22

mark92382

chapter1.pdf

Home >Psychology homework help >22

Introduction 1 CHAPTER

KEY TERMS mental lexicon

process

psycholinguistics

representation

PREVIEW

This chapter provides an overview both of the field of

psycholinguistics and of the book itself. By the end of the

chapter you should have a better understanding of, amongst

other things, the appeal, significance and subject matter of

psycholinguistics.

https://www.cambridge.org/core/terms. https://doi.org/10.1017/CBO9780511978531.002 Downloaded from https://www.cambridge.org/core. University of California, Santa Cruz, on 29 Jul 2020 at 09:59:24, subject to the Cambridge Core terms of use, available at

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

2 INTRODUCING PSYCHOLINGUISTICS

1.1 Introduction

Although there is little agreement on how many words an adult native

speaker of English might know, let us take a conservative estimate of

20,000 (based on Nation, 2006, but see Section 8.2 ), and let us also

assume that splundle is not one of them. Imagine such a native speaker

searching through their 20,000-word mental lexicon, as we call the dic-

tionary in our heads. Let’s assume that they do so at a seemingly impres-

sive rate of 100 words per second (which incidentally is 20 times faster

than a good reading pace of 300 words per minute). At this rate, if they

searched exhaustively through their mental lexicon, it would still take

over 3 minutes to confirm that splundle is not there. Yet if you ask some-

one if they know the word splundle , they will be able to tell you more or

less instantly that they do not. Clearly something is wrong with the

assumptions we have just made about how rapidly we can look up words

in our mental lexicon, or about the way in which we search through it,

or both.

Ask someone to tell you what they had for breakfast, and then ask them

to explain the workings of their coffee machine, and you are likely to find

that their speech is much more hesitant in the second task than in the

first, with more errors and restarts as well as pauses and uhms and ahs . The

differences here reveal something about the nature of planning involved

in different speaking tasks. The locations of the hesitations might also tell

you which kinds of words and/or sentence structures the speaker finds

more difficult to find or to put together.

Have a conversation with someone while they are carrying out a diffi-

cult task like driving a car, and you will find that both their language

production and their language comprehension is less fluent than at other

times (Becic et al ., 2010 ).

If you listen out for speech errors, or slips of the tongue, you are much

more likely to hear an error where the beginning sounds of two words

have been swapped over, as in ‘tip of the slongue’, than one where the

beginning of one word and the end of another have been exchanged, as in

‘slit of the pongue’.

When you see a sign like that in Figure 1.1 , or one that says ‘Please go

slowly round the bend’, you might well chuckle, but your experience of

misreading the sign shows that there are certain preferred patterns of

analysing the structure and meaning of sentences in English.

Consider what is wrong with the following interaction between speak-

ers A and B, in which <SILENCE> indicates noticeably long periods of

silence, and CAPITALS indicates the part of the word that receives word

stress:

A. Could you tell me the time of the next train to Palmerston North and

how much a single fare costs?

B. <SILENCE>

A . Hello?

B. <SILENCE> Invisible tables sleeeeep violently

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

Introduction 3

A. When does the next train go to Palmerston North and what is a sin-

gle ticket?

B. <SILENCE> PalMERston North trains from platform 3 leave. A single

ticket licences a one-way journey.

There are of course a number of odd features here – there is meaningless and

irrelevant content, there are long silences with odd patterns of pausing,

there are infelicities of vocabulary and structure (‘A single ticket licences a

one-way journey’, ‘Palmerston North trains from platform 3 leave’), there is

strange stress placement (‘PalMERston’) and another odd pronunciation

(‘sleeeeep’). Looking at ‘odd’ speech like this (the example is fabricated but

the principle remains valid) sheds some light on what we need to do in order

to speak. That is, speakers usually aim to produce utterances that have appro-

priate meaningful content, that use appropriate lexical items and gram-

matical structures, and that have appropriate pronunciation, intonation

and phrasing. Speakers aim to do this fluently and in real time, and rarely

have opportunity to rehearse. They also have to relate what they say to the

context, including to previous speech in a conversation.

Speakers are generally very good at doing this. For example, one experi-

ment was explicitly designed to elicit subject-verb agreement errors like

(1.1), where the subject is plural ( efforts ) but is followed by a singular verb

( is ). The reason for the error is that the most recent noun before the verb

is singular ( language ). In the experiment, fewer than 5% of the stimuli

designed to produce agreement errors like this actually did so (Bock &

Miller, 1991 ).

(1.1) Efforts to make English the official language is gaining strength

thro ughout the US.

In spontaneous speech, it has been found that sound errors (e.g. saying

‘par cark’ for ‘car park’) occur only about 1.5 times per 10,000 words, and

word errors (where the wrong word is chosen or words change positions

in a sentence) occur only about 2.5 times per 10,000 words (Deese, 1984 ).

Clearly we are generally pretty good at what we do when we produce and

understand language.

Figure 1.1 Thank goodness our young ones are not

breaking the speed limit.

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

4 INTRODUCING PSYCHOLINGUISTICS

1.2 What is psycholinguistics?

The rather disparate observations in the preceding section illustrate just

a few of the areas of interest in psycholinguistics. Psycholinguistics can be

defined as the study of the mental representations and processes

involved in language use, including the production, comprehension and

storage of spoken and written language. A number of issues arise from

this definition. Some are to do with representations , such as:

• How are words stored in the mental lexicon, i.e. the dictionary in our heads? Is the mental lexicon like a dictionary, or more like a thesau-

rus? For instance, is cat listed near the similar sounding word catch or

near the meaning-related word dog ? Or neither? Or both?

• Do we have phoneme-sized chunks of language in our heads? That is, as part of recognising the word cat do we also recognise the compo-

nent sounds /k/, /æ/ and /t/?

• Do literate people have letter-sized chunks filling equivalent roles for the processing of written language?

• How is the meaning of a sentence represented in our memory? • Is government a single word or govern + ment ? • Is the plural form cats represented in the lexicon, or just the singular

cat ?

Other questions concern the processes that might operate on those

representations:

• How do we recognise words so effortlessly? • Do we analyse the speech signal phoneme-by-phoneme or do we iden-

tify complete syllables or even larger units?

• Do we recognise government as a complete form or do we have to con- struct it from govern + ment ?

• If cats is not represented in the lexicon, does that mean that we use a rule to get the plural form of cat , and how does this work for irregular

plurals like children ?

• When we speak, how do we convert an idea into an utterance? • As listeners, how do we get from hearing an utterance to developing

our own representation of the idea(s) being expressed by that utter-

ance?

• What stages do we have to go through during the construction of utterances? For example, do we first generate a sentence structure

and only then populate it with words from our mental lexicon, or do

we first choose words and then build a structure around those words?

• Do the processes involved in language production and comprehension influence one another, and if so in what ways?

The outline sketch of language use in Figure 1.2 gives an overview of areas

of interest in psycholinguistics as well as providing the basic structure of

this book, which has chapters on language production (roughly following

the progression shown in the second column) and on comprehension

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

Introduction 5

(fourth column). Psycholinguistics clearly has links to other areas of linguis-

tic study, and some of these are shown in the final column.

From the language producer’s (speaker’s, writer’s) perspective, the pro-

duction of a message takes us from an underlying intention, through

stages of planning sentence structures and selecting words, to the articula-

tion of that intention as a sequence of sounds or letters, as shown by the

arrow.

From the comprehender’s (listener’s, reader’s) viewpoint, the goal is to

perceive or recognise elements such as letters and sounds in the input, to

recognise words and to work out the connections between these words in

sentence structures, in order to arrive at a message-level interpretation.

The arrow in the ‘comprehension’ column shows such a ‘bottom-up’ flow

of information from the input to an interpretation. This is a simplifica-

tion, though, as there is evidence for ‘top-down’ information flow too, e.g.

when a listener starts to gain an understanding of the sentence they are

hearing this can influence the efficiency with which they recognise sub-

sequent words in the sentence. Most psycholinguists today support the

idea of interactive processing in both production and comprehension,

with information flowing in both directions (bottom-up and top-down) as

well as between elements at the same level (so recognising one word has

an effect on the likelihood of recognising similar words).

It is reasonable to claim that the main focal areas of psycholinguistics

have tended to be sentences and words . So production studies have focused

on the generation of sentence structure and on syntactic planning, as well

as on word finding and word building. Similarly, much of the study of

comprehension has dealt with word recognition and sentence parsing

(working out the syntactic structure of sentences). The study of these pro-

cesses has in turn involved consideration of the representation of words

and of grammar.

In effect there are many more ‘fields’ than this, dealing with many

subprocesses, such as the perception of letters and of speech sounds, the

processing of different kinds of word endings, and so on. In addition,

these fields are not always clearly distinct, neither in terms of research

nor in terms of the ways in which the processor operates. For instance

does our choice of words influence our choice of sentence structure, does

sentence-level interpretation influence phonetic perception, does the

spoken form of a word affect our silent reading of it, and so on?

Bottom-up:

processing based

on information

flow from lower

levels of

processing to

higher levels, e.g.

from the sensory

input (the speech

signal) to the

lexicon (words).

Top-down:

processing guided

by information

flow from higher

levels (e.g.

sentence

interpretation) to

lower levels (e.g.

words).

Interactive:

processing that

involves a

combination of

bottom-up and

top-down flow of

information, as

well as links

between elements

at the same level.

Processes Production Comprehension Areas of linguistics

discourse analysis

central intention message (‘idea’) interpretation

syntax/semantics

planning sentence structure parsing

morphology/syntax

lexicalisation words word recognition

phonetics/phonology

peripheral articulation sounds/letters perception

Figure 1.2 An outline sketch of

language use.

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

6 INTRODUCING PSYCHOLINGUISTICS

Although the discussion above has tried to be inclusive of both spoken

and written language, the content of this book shows a bias towards the

production and comprehension of spoken language. To an extent this is a

personal bias, since the author’s own research interests are in this area,

but it also reflects the primacy of the spoken language in the develop-

ment both of the individual and of the species, as well as a general bias in

the research literature.

1.3 Who does psycholinguistics?

The term psycholinguistics appears to have first been used to refer to the

psychology of language in the 1920s. However, the birth of psycholinguis-

tics as a discipline is often linked to a seminar at Cornell University in

1951. The interdisciplinary nature of the field is reflected in the fact that

this seminar was reported in both a psychological journal (Osgood &

Sebeok, 1954a ) and a linguistics one (Osgood & Sebeok, 1954b ).

Psycholinguistics continues to be taught in both of these disciplinary

areas, and is influenced by research and teaching practices and method-

ologies in both. Courses and research in psycholinguistics are also found

in multidisciplinary cognitive science units throughout the world, and

the issues studied in this area are also of interest to computer scientists

(particularly those interested in Artificial Intelligence and Human–

Computer Interaction), philosophers (especially in areas concerning the

development and representation of meaning) and others.

Of course, our reasons for studying the psychology of language may dif-

fer, according to our starting point, and this will be reflected in the

nature of courses and of research agendas. Thus a characterisation of the

psychological perspective might be that studying language use can con-

tribute to our knowledge of the workings of the mind. Neuropsychologists

might especially be interested in locating the language faculties within

the physical structures of the brain. A more linguistic perspective may be

that studying language use can inform our theories of language structure,

it can provide the performance data to support theories of competence, it

can provide psychological validity for linguistic constructs, and so on.

1.4 How do psycholinguists do psycholinguistics?

Although some of our knowledge in this field comes from introspection

and the observation of daily behaviour, most of the major insights have

come through the generation and testing of hypotheses through experi-

ment. More recently, there has been a marked increase in high-tech obser-

vation, measuring brain activity while participants are engaged in

language-related tasks. Because of this combination of types of evidence,

psycholinguistics tends to blend the theoretical and descriptive insights

of linguistics with the experimental methodology and rigour of

psychology.

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

Introduction 7

Observation Sophisticated experimental procedures and equipment have only become

available relatively recently, and so it is no surprise that early discoveries

in psycholinguistics were based on more observational approaches. Early

approaches also focused in particular on speech production, since the

spoken output is most easily observed. For instance, we can note and ana-

lyse the occurrence of pauses and hesitations in speech, and assess these

as indicators of the planning and sequencing of the spoken output. We

can observe where speakers make errors, and relate these errors to hypoth-

eses about the speech planning and production processes. We can also

look at the correction of errors as an indication that speakers are monitor-

ing their own output. Rather less often, we can note when speakers are in

a ‘tip-of-the-tongue’ state and find out from them what they can remem-

ber of an elusive word, and see what this might tell us about the processes

involved in finding words.

Some of these observational techniques can be combined with experi-

mental intervention. We will see in later chapters, for example, that

researchers have run experiments designed to elicit speech errors or to

induce the tip-of-the-tongue state. A range of additional tasks has been

used to study both production and comprehension.

Experiment Experimental methods have become highly sophisticated over the past

half century, in particular with the ready availability of software that

allows a high level of control over the presentation of stimuli and collec-

tion of data, with precise timing. Advantages of experimental over obser-

vational approaches include the reduction of observer bias and the

increased control over what participants are required to do. This last point

means that researchers are more easily able to isolate aspects of the pro-

duction or comprehension processes that they are interested in. The dis-

advantage of experimental approaches, however, is the relative lack of

ecological validity – it is usually only in experimental settings that par-

ticipants are asked to make an explicit judgement about whether or not a

stimulus they have heard is a word of their language, or makes sense in

their language.

Typically, experimental techniques involve some kind of response time

measure, where the time taken by participants to carry out some lan-

guage-related task is recorded, often together with their accuracy in that

task. Examples of some of these tasks can be found on the website for this

book. Some response time tasks involve the notion of competing atten-

tion. That is, it is assumed that the time taken by the participant to com-

plete the task depends on how much spare capacity they have, which will

in turn depend on the nature of the linguistic processing going on at the

same time. So for instance in the click location task participants listen to

an utterance that has had a non-linguistic click or beep sound added to it.

Their task is to listen to and understand the utterance and at the same

time to mark on a transcript of the utterance where they think the click

occurred. Results from such tasks show ‘migration’ of the click to

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

8 INTRODUCING PSYCHOLINGUISTICS

locations other than where it actually occurred. This migration is at least

in part a consequence of the fact that participants’ attention was devoted

to linguistic processing. Similar effects are found in dichotic switch

monitoring. In this task participants have to indicate when the speech

they are listening to moves from one headphone channel to the other.

They do not always do this accurately because of the linguistic processing

load at the point of the switch. (See Chapters 10 and 11 for examples of

such tasks and how they contribute to our understanding of sentence

comprehension.)

Other response time tasks use reaction times as a measure of the process-

ing speed for the item being attended to. In its simplest form, perhaps, the

response time task requires participants to press a button as quickly as pos-

sible when they see or hear a word. There may be a choice of buttons, one

for a positive response when the word is an actual word of the language and

the other for a negative response, when the word is not an actual word.

Chapters 8 and 9 discuss how the speed and accuracy of participants’

responses depend on factors such as familiarity – well-known words result

in faster and more accurate responses than less well-known words.

In another button-pressing task, a participant reads a sentence on-

screen, with presentation of each successive word controlled by the par-

ticipant’s pressing of the button. Chapters 10 and 11 give illustrations of

how difficulty in interpreting the sentence at any point is reflected in the

time taken to press the button to see the next word.

Carrying out a response task involves many components, some of which

are assumed to be relatively invariant across different stimuli (e.g. the

time it takes to send a motor command from the brain to the hand to

execute a button press). One study (Indefrey & Levelt, 2004 ) broke down

the supposedly simple task of saying a word that names a pictured object

into the components listed in Table 1.1 , starting with a mapping from the

picture onto a mental concept and finishing with the initiation of the

spoken word. Details will be given in the following chapters, but note for

now that the most obvious influence on the time taken to initiate the

response is the length of the word, since the stage labelled syllabification

in the table has a duration that depends on the number of phonemes, or

speech sounds, in the word. But in addition, it is important to remember

that the other stages will have variable durations, depending for example

on the accessibility of the word, which will vary with factors such as how

often or how recently that word has been used. These are the factors that

Table 1.1 Approximate timing of components involved in picture naming

(based on Indefrey & Levelt, 2004 ).

Conceptual preparation 175 msec

Word retrieval 75 msec

Phonological code retrieval 80 msec

Syllabification 25 msec per phoneme

Phonetic encoding 145 msec

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

Introduction 9

are often explored in experiments, and which require other aspects of the

experiment to be held constant.

Language and the brain It is becoming more and more usual to find studies which include moni-

toring of the patterns of blood flow and/or electrical activity associated

with neural stimulation in the brain while a participant is carrying out

some language production or comprehension task. These measures have a

clear advantage over behavioural methods such as response button press-

ing, since they provide a more direct measure of psycholinguistic process-

ing. Imaging techniques that measure blood flow give an indication of the

areas of the brain involved in the tasks being carried out (and of brain

areas that have been damaged after strokes or accidents). However, they

typically have relatively poor temporal resolution, and so are not ideally

suited to tracking the time course of the very fast processes involved in

language processing. On the other hand, techniques that measure electri-

cal activity provide better temporal resolution but poor spatial resolution.

They are therefore better suited to the measurement of the time course of

processing than to the precise determination of which brain areas are

responsible for various tasks.

As an example of what we can learn from neurophysiological tech-

niques, Figure 1.3 gives a very approximate depiction of differences in

brain activity levels measured in a range of tasks carried out by German

participants (Heim, Opitz & Friederici, 2002 ). Brain activity was measured

using functional Magnetic Resonance Imaging (fMRI), which tracks

changes in blood flow (see sidebar). An important aspect of German that

is being examined here is the fact that nouns have grammatical gender

which is marked through the forms of articles (e.g. the word for the ). For

instance the word for cat has feminine gender, so the cat is ‘die Katze’,

while the word for dog has masculine gender and the dog is ‘der Hund’.

The shaded shapes in each panel show differences in brain activity

when two tasks are compared. The first panel shows areas with an increase

in activity in a simple baseline speaking task ( BASE ), compared to when the

participants are at rest ( NULL ). The baseline task is simply to say ‘ja ja’ (‘yes

yes’) when they see a simple visual stimulus. The second panel shows areas

that are active in a picture naming task ( NAME ), compared with the

fMRI –

functional

Magnetic

Resonance

Imaging

Neurons that

are active

consume more

oxygen from their

blood supply than

those that are

inactive.

Oxygenated and

deoxygenated

blood have

different magnetic

properties, and

though these

differences are

small they are

sufficient to show

up in a scanner

that generates

high magnetic

fields. The

participant lies in

a scanner and

changes in neural

activity in

different brain

areas are

measured while

they perform

some language

task.

BASE-NULL NAME-BASE GENDER-NAME

Figure 1.3 Schematic representation of differences in brain activity during different tasks (based on fMRI data published by Heim et al., 2002 ; see text for explanation).

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

10 INTRODUCING PSYCHOLINGUISTICS

baseline task. In a naming task the participants are not just speaking but

are also retrieving a word from memory. For instance, if they see a picture

of a cat, then they have to retrieve and produce the word that names that

picture, i.e. ‘Katze’ ( cat ). Clearly additional areas of the brain are involved

in this task when compared with the baseline task. Finally, the third panel

shows a small additional area (when compared with the naming task) that

is active when the participant produces the appropriate definite article.

As this is marked for the grammatical gender of the noun in German, the

correct form has to be selected, and this seems to involve this small area

of the brain (hence the label GENDER in the figure).

fMRI, as used to obtain the data above, is one of the imaging techniques

that measures blood flow in the brain. Other types include PET scans

(positron emission tomography), which measure emissions from radioac-

tive substances injected into the bloodstream, and CAT scans (computed

axial tomography), which are based on a series of X-rays taken from differ-

ent angles.

Imaging techniques measuring electrical activity in the brain include

MEG (magnetoencephalography) and EEG (electroencephalography).

The former measures the magnetic fields created by naturally occurring

electrical activity in the brain, while the latter measures electrical activ-

ity on the scalp. An advantage of the MEG technique is that the magnetic

fields it measures are not distorted by surrounding tissues, which can

affect the electrical activity measured by EEG. These techniques measure

event-related potentials (ERP), the electrical currents passing through the

fluid that surrounds neurons as they respond to an event, or event-related

magnetic fields (ERF) which measure currents over larger neuronal

structures.

Despite the technological advances that now allow us to look in great

detail at neurophysiological aspects of language production and compre-

hension, there are areas of psycholinguistics in which informal observation

and relatively simple experimental techniques still provide appropriate

means for learning more about the production, comprehension and repre-

sentation of language. A clear advantage of such techniques for many

researchers, especially those housed in linguistics departments in humani-

ties faculties, is their low cost compared with the hugely expensive imaging

techniques mentioned above. Where possible, such approaches have been

used in the exercise material included in each chapter of this book.

Summary

This chapter has given a short overview of some of the subject matter

of psycholinguistics. It has given some examples of the types of phe-

nomena that psycholinguistics explores, as well as examples of how

psycholinguists carry out that exploration.

Some of the areas of psycholinguistics that have been highlighted as

of relevance to this book are:

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

Introduction 11

• planning a language utterance • the development of a syntactic structure for the expression of an

idea

• the selection of words to express that idea and their insertion into sentences

• the construction of words from smaller parts such as stems and endings, and from sounds and letters

• language-specific aspects of perception • the recognition of words, whether in speech or writing • the construction of a sentence structure during comprehension,

and the part played in that by meaning and other aspects of sen-

tences

Some of the techniques used by psycholinguists include:

• observation of speakers and listeners, including self-observation • conducting experiments in production and comprehension,

including response time experiments

• measuring brain activity during language tasks

The remaining chapters of this book look in more detail at many

aspects of the exploration of language processing set out in this

chapter.

Exercises The exercises below are intended to encourage you to use everyday contexts to explore some of the phenomena investigated in psycholinguistics. Exercises suggested in other chapters will involve more detailed examinations of these and other phenomena.

Exercise 1.1 Newspaper headlines are often incomplete sentences or leave out shorter or less important words. This can result in unintended amusing or ambiguous results, as in ‘Man gets nine years in violin case’, or ‘Sisters reunited after twenty years in checkout queue.’ Look out for such instances when you read newspapers, and try to work out why the headlines have the effect they do. Alternatively, Google ‘Amusing headlines’ to find a number of collections. Think about what it is that we do when we read texts that result in such amusing interpretations.

Exercise 1.2 Listen out for speech errors in the conversations you have during the day. If you can do this without disrupting the conversations, keep a record of these errors in a notebook. These can be errors that other people make, or errors you notice yourself making. See if you can identify any patterns in the errors. In very general terms (we will look at the details later), what do the errors tell you about speech production? Is it always easy to say what the error involves or why the error has come about? How representative and reliable do you think your examples are? Can you think of problems with this way of collect- ing speech error data?

https://www.cambridge.org/core/terms

https://doi.org/10.1017/CBO9780511978531.002

https://www.cambridge.org/core

12 INTRODUCING PSYCHOLINGUISTICS

Exercise 1.3 If you find yourself (or a conversational partner) in a tip-of-the-tongue state, i.e. you know there is a word for what you want to say but you just can’t find it, then write down what you (or the speaker if it is not you) can remember about the word. I.e. how long the word is, what kinds of sounds it has, what the stress pattern of the word is, what other words are similar in sound or meaning. If you subsequently are able to recall the word, how much overlap does it have with the characteristics you were able to remember?