

though, that this is not a strong defense of the Unitary Base Hypothesis, and that the hypothesis is therefore

almost certainly incorrect.

202 INTRODUCING MORPHOLOGY

proving hypotheses to be true. In fact, it is never possible to prove a

scientific hypothesis to be true: since there will always be some lin-

guistic data we have not yet looked at, there is always some chance

that further study will prove a hypothesis to be false. We take a theo-

retical hypothesis to be sound just as long as we have not yet found

evidence against it.

Good hypotheses must therefore be testable, and we would like them to

explain a wide range of data. But there is one more thing we would want

of a good theoretical hypothesis: it must also be simple. Since generative

linguists are ultimately concerned with the mental representation of lan-

guage, part of their concern is to explain how children acquire knowledge

of those mental representations so quickly and with such ease. We assume

that the simpler our proposed mental representations, the easier it would

be for children to acquire them, and therefore the more plausible they

should be.

With this in mind, we can now go on to look at a number of other theo-

retical proposals for characterizing our mental representation of mor-

phology that are less easy to dismiss. Keep in mind that we can look at

only a few interesting points of morphological theory here; in the last

three decades there has been a great deal written about morphological

theory that we will not be able to cover. So what I hope to do here is to give

you a taste of theory and to whet your appetite for further study. With

that in mind, as we go along I’ll suggest some original sources that the

interested student might take a look at in order to delve more deeply into

morphological theory.

10.2 The nature of morphological rules

Up to this point we’ve talked about morphological rules for affixation,

compounding, internal stem change, and other means of creating new

words, but we have only characterized those rules informally. One of the

important parts of modeling the mental lexicon is to characterize mor-

phological rules formally. In this section we will look at different formal

systems for characterizing morphological rules and try to see how they

make different claims about the sorts of morphology we ought to find in

the languages of the world.

10.2.1 Morphemes as lexical items: Item and arrangement morphology Let’s take another look at one of the informal rules of word formation

that we proposed in Chapter 2 :

(6) -ize attaches to adjectives or nouns of two or more syllables where

the final syllable does not bear primary stress. For a base ‘X’ it

produces verbs that mean ‘make/put into X’.

One way of making this sort of rule formal is to assume that in our mental

lexicons the morpheme -ize has a lexical entry, just as free morphemes do,

and that part of its lexical entry is the following:

Theoretical challenges 203

(7) The -ize rule (more formal version)

-ize structural information: [ [ ] A,N __] V

semantic information: ‘make A; make/put into N’

phonological information: […σσ W aiz]

The first line of this rule gives structural information: it says that

-ize is a suffix that attaches to nouns or adjectives, and produces

verbs. In fact, it says somewhat more than this, as the brackets indi-

cate that when the suffix is added, a bit of hierarchical structure is

formed:

(8) V

union ize

The second line of the rule tells us what the resulting word means;

this part of the rule can be formalized as well, using special nota-

tion, but we will not do so here. We’ll merely say that when the

piece - ize is added to a base, it also adds the meaning ‘make A or

make/put into N’. Finally, the third line uses the Greek letter sigma

(σ) to stand for ‘syllable’, and the subscript W to stand for a ‘weak’ or unstressed syllable. This, then, encodes the information that - ize

requires a base that has at least two syllables, the last of which must

not bear stress.

This kind of theory in effect makes a claim that affixes are just like

free morphemes in that they have lexical entries that include various

types of information. The only difference between the entry for an

affix and for a base is that the affix is a bound morpheme, and there-

fore as part of its structural information requires another category to

attach to. Theories that propose rules of this sort are traditionally

referred to as Item and Arrangement (IA) theories, because they claim

that morphemes have independent existence in the mental lexicon

with their own structural, semantic, and phonological information,

and that they can be arranged hierarchically into words. In a way, such

theories make the claim that complex words have a sort of internal

syntactic structure.

10.2.2 Morphemes as processes and realizational morphology This is, of course, not the only way of formalizing morphological rules.

Indeed, many morphologists believe that it is a mistake to count mor-

phemes as ‘things’ that have their own independent existence in our

mental lexicons and to treat morphology as the internal syntax of words.

The alternative, they argue, is to allow only free morphemes to have lexi-

cal entries of their own, and to introduce bound morphemes using rules

that contain the phonological form and the semantic content of the

bound morpheme. The - ize rule might look like (9) in such a theoretical

framework:

204 INTRODUCING MORPHOLOGY

2. For the moment, we can set aside how the phonological information is stated in this sort of theory.

(9) The -ize rule

X → Xize, where X = N,A and Xize = V meaning ‘make, put into X’ 2

In the case of the -ize rule, this may not look terribly different than the

lexical entry we proposed above. But consider the sort of rule we

would have to propose for an irregular past tense form like sang in

English. Since the past tense in this form is created by a process of

ablaut, we can propose a rule that changes the vowel [i] to [æ] to pro-

duce sang .

(10) Irregular past rule for the sing, swim, ring class of verbs

CiN → CæN

[−past] [+past]

If C stands for any consonant, and N for any nasal consonant, we can

express the vowel change that takes place in the past tense very simply

with such a rule. It is not so easy to see how to express an internal vowel

change using the ‘morpheme as thing’ model; it doesn’t seem to make

sense to say that the past tense in such verbs is a morpheme that consists

of only the vowel [æ]. The sort of theoretical framework that treats mor-

phemes as parts of rules is sometimes called an Item and Process (IP)

theory, because morphological rules are conceived as operations or pro-

cesses that act on free morphemes.

Related to the Item and Process model is the Word and Paradigm (WP)

model. WP models are also sometimes known as realizational models .

They are often proposed to account for inf lectional word formation in

languages that have complex paradigms, especially the sort of paradigm

which exhibits a characteristic called multiple exponence . Multiple

exponence occurs when particular inf lectional characteristics – say past

tense, or third person – are signaled by more than one morpheme in a

word. For example, in the Latin second person singular past tense verb

form am ā visti ‘you-sg loved’ we might say that the root is am and the stem with theme vowel am ā. The past tense is signaled by the morpheme - v , and the second person singular morpheme is - isti . But this is not quite

correct, because the morpheme - isti is a person/number ending that is

only used in the past tense; in some sense, -isti bears the meaning of past

tense along with its person/number meaning. The past tense meaning is

signaled twice in the word ām ā visti . This is what we mean by multiple exponence. In an IA theory of morphology, multiple exponence is prob-

lematic. IA models work best when there is a one-to-one correspondence

between morphemes and meanings. In other words, each morpheme

expresses one and only one inf lectional or derivational meaning. WP or

realizational models , on the other hand, do not separate out mor-

phemes into discrete pieces, but rather state rules that associate mean-

ings (single or multiple) with complex forms. For example, a realiza-

tional rule for the second person singular past tense form of the verb in

Latin might be (11) :

Theoretical challenges 205

(11) X ]

+past

2nd person

sg.

[Xvisti 



   





   

→

The realizational model does not recognize - v and - isti as separate mor-

phemes. Instead, the form am ā visti is conceived as a unit that expresses past tense, second person, and singular number conjointly. The existence

of multiple exponence causes no problems in a realizational model,

because inflected words need not be segmented into discrete pieces.

10.2.3 Can we decide between them? You might at this point be wondering if it really makes a difference

whether we conceive of morphemes as things with discrete meanings and

their own lexical entries, or as parts of morphological rules that realize

unanalyzable words with complex meanings. Our conception of what

morphological rules look like really does matter, because it makes predic-

tions about what sorts of morphology we should expect to find in the

languages of the world. On the one hand, Item and Arrangement theories

predict that morphology ideally should be agglutinative, with words seg-

mentable into several pieces, each of which has a distinct meaning. In

some languages this is the case – recall our sketch of Turkish in Chapter 7 .

On the other hand, Word and Paradigm theories, although they do not

strictly preclude agglutinative morphology, lead us to expect that mor-

phology typically ought not to be agglutinative; rather, it should contain

lots of multiple exponence, as is the case in Latin, but not in Turkish. The

problem, of course, is that neither of these predictions is quite right.

Some languages are more agglutinative than others. Some languages have

lots of multiple exponence, and others have little or none.

So it does make a difference which kind of theory we choose. But the

choice is made difficult by the complexity and variety of morphology we

actually do find in the languages of the world. In addition, nothing rules

out the possibility that both models might have merits. For example, it is

possible that IA theories do a better job modeling derivational morphol-

ogy in many languages and IP or WP models a better job of modeling

inflectional morphology (see Borer 2013 for an argument to this effect).

The choice between models is made even more difficult by the fact that

each of these models is not really just a single theory, but a kind of umbrella

that encompasses a number of different theories. For example, in their sim-

plest form IA models say that the correspondence between meanings and

‘pieces’ ought to be one-to-one, but IA models need not say this. It is possible

to propose an IA model in which the relationship between pieces and mean-

ings is ideally, but not strictly, one-to-one. Similarly, there are a number of

different versions of WP or realizational models that are plausible. We will

not be able to go into the various theoretical possibilities here, but you will

no doubt encounter them in a more advanced course in morphology. It

206 INTRODUCING MORPHOLOGY

seems safe to say, though, that each theory must be tested against a wide

variety of languages with all different sorts of word formation, and a wide

variety of inflectional and derivational word formation processes including

not only affixation, but also compounding, conversion, reduplication, and

templatic morphology. We will end this section by simply saying that the jury

is still out on whether IA, IP, or realizational models of morphology consti-

tute better models of how morphology is organized in the human mind.

Challenge

Review the examples of reduplication that we discussed in Chapter 5 .

Which of the three models discussed above (IA, IP, or WP) is best

suited to modeling rules of reduplication?

For students interested in studying questions of theoretical models in

more detail, a good place to start is Spencer ( 1991 ). Early generative theo-

ries that follow the Item and Arrangement model are Lieber ( 1980 ), Selkirk

( 1982 ), DiSciullo and Williams ( 1987 ), and Lieber ( 1992 ). Generative theo-

ries that take an IP, WP, or generally realizational perspective are Aronoff

( 1976 ), Anderson ( 1982 , 1992 ), and Stump ( 2001 ). A more recent theoretical

development that embraces aspects of both models is called Distributed

Morphology. The interested student can read about Distributed Morphology

in Halle and Marantz ( 1993 ) and Harley and Noyer ( 1999 ). And this only

begins to scratch the surface with respect to theoretical models, so if you

go on to study morphology further, you will find that there are many

competing theoretical models that I have not mentioned here.

10.3 Lexical integrity

In Chapter 8 we considered the relationship between morphology and syn-

tax and saw several ways in which these two segments of the grammar are

closely intertwined. The relationship between morphology and syntax

indeed has given rise to one of the most interesting and longest-standing

theoretical controversies among morphologists. Early in the history of gen-

erative morphology, several theorists proposed what has come to be called

the Lexical Integrity Hypothesis . One version of this hypothesis is (12) :

(12) Lexical Integrity Hypothesis

“No syntactic rule can refer to elements of morphological structure.”

(Lapointe 1980 : 8)

What this means is that syntactic rules – phrase structure or movement rules,

for example – cannot look into words and manipulate their internal structures.

The Lexical Integrity Hypothesis requires that morphological and syntactic

rules be fundamentally different. Morphological rules are concerned with

affixes and bases, with rules of reduplication and ablaut, and so on, that affect

the internal structure of words. Rules of syntax take words as unanalyzable

wholes and form them into phrases and sentences. One way of ensuring this

separation between morphology and syntax that was proposed early in the

Theoretical challenges 207

history of generative morphology was to order morphological rules before syntac-

tic rules, as if there were something like a linguistic assembly line that started

with the smallest units of structure and proceeded to larger and larger units:

FIGURE 10.1 One way of organizing the grammar

Rules of syntax

Rules of morphology

The model in Figure 10.1 ensures the separation of morphology and

syntax because syntax only gets to look at already-formed words. Many

linguists no longer believe that rules operate in a strict ‘assembly-line’

fashion, but nevertheless continue to maintain that morphological and

syntactic rules must be kept separate from one another.

The Lexical Integrity Hypothesis is both plausible and testable: indeed

it makes a clear prediction that we should never find fully formed phrases

or sentences inside words. If phrases are formed by syntactic rules and

syntactic rules are separate from morphological rules, words should not

contain phrases or sentences. However, as we saw in Chapter 8 , there are

some sorts of words that do seem to contain phrases and even sentences.

Among these are phrasal compounds like those in (13) :

(13) stuff-blowing-up effects

bikini-girls-in-trouble genre

comic-book-and-science-fiction fans

God-is-dead theology

It is also possible in some languages – including English – to attach pre-

fixes or suffixes to whole phrases or even whole sentences (see Chapter 8

for more examples):

(14) a. English

mouse- and rat-like

pre- and post-war

I-can-do-it-too-ness

b. Turkish (Lewis 1967 : 41)

Tebrik ve teşekkür-ler-im-i congratulation and thank- PL-MY-ACC

‘my congratulations and thanks’

As Bauer, Lieber, and Plag ( 2013 ) show, in English quite a few affixes can

attach to phrases and sentences and a few prefixes ( pre-, post-, hyper-, hypo- ) can

208 INTRODUCING MORPHOLOGY

3. In (16) ‘e’ indicates that the subject NP is empty, and ‘t’ stands for ‘trace’, which marks the place from which

a constituent has been moved.

themselves be conjoined and attached to a base. Turkish allows some of its

inflectional endings to apply equally to two conjoined bases. If syntactic

phrases and sentences can act as the bases for complex words, then the strict

separation of morphological and syntactic rules required by the Lexical

Integrity Hypothesis cannot be correct.

Linguists have therefore proposed alternatives to the Lexical Integrity

Hypothesis. Some linguists have proposed models in which limited inter-

action between morphology and syntax is possible. Others, however, have

taken a more radical approach, arguing that there should be no separa-

tion between morphology and syntax, and that syntactic rules should be

responsible for at least some sorts of word formation.

One sort of word formation that has been used to argue for this hypothesis

is noun incorporation, which you looked at briefly in Chapter 8 . To refresh

your memory, consider the Mohawk examples in (15) (Baker 1988 : 20):

(15) a. Ka-rakv ne sawatis hrao-nuhs-aʔ

3 N - be.white DET John 3 M -house- SUF

‘John’s house is white’

b. Hrao-nuhs- rakv ne sawatis

3 M - house-be.white DET John

‘John’s house is white’

In (15a) the noun ‘house’ is part of an independent noun phrase (NP). In

(15b) , however, it has been incorporated into the verb ‘be white’ so that

together they form a single word. In a syntactic analysis of noun

incorporation (15b) starts out with the noun ‘house’ part of an NP with

‘John’s’. However, a syntactic movement rule plucks ‘house’ from its NP

and attaches it to the verb ‘be white’, as (16) illustrates: 3

(16) From Baker ( 1988 : 20)

S S

NP N N V NP N

NP VP

V NP NPV

NP VP

e e tbe.white be.whiteJohn’s John’shouse house

There is a great deal that might be said about the pros and cons of this

analysis, although we cannot do so here. I should point out, though, that

while some linguists find the evidence for this analysis convincing, others

are less convinced and prefer to work within theoretical models that treat

noun incorporation as the result of morphological rules, and allow less

Theoretical challenges 209

interaction between morphology and syntax. As with many other theo-

retical issues in morphology, the jury is still out on the best way to treat

the relationship between morphology and syntax.

Again, the interested student can read further on this topic. A general

article on the status of the Lexical Integrity Principle is Lieber and Scalise

( 2007 ). For the debate on the analysis of noun incorporation, you might

want to compare the arguments in Baker ( 1988 , 1996 ) to those in Mithun

( 1999 ). A good general overview of the topic of noun incorporation can be

found in Massam ( 2009 ).

10.4 Blocking, competition, and affix rivalry

Consider the data in (17) :

(17) a. curious curiosity

generous generosity

impetuous impetuosity

b. glorious ?gloriosity

furious ?furiosity

gracious ?graciosity

Generally, as the examples in (17a) suggest, it seems possible in English

to derive an - ity noun from an adjective that ends in the suffix - ous . But in

some cases – for example, those in (17b) – the - ity form seems very odd, if

not downright impossible. Why should this be? The examples in (17) illus-

trate a phenomenon that morphologists call blocking . In its simplest

form blocking is said to occur when there is a simplex word that bears the

same meaning or fulfills the same function as the purportedly non-

existent derived word. In the case of the examples in (17b) , but not (17a) ,

there are simple, underived nouns from which the -ous adjectives are

formed: glory, fury , and grace . Some morphologists hypothesize that since

those words exist in English, they block or preclude the formation of the

- ity nouns. The - ous adjectives in (17a) do not derive from free morphemes

(they are built on bound bases), and therefore can form nouns by affixa-

tion of - ity .

We might also wonder whether blocking occurs in the formation of

nominalized verbs in English:

(18) ?occuration occurrence ?occurment

reservation ?reservance ?reservment

?amusation ?amusance amusement

As the examples in (18) suggest, there are a number of different suf-

fixes that form nouns from verbs, among them -ation, -ance, - ure , and - ment .

Sets of affixes like these that have the same function or meaning are

sometimes referred to as rival affixes . Rival affixes are sometimes said to

be in competition with each other, the idea being that for each verbal

base, only one affix is possible. Other nominalizing affixes cannot attach,

or at least they sound very odd to the ears of a native speaker. The idea is

210 INTRODUCING MORPHOLOGY

that the existence of one nominalization blocks or precludes the exis-

tence of others.

Blocking is also said to occur in inflectional paradigms as well as in

cases of derivation. For example, in English the suppletive past tense form

went is said to block the formation of a regular past tense form *goed , and

the existence of the irregular plural form children is said to block the for-

mation of the irregular plural *childs .

The obvious question that a theorist might ask is why blocking should

occur. One possible answer is that languages tend to avoid synonymy. Once

we have a word that means ‘more than one child’ or ‘the process or result of

reserving’, why would we need another? But is it true that languages always

avoid synonymy? Some evidence that seems to support this hypothesis comes

from the ‘doublets’ or ‘triplets’ like the ones in (19) – bases that do take more

than one plural or nominalizing affix. Consider the examples in (19) :

(19) a. brothers brethren

b. commission committal commitment

Superficially, these examples seem to provide evidence against block-

ing, but actually they don’t. In (19a) we have two plurals of the word

brother , one the regular plural brothers and the other an archaic plural

brethren . In (19b) we can see that the verb commit has three different nomi-

nalizations, not just one. But these examples do not really argue against

blocking, because we don’t in fact have cases of perfectly synonymous

forms. Although brethren was at one time just a plural of brother , it has

specialized in meaning and is used in religious contexts to refer to mem-

bers of the same church. In the case of the different nominalizations of

the verb commit , each one has a specific lexicalized meaning. Commission ,

for example, is the act of committing, and a commission an order to create

a piece of art. A committal is an order to send someone to prison or to the

hospital. And a commitment is a pledge of certain sorts. Blocking doesn’t

occur in these cases because we do not have words that are synonymous.

But perhaps we shouldn’t be too hasty to draw conclusions here.

Challenge

Is it really true that languages avoid synonymy? Try to think of

examples of words that you might consider to be perfectly

synonymous. You may consider simplex as well as complex words.

The answer to my challenge is that it’s not all that clear that synonymy

avoidance is anything more than a tendency. For example, in English we

have simplex items like couch and sofa that for all intents and purposes

tend to be synonyms, as are the compounds groundhog and woodchuck in

North American English. With respect to derivation and inflection, at

least in English, it turns out that there is evidence that synonymy is not

always avoided, and indeed that blocking need not occur. Consider the

examples in (20) :

Theoretical challenges 211

4. In COCA, examples with these suffi xes are sometimes hyphenated and sometimes not. Here, I have pre-

served the hyphenation found in COCA.

(20) purity pureness

generosity generousness

impetuosity impetuousness

As the examples in (20) show, alongside a noun formed with -ity it is

always possible to form a noun with -ness . Although occasionally the two

words have distinct meanings (e.g., monstrosity and monstrousness mean dif-

ferent things), much of the time the -ity and -ness words do appear to be

synonymous, contrary to our hypothesis. It is therefore not possible to say

that blocking always occurs to avoid synonymy

Further examples of synonymous derivation in English can be found

with other sets of rival affixes. For example, for the suffixes that derive

nouns from verbs that we mentioned above, COCA gives us examples like

omitment alongside the more frequent omission , or disrupture alongside

disruption , with no apparent difference in meaning. The affixes - ish ,

- esque , and - like are another set of rival affixes. Again, some searching in

COCA turns up cases of the same base with each of these suffixes where

the derived words in context seem to have the same meaning, for exam-

ple, Barbie-ish, Barbieesque , and Barbie-like or iPod-ish, iPod-esque , and iPod-

like . 4 A final rival set of affixes is - hood , - ship , and - dom which form

abstract nouns from other nouns. These can be found in COCA on bases

like guru or student , with the resulting derived words guruhood , guruship ,

and gurudom , or studenthood , studentship , and studentdom apparently all

used with the same meaning. Bauer, Lieber, and Plag ( 2013 ) give more

examples like those above. The existence of examples like these calls the

principle of blocking into question, at least with respect to derivational

morphology.

There is also some evidence that blocking isn’t a hard and fast rule in

inflection either. For example, although adult English speakers always use

the irregular plural mice when referring to furry little vermin, they vary

between mice and mouses when referring to computer pointing devices (the

interested student can check this out for herself in COCA). And there’s a

great deal of variation between the past participle forms of the verb dive

(either dived or dove ) and forget ( forgot, forgotten ). So blocking does not seem

to be inevitable in inflectional forms either.

Does blocking exist? On the basis of our intuitions, morphologists have

long had the feeling that it does, but looking at actual examples in a large

corpus calls that feeling into question. Why do our intuitions about block-

ing and synonymy avoidance feel so real, then? It may be that our mental

lexicons allow us easiest and quickest access to the words we are exposed

to the most frequently, so that alternative forms sound odd to us. Only a

combination of further study of the organization of the mental lexicon

and further analysis of data from corpora are likely to shed new light on

the subject.

212 INTRODUCING MORPHOLOGY

5. In the literature on morphology, the term Level Ordering Hypothesis has frequently been used, for reasons

we do not need to go into here.

10.5 Constraints on affix ordering

At various points in this book we have talked about how affixes are

ordered with respect to each other. For example, in Chapter 6 we noted

that inflectional affixes generally come outside of derivational affixes in

the languages of the world. We also mentioned in Chapter 7 (Bybee 1985 )

that among inflectional affixes, tense and aspect inflection tends to come

closer to the stem than person and number inflection, the reason being

that tense and aspect are more closely relevant to the meaning of the verb

than person and number.

In this section, we will look more closely at the issue of how English

derivational affixes are ordered with respect to each other, as this has

been a matter of theoretical dispute for some time. The problem is this: if

the only thing that constrained the ordering of affixes in English were the

categorial restrictions on their attachment (e.g., that -ness only attaches to

adjectives, or that -ize attaches to both nouns and adjectives), we would

expect to find many more combinations of affixes than we do find. In fact,

we find very few of the potential combinations of affixes. The theoretical

issue, then, is what restricts the combination of affixes, and how we

explain those restrictions.

We have already touched upon this problem. In Chapter 9 we noted that

native derivational affixes in English tend to come outside of non-native

ones. One explanation that has been proposed for this generalization is

that the two strata of English derivational morphology are ordered with

respect to each other such that non-native affixation precedes native

affixation. If the two strata are strictly ordered with respect to each other,

then non-native affixes will never be able to attach outside of native ones.

We might call this the Stratal Ordering Hypothesis . 5

There are two problems with this hypothesis, however. One is that it’s

not quite accurate. There are non-native affixes that do not cause stress or

phonological changes, like other non-native affixes, and that are perfectly

happy attaching to native bases, for example -ee ( standee ), - ize ( winterize ),

-able ( singable ). Occasionally it is even possible to attach a non-native affix

to a word formed with a native suffix; the words softenable or whitenable ,

for example, have native -en followed by non-native - able . This has led theo-

rists to lump -ee, -ize , and -able in with native affixes, thus blurring the

lines between the strata.

More seriously, if the only thing which constrained the ordering of deri-

vational affixes in English were the ordering of the two strata, we would

still expect to find many more combinations of affixes than we do. For

example, the suffixes -age and -ize are both non-native. The suffix -age forms

nouns, and - ize attaches to nouns, so -ize should attach to -age words. But we

never find words with the combination -ageize , and words we might coin

on the spot sound quite odd ( orphanageize?, baggageize? ). Similarly, - ify forms

verbs, and non-native -ance forms nouns from verbs, but we never get nouns

Theoretical challenges 213

6. Note that the term base in this context is used in a broader sense than I have used it in this book. For Plag

and Giegerich the base of a word is whatever simple or complex form an affi x attaches to.

like purifiance . So another problem for the theory of stratal ordering is that

the combinations of affixes within strata are more limited than the Stratal

Ordering Hypothesis would lead us to expect.

How else might we constrain the ordering of affixes? One possibility

that has been proposed (Plag 1999 ; Giegerich 1999 ) is that affixes cannot

only select what they attach to (native or non-native bases of particular

categories), but also what attaches to them. For example, according to

this hypothesis, the reason that we don’t find words like purifiance is

that the suffix -ify selects the suffix - ation as its nominalizer. So we find

purification (with an extender - c -), and we predict that any new verb in -ify

that we create (say, Bushify ), will allow -ation to attach to form its nomi-

nalization (therefore Bushification ). Similarly, any verb formed with the

prefix en- in English (e.g., entomb ), will form its nominalization with

-ment , because en- selects -ment as its nominalizer (so entombment , rather

than entombal or entombation ). This sort of selection is called base-driven

selection . 6

Another proposal is called Complexity Based Ordering . According to

Hay and Plag ( 2004 : 571), the gist of this proposal is that “the less phono-

logically segmentable, the less transparent, the less frequent, and the less

productive an affix is, the more resistant it will be to attaching to already

affixed words.” For example, the suffix -ness is extremely productive, its

meaning is always transparent, and it’s easily segmentable from its bases.

According to this hypothesis, it should not be resistant at all to attaching

to other affixes, and this is indeed what we find, as the examples in (21a)

show. By contrast, the verb-forming suffix -en (as in shorten, deepen ), is not

terribly productive or frequent, and because it is vowel-initial, is less eas-

ily segmentable from its base than is -ness , which is consonant-initial. As

the hypothetical examples in (21b) show, it is difficult to find any suffixes

that -en can attach to:

(21) a. courtliness, amateurishness, aimlessness, carefulness

b. *hopefulen, *happinessen, *shoelessen

Indeed the only affix that -en can attach to is -th ( lengthen, strengthen ),

which is completely unproductive in English, and even less segmentable

from its bases than -en itself is.

What we can see is that there are a number of hypotheses that partially

explain how derivational affixes are ordered with respect to each other in

English, but that this question is by no means settled. Theorists will con-

tinue to work on this problem for some time to come.

10.6 Bracketing paradoxes

Bracketing paradoxes are cases in which either the semantic interpreta-

tion or the phonological organization of a word seems to conflict with its

internal structure. Consider the words in (22) :

214 INTRODUCING MORPHOLOGY

(22) ungrammaticality

blue-eyed

unhappier

At first glance, these are unremarkable words. But if we look more closely

at them, you’ll see that they raise some problems. The word ungrammati-

cality needs to have the structure in (23) :

(23) N

un grammatical ity

Since the prefix un- attaches to adjectives and does not change category, it

must attach first to the base grammatical . 7 The suffix -ity attaches to adjec-

tives and forms nouns. So the structure in (23) would be justified accord-

ing to the structural requirements of the affixes. But un- is a native prefix,

and -ity a non-native suffix. The structure in (21) therefore requires that a

native prefix go inside a non-native suffix, contrary to the generalization

we saw in section 10.5 . In other words, if there really is some sort of con-

straint against non-native suffixes appearing outside native suffixes, the

word ungrammaticality is paradoxical.

Of course, this example may not be such a problem after all, since – as

we saw above – there are other cases in which a non-native affix appears

outside of a native one. The word blue-eyed , however, is paradoxical in a

way that cannot be attributed to stratal ordering. It appears to be a com-

pound of the adjectives blue and eyed , the second of which is itself a com-

plex word consisting of the noun eye and an adjective-forming suffix -ed :

(24) A

A A

blue eye ed

But the structure in (24) implies that there is an independent adjective

eyed and that seems not to be the case. Besides, the word blue-eyed seems to

mean ‘having blue eyes’, which would suggest the structure in (25) , rather

than the one in (24) :

(25) [[blue eye] ed]

7. This is of course itself a complex word, but as its internal structure is not relevant to the issue we're discuss-

ing here, we will ignore that internal structure.

Theoretical challenges 215

How do we explain this paradox? In fact, a number of solutions to this

paradox have been suggested. The most plausible is that the structure in

(24) is in fact correct, and that there is a pragmatic reason why we don’t

find an independent word eyed . We don’t find such a word because it’s

usually not a useful concept. People assume that living organisms have

eyes, so we’d never have a reason to point out the ‘eyed one’ as opposed

to the one without eyes. But given a context in which such a contrast is

plausible – say, comparing two space aliens – it no longer seems so absurd

to think of an independent word eyed .

Our final example of a bracketing paradox is the word unhappier , which

is paradoxical for yet a different reason. Here, the semantic interpretation

of the word would suggest the structure in (26) :

(26) A

un happy er

The word unhappier seems to mean ‘more unhappy’, with the comparative

suffix -er applying to the negated adjective, so the semantic interpretation

of the word corresponds to this structure. However, as we saw in Chapter

8 , the comparative suffix -er has a phonological restriction on its attach-

ment that calls the structure in (26) into question. Recall that the com-

parative suffix attaches to one-syllable adjectives, and to two-syllable adjec-

tives whose second syllable is unstressed. Two-syllable adjectives whose

stress falls on the second syllable (e.g., aghast or upset ) cannot take the

comaparative -er suffix, but generally prefer the periphrastic comparative

( more aghast, more upset ). And three-syllable adjectives almost never form

their comparatives with -er . The problem with the structure in (26) , then, is

that -er looks like it has attached to the complex adjective unhappy , which

consists of three syllables. Of course, if the -er were first attached to happy

and then un- attached outside that, as in (27) , there would be no problem:

(27) [un [[happy]er]]

But this does not accurately reflect the meaning of the word. The struc-

ture in (28) suggests that the word means ‘not happier’ rather than ‘more

unhappy’.

Here too, there is a potential solution to the paradox. Many morphologists

believe that words have two separate structures, one which reflects the syntax

and semantics of the word, and a separate structure which reflects the pro-

sodic organization of the word into syllables, feet, and higher levels of phono-

logical organization. We will not go into the details of different levels of

phonological organization here, but I can give you just a suggestion of what

I mean. Suppose that the word unhappier has two simultaneous structures:

216 INTRODUCING MORPHOLOGY

(28) A

un happy er

σs σwσs σ

The top structure is identical to the one in (27) and reflects the seman-

tic organization of the word. The one on the bottom reflects the phono-

logical organization of the word, where σ stands for ‘syllable’, σ s for ‘stressed syllable’, and σ w for ‘unstressed syllable’. If words are allowed to have two separate and simultaneous representations, one for their syntac-

tic and semantic structure, and another for their phonological structure,

the word unhappier is no longer paradoxical.

10.7 The nature of affixal polysemy

Unlike the problems of affix order and bracketing paradoxes, which

have received a great deal of attention from morphologists, the prob-

lem we will take on in this section has received little attention, but it is

no less interesting. In Chapter 3 we brief ly touched upon the issue of

affixal polysemy. To refresh your memory, affixal polysemy is the ten-

dency for affixes to have several closely related meanings. For example,

we pointed out in Chapter 3 that it is a curious fact that the affix that

is used for making agent nouns in languages is frequently also used for

making instrument nouns. As I pointed out, this is the case in English ,

and Dutch – not surprising, as they are closely related languages – but

also in Yoruba , a Niger-Congo language, Turkish (Lewis 1967 : 225),

Kannada (Sridhar 1990 : 273), and many other languages. It cannot be an

accident that agent nouns and instrument nouns are so often created

by the same affixes. So the theoretical question that arises is why this

should be. To answer this question, let’s again look a bit more closely at

English.

(29) - er agent writer, driver, thinker, walker

instrument opener, printer, pager

experiencer hearer

patient/theme fryer, sinker

Theoretical challenges 217

- ant/-ent agent accountant, claimant, servant

instrument adulterant, irritant

experiencer discernant

patient/theme descendant

It appears that in English the suffix -er forms not only agent and instru-

ment nouns, but also nouns that denote the experiencer of an action, or

even the patient or theme of an action. Even more curious, the suffix -ant

covers the same range of meanings. The theoretical question we must

raise in light of these data is why - er and -ant nouns cover just this range

of meanings and not some others.

The first step in explaining this affixal polysemy is to figure out just

what -er and -ant mean. One suggestion that has been made (Lieber 2004 )

is that these affixes don’t actually mean ‘agent’ or ‘instrument’, but some-

thing much more abstract – something like ‘concrete noun concerned

with a process or event’. In this, they are semantically analogous to simple

nouns like poet or awl that denote people or things defined by what they

do. One piece of evidence for this claim is that -er can attach to nouns as

well as to verbs, and when it does, it always adds an active or eventive ele-

ment of meaning to its noun base. So, for example, a villager is someone

who lives in a village, and a freighter is something that carries freight. No

verb is necessary for the active part of the meaning – this comes directly

from the suffix.

The second part of the answer to our question has to do with under-

standing the argument structures of verbs. You’ll recall from Chapter 8

that the argument structure of a verb consists of those arguments that are

semantically necessary to the verb (see section 8.2 ). What is most interest-

ing for our purposes is that in English there is a range of semantic roles

that the subject of a verb can play:

(30) a. Fenster ate the pizza.

b. The key opened the door.

c. We heard the neighbors fighting.

d. The boat sank rapidly.

In (30a) , the subject of the sentence is the agent or ‘doer’ of the action.

In (30b) , the subject is called an instrument rather than an agent, because

it is an inanimate noun that does something. In (30c) , we call the seman-

tic role that the subject plays the experiencer rather than agent, because

one can hear something without doing anything at all – to be an agent,

one must act intentionally. Finally, the subject in (30d) is not the agent,

but the theme , in other words, the noun that undergoes or is moved by

the action.

What is interesting is that -er and -ant nouns denote exactly the seman-

tic role conveyed by the subject of their base verb. So an eater is an agent,

just as the subject of the verb eat is an agent, an opener is an instrument,

just as the subject of open is an instrument, a hearer is an experiencer, just

as the subject of hear is an experiencer, and a sinker is a theme, just as the

subject of sink (intransitive) is a theme. The reason that -er and -ant display

(agent)

(instrument)

(experiencer)

(theme)

218 INTRODUCING MORPHOLOGY

exactly the range of meanings that they do is that they are linked to the

subject argument of the verb. They can mean whatever the subject of a

verb can mean.

10.8 Reprise: what’s theory?

The few topics I’ve touched upon here just barely scratch the surface of

theoretical questions that linguists can raise about morphology. Most of

the issues we’ve looked at in this chapter concern English, although if

we had time to delve further into them, we’d see that they concern many

other languages as well. What is important to keep in mind, though, is

that there are many theoretical issues that come up only when we look

beyond English to the analysis of other languages. You have also barely

gotten the chance to see how one argues for or against theoretical pro-

posals. In this book, you have learned to find, analyze, and think about

morphological data. You are now ready to embark on further theoretical

challenges, both looking at issues raised by word formation in the lan-

guages of the world, and learning how to make theoretical proposals

and support them. We leave these challenges to your next course in

morphology.

Summary In this chapter we have first considered what we mean by morphological theory, and then explored a number of theoretical topics that have been

important to generative morphologists over the years. We have looked at

the nature of morphological rules and the predictions specific models

of morphology make about the sorts of morphology we ought to find in

the languages of the world. We have explored the issue of the relation

between morphology and syntax and the extent to which these two

levels of linguistic organization can be kept separate from one another.

We have looked at how the patterns of ordering in derivational affixes

can be explained, how bracketing paradoxes can be resolved and how

affixal polysemy can be explained.

Exercises 1. Consider the sort of templatic morphology that we looked at in section

5.6 . Do you think that templatic morphology presents any problems for Item and Arrangement theories of morphology?

2. Consider the verb shit . What are the past tense and past participle forms of this verb. Check a dictionary if you're not sure. Does this verb give us any evidence for or against blocking?

3. Consider the words pomp, pompous, and pomposity. Do they offer evidence for or against blocking?

4. Why might the following words be considered bracketing paradoxes? three-wheeler

Theoretical challenges 219

whitewashed transformational grammarian nuclear physicist

5. In section 10.7 I suggested that the suffix -er can have whatever semantic role is carried by the subject of its base verb. Consider the words loaner and keeper in the sentences below:

i. My car was in the garage so they gave me a loaner. ii. This book is a keeper.

What challenge do these forms present for the hypothesis in 10.7?

6. The suffix -ee is usually said to form ‘patient nouns’, that is, nouns that denote the person who undergoes or is subject to the action denoted by the base verb. Consider the following examples, and discuss the extent to which -ee exhibits affixal polysemy:

employee nominee standee escapee addressee amputee

7. Consider the following prefixed words (from Bauer, Lieber, and Plag 2013 : 600, data from COCA). Each word has two prefixes. First use the online OED or another dictionary to identify which prefixes are native and which are non-native. Then discuss what problems, if any, these examples pose for the Stratal Ordering Hpothesis.

ex-betrothed hyper-unemployment mini-outbreak polyunsaturated post-midnight

Glossary ablative: The case typically assigned to objects of prepositions denoting

instruments or sources.

ablaut: Internal vowel change. Also known as apophony .

absolutive: In an ergative-absolutive case system, the case that is assigned to the

subject of an intransitive clause and the object of a transitive clause.

accusative: In a nominative–accusative case system, the case assigned to the direct

object of the clause, and in some languages to objects of prepositions.

acronym: A word made up of the initial letter or letters of a phrase and

pronounced as a word. For example, from self-contained underwater

breathing apparatus we get the acronym scuba , pronounced [skubə].

active: A voice in which the subject of the clause is (typically) the agent,

instrument, or experiencer and the direct object the theme or patient.

In English an active clause would be Fenster ate the pizza , as opposed to

a passive The pizza was eaten .

adjuncts: Non-argumental phrases that are not necessary to the meaning of a

verb.

affix: A bound morpheme that consists of one or more segments that

typically appear before, after, or within a base morpheme.

affixal polysemy: Multiple related meanings of an affix.

affixation: Formation of words by the addition of prefixes, suffixes, infixes, and

circumfixes.

agent: The argument of the verb that performs or does the action. Agents

typically are sentient and have intentional or volitional control of

actions.

agglutinative: One of the four traditional classifications of morphological systems.

Agglutinative systems are characterized by sequences of affixes each of

which is easily segmentable from the base and associated with a single

meaning or grammatical function.

agrammatism: A form of aphasia in which comprehension is good, production is

labored, and grammatical or function words largely absent. Also

known as non-fluent aphasia.

agreement: Contextual inflection of elements of a phrase or sentence to match

another element of that phrase or sentence. For example, in the

Romance languages the inflection of adjectives in a noun phrase must

match the gender and number of the head noun. In Latin the verb

must be inf lected to match the person and number of its subject.

allomorph: A phonologically distinct variant of a morpheme.

analytic: One of the traditional four classifications of morphological systems. In

analytic systems words consist of only one morpheme. Also known as

isolating .

anti-passive: Morphology that decreases the valency of verbs by eliminating the

object argument.

222 GLOSSARY

aphasia: A language disorder that results from injury to the brain.

apophony: Internal vowel change. Also known as ablaut .

applicative: Morphology that increases the valency of a verb by adding an object

argument.

argument: A noun phrase that is semantically and often syntactically necessary

to the meaning of a verb. The arguments of a verb consist of its subject

and complement(s).

aspect: A type of inflection that conveys information about the internal

composition of an event.

assimilation: A phonological process in which segments come to be more like each

other in some phonological feature such as voicing or nasality.

attenuative affixes: Affixes that denote ‘sort of X’ or ‘a little X’.

attributive compound: A compound in which the two elements bear a modifier–modified

relationship to one another.

augmentative: A kind of expressive morphology which conveys notions of larger size

and sometimes pejorative tone.

backformation: A morphological process in which a word is formed by subtracting a

piece, usually an affix, from a word which is or appears to be complex.

In English, for example, the verb peddle was created by back formation

from peddler (originally spelled peddlar ).

base-driven selection: Choice of an affix by its base, whether a simple or complex word. For

example, in English, words prefixed by en- always form nouns by

suffixation of -ment . The complex base enX therefore selects its affix.

binyan: A templatic pattern associated with a specific meaning or function.

blend: A type of word formation in which parts of words that are not

themselves morphemes are combined to form a new word. For

example, the word smog is a blend of smoke and fog.

blocking: The tendency of an already existent word to preclude the derivation of

another word that would have the same meaning. For example, the

existence of the word glory precludes the derivation of gloriosity and

the existence of went precludes the formation of the regular past tense

goed.

bound base: A morpheme which is not an affix but which nevertheless cannot

stand on its own. In English, bound bases are items like endo , derm ,

and ology , from which neo-classical compounds like endoderm and

dermatology are formed.

bracketing paradoxes: Complex words in which there is a mismatch between syntactic

structure and phonological form or between syntactic structure and

semantic interpretation. Within theories that admit stratal ordering,

bracketing paradoxes can also involve mismatches between the

structure required on the basis of word formation rules and the

structure consistent with stratal ordering.

case: Inflectional marking which signals the function of noun phrases in

sentences.

causative: Valency-changing morphology that adds an external causer to a verb.

circumfix: A morpheme that consists of the simultaneous attachment of a prefix

and a suffix which convey meaning or function only when they

appear together.

Glossary 223

classifier: A word or morpheme used in a noun phrase, often in the context of a

numeral or determiner, that designates the grammatical or semantic

class of the noun.

clipping: A word formed by subtraction of part of a larger word. For example, in

English math is a clipping from mathematics and ad is a clipping from

advertisement.

clitic: Small grammatical elements that cannot occur independently but are

not as closely bound to their hosts as inf lectional affixes are.

closed class: A fixed list from which particular forms can be lost, but to which no

new forms can be added.

coda: The consonants that come after the nucleus in a syllable.

coinage: A word that is made up from whole cloth rather than by affixation,

compounding, conversion, blending, reduplication, or other

processes.

competition: When rival affixes interact such that a derived word can be formed

with one of the set, precluding the others from forming synonymous

words.

completive: An aspectual distinction that focuses on the end of an event.

complex word: A word made up of more than one morpheme.

Complexity Based The hypothesis that suffixes which are more transparent, more

Ordering: productive, and more easily segmented from their bases will occur

outside those that are less transparent, less productive, and less easily

segmented from their bases.

compositional: The semantic interpretation of a word is compositional to the extent

that it can be computed as the sum of the meanings of each of its

morphemes.

compound: A word made up of two or more separate lexemes.

compounding : Formation of words by combining bases.

conjugation: The traditional name for the inflectional paradigm of a verb.

consonant mutation: A form of internal stem change in which consonants of a base differ

systematically in different morphological contexts.

contextual inflection: Inflection which is determined by the syntactic construction in which

a word finds itself.

continuative: An aspectual distinction that focuses on the middle of an event as it

progresses.

conversion: A type of word formation in which the category of a base is changed

with no corresponding change in its form. For example, in English the

verb to chair is formed by conversion from the noun chair . Also called

functional shift .

coordinative A type of compound in which the two elements have equal semantic

compound: weight. Examples in English are producer-director or blue-green .

corpus: A database comprised of spoken language and/or written texts that

can be mined for various forms of linguistic study.

cran morph: A bound morpheme that occurs in only one word. An example in

English is cran in cranberry .

creativity: The conscious use of unproductive word formation processes to form

new words that are often perceived as humorous, annoying, or

otherwise worthy of note.

224 GLOSSARY

cumulative exponence: The association of several inflectional meanings with a single

inflectional morpheme.

dative: In languages which mark case, the case assigned to the indirect object

and frequently to objects of prepositions.

declarative: The mood/modality of ordinary statements (as opposed to questions or

imperatives, for example).

declension: The traditional name for the inflectional paradigm of a noun,

especially in languages that display case marking.

default endings: Inflectional markings that are used when no more specific marking is

applicable.

dependent: Modifiers and complements of the head of a phrase.

dependent-marking: Morphological marking of the dependents of a phrase rather than its

head. For example, in noun phrases marking occurs on determiners

and adjectives rather than the noun.

derivation: Lexeme formation processes that either change syntactic category or

add substantial meaning or both.

deverbal compound: See synthetic compound .

diminutive: Evaluative morphology that expresses smallness, youth, and/or

affection.

ditransitive The valency of a verb that has two complements.

double marking: Morphological marking of both the head of a phrase and its

dependents. For example, in a noun phrase marking would occur on

both the head noun and on adjectives and/or determiners that

modify it.

dual: Number-marking that denotes exactly two objects.

electroencephalography

(EEG): A device for measuring electrical activity on the scalp.

enclitic: A clitic that is positioned after its host.

endocentric: Having a head. In endocentric compounds the compound as a whole is

the same category and semantic type as its head.

epenthesis: A phonological process in which a segment is inserted between two

other segments.

ergative: In an ergative/absolutive case system, the marking of the subject of a

transitive verb.

ergative/absolutive A case-marking system in which the subject of an intransitive verb is

case system: marked with the same case as the object of a transitive verb, and the

subject of a transitive verb receives a different marking.

etymology: The study of the origins and development of words.

evaluative affixes: Affixes, including diminutives and augmentatives, that denote size

and/or negative or positive associations.

evaluative morphology: Morphology that conveys information about size and frequently also

about positive or negative valuation. See evaluative affixes .

evidentiality: An inflectional category in which speakers must specify what the

source of knowledge is.

exclusive: Person-marking in which the hearer is not included.

exocentric: Lacking a head. In exocentric compounds the compound as a whole is

not of the category or semantic type of either of its elements.

Glossary 225

experiencer: The argument of the verb that experiences the action. Like agents,

experiencers are sentient, but unlike agents they are not in volitional

control of the action. They are often the subjects of verbs of perception

or emotion.

exponence: The association of inflectional meanings with inflectional forms.

extender: An extra segment that occasionally appears in derived words in

English, but which has no meaning and does not obviously belong to

either the base or the affix, for example, n in Platonic ( Plato+n+ic ).

eye tracking: An experimental method in which the movements of a subject's eyes

are followed and recorded while they are reading words or sentences.

fast mapping: The ability of language-learners to rapidly create lexical entries for

new words that they hear.

formative: An element that is used to form words but which lacks clear meaning

or function.

free base: A base that can occur as an independent word.

frequency of base type: The number of different bases that are available for an affix to attach

to, thus resulting in new words.

frequentative: Aspectual marking that signals repetition of an action. See also

iterative .

full reduplication: A word formation process in which whole words are repeated to

denote some inf lectional or derivational meaning.

functional magnetic

resonance imaging (fMRI): A device which maps brain activity over the course of time.

functional shift: See conversion.

fusional: One of the four traditional classifications of morphological systems. In

fusional systems words are complex but not easily segmentable into

distinct morphemes. Morphological markings may bear more than

one function or meaning.

Gavagai problem: A philosophical problem concerning how children come to associate

the meaning of a word with the action or entity the word denotes.

gender: Inflectional classes of noun that may be either arbitrary ( grammatical

gender ) or semantically based ( natural gender ). See also noun classes .

genitive: The case assigned to the possessor of a noun.

gloss: A morpheme-by-morpheme translation of a word or sentence.

habilitative: A verb form meaning ‘can V’.

habitual aspect: Aspectual marking that designates that an action is usually or

characteristically done.

hapax legomenon: A word that occurs only once in a corpus.

head: The morpheme that determines the category and semantic type of the

word or phrase.

head-marking: Morphological marking of the head of a phrase rather than its

dependents. For example, in noun phrases marking occurs on the

noun itself, rather than on determiners and adjectives that modify the

noun.

heavy syllable: A syllable which has either a coda or a long vowel or both.

hypocoristic: A nickname.

imperative: The mood/modality used for commands.