Need help ASAP plz!!!1
Ayzee94-
2.pdf
When knowledge causes failure: Children’s extension of novel
adjectives and the interpretation of one
Kazuko Hiramatsu a,*, Keli E. Rulf
b , Samuel D. Epstein
c
a English Department, University of Michigan-Flint, 303 E. Kearsley St., Flint, MI 48502-1950, USA
b Department of Communication Sciences and Disorders, Northwestern University, Evanston, IL 60208, USA
c Department of Linguistics, University of Michigan, Ann Arbor, MI 48109-1220, USA
Received 7 July 2009; received in revised form 8 July 2009; accepted 8 July 2009
Available online 26 August 2009
Abstract
When studying language acquisition, the interaction of different modules (e.g. syntax, semantics, pragmatics) must be
considered. In this case study we scrutinize a body of experimental work on adjectival acquisition and explore how this interaction
affects the assessment of children’s linguistic knowledge.
Previous research indicates that 3-year-old children extend novel adjectives to objects within but not across the same basic-level
category. However, Rulf (2004) argues that the pronoun one in the experimental prompt used is pragmatically odd in many of these
studies (Waxman and Markow, 1998; Klibanoff and Waxman, 1999; Waxman and Klibanoff, 2000; Waxman and Booth, 2000). In
the first experiment, we investigated how adults interpret the pronoun one. Our results strongly suggest that even adults would have
failed the experimental task. In a second experiment with children, we eliminated the pragmatically odd pronoun usage and found
that both 3- and 4-year olds were able to extend novel adjectives across categories. This suggests that previous task failure by 3-year
olds was due not to a non-adult interpretation of adjectives, but to an adult-like interpretation of the pronoun. If so, previous studies
have underestimated the grammar of these children, both with respect to adjectival interpretation and the interpretation of the
pronoun one.
# 2009 Elsevier B.V. All rights reserved.
Keywords: Language acquisition; Adjective; Pronoun; Word learning; Methodology
1. Introduction
In the study of child language acquisition, we attempt to understand children’s underlying non-observable and
modularized knowledge based on observable behavior such as data from a controlled experiment or spontaneous
speech. The task of determining a child’s linguistic knowledge becomes even more challenging when we must take
into consideration the interaction of multiple modules such as syntax, semantics and pragmatics. Because of this, the
linguistic probes or instructions that are used to investigate children’s knowledge are extremely important. If our goal
is to determine the linguistic knowledge state of the child, we need to provide conditions that best reflect it. Once we
identify the child’s grammar, we must then also answer crucial questions about learnability. Does the experimental
www.elsevier.com/locate/lingua
Available online at www.sciencedirect.com
Lingua 120 (2010) 1209–1218
* Corresponding author. Tel.: +1 810 762 3285; fax: +1 810 237 6666.
E-mail addresses: [email protected] (K. Hiramatsu), [email protected] (K.E. Rulf), [email protected] (S.D. Epstein).
0024-3841/$ – see front matter # 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.lingua.2009.07.002
evidence indicate that the child has or lacks the adult grammar? Of course, this determination is dependent on a
determination of the adult grammar itself, i.e. the ‘‘end-sate’’ of child grammar development. If, by hypothesis, a child
does not have the target (adult) grammar, how and when does she acquire it? (See Hamburger and Crain, 1982, 1984;
Crain, 1991; Crain and Thornton, 1998; and the references they cite, for more discussion.) In this case study we
scrutinize a body of experimental work on the acquisition of adjectives and we explore how the interaction of different
modules affects the assessment of children’s linguistic knowledge.
2. The acquisition of adjectives
It is a fundamental goal of developmental psycholinguistic theory to determine children’s early interpretation of
lexical categories, such as adjectives and nouns, and the modificational semantic relations imposed upon these
categories in varying syntactic constructions. (This of course does not necessitate the adoption of a theory of human
grammars that is itself construction-specific.) In a series of important and intriguing studies, Klibanoff and Waxman
(1998, 1999, 2000) showed that 4-year olds extended novel adjectives to objects of both the same and different basic-
level categories. In contrast, 3-year olds extended novel adjectives only to objects of the same basic-level category. In
subsequent studies Waxman and her colleagues have investigated what conditions, such as offering comparisons or
multiple examples, facilitate across-category extension of novel adjectives. (See for example Waxman and Klibanoff,
2000; Waxman and Booth, 2000; Waxman, 2002.)
All of these studies have involved the same basic task. The child was presented with a target object and asked to
choose between two test objects: one with the same property as the target and another with a contrasting property. In
the within-category condition, a bumpy toy horse was presented and the child was then asked to choose between a
bumpy toy horse and a smooth toy horse. In the across-category condition, a bumpy toy rhinoceros was initially
presented instead of a bumpy toy horse.
To study the linguistic cues that children use to differentiate adjectives from nouns, researchers have manipulated
the syntactic frame that introduces the novel adjective. Waxman’s earlier studies (Klibanoff and Waxman, 1998, 2000)
used prompts with an explicit mention of the object being described by the novel adjective.
(1) Within-category condition
Let’s look at this horse. Gogi says this is a very blickish horse.
Can you give Gogi another horse that’s blickish?
(2) Across-category condition
Let’s look at this rhinoceros. Gogi says this is a very blickish rhinoceros.
Can you give Gogi a horse that’s blickish?
In later studies, the concrete noun used in the prompt (e.g. ‘‘horse’’) was changed to the pronoun one, as shown
below.
(3) Let’s look at this one. Gogi says this is a very blickish one.
Can you give Gogi another one that’s blickish? (Klibanoff and Waxman, 1999)
Klibanoff and Waxman (1999) initially made this change to test the possible biasing effect of the explicit mention of
the basic category (e.g. ‘rhinoceros’ in (2)) on 3-year olds in the across-category condition. However, even when the
explicit noun was replaced with the pronoun one, the results were comparable. Three-year olds were able to extend
novel adjectives only in the within-category condition, while 4-year olds were able to do so in both conditions.
Waxman’s central hypothesis based on these results is that children fail to extend novel adjectives across categories
early in the course of psycholinguistic development unless there is additional input such as a comparison or multiple
examples. Since there was no difference in the results using the new prompt in (3), and under the assumption that the
pronoun is preferable to the explicit noun, Waxman and colleagues used the pronoun prompt in all subsequent studies.
We agree that it is important to eliminate the explicit mention of the noun in the original prompts for the reasons
mentioned above. However, we believe the change they implemented is potentially problematic. Although a pronoun
is, in this respect, arguably preferable to an explicit noun, the use of ‘‘another one’’ in the prompt is pragmatically odd.
K. Hiramatsu et al. / Lingua 120 (2010) 1209–12181210
When the pronoun one is used, there is normally a linguistic antecedent. The interpretation of the pronoun one in
English has received much attention within the generative syntactic literature. (See for example, Baker, 1978;
Hornstein and Lightfoot, 1981; Radford, 1981, 1988; Hamburger and Crain, 1984, as well as the references they cite,
who discuss the acquisition of the X-bar theory of syntactic structure. See also Crain and Lillo-Martin, 1999; and the
appendix to Epstein et al., 1996; for explication of the X-bar theory in this domain.)
In the example below, the linguistic antecedent of one is ‘horse’.
(4) Gogi says this is a very blickish horse. Can you give Gogi another one that’s blickish?
When there is no linguistic antecedent, as with the prompts in Waxman’s experiments, the so-called exophoric
interpretation of one must be found. 1
Consider the within- and across-category conditions again, repeated below with Waxman’s revised prompt with one. 2
(5) Within-category condition
[Present child with target object (a toy horse).]
Let’s look at this one. Gogi says this is a very blickish one.
Can you give Gogi another one that’s blickish?
[Present child a choice of two test objects (2 toy horses).]
(6) Across-category condition [Present child with target object (a toy rhinoceros).]
Let’s look at this one. Gogi says this is a very blickish one.
Can you give Gogi another one that’s blickish?
[Present child a choice of two test objects (2 toy horses).]
The most plausible antecedent of one in the question part of the prompt is the nominal name of the first object shown
(e.g. ‘horse’ in (5), ‘rhinoceros’ in (6)). The question is whether a generic interpretation of one as ‘animal’ or ‘thing’,
constitutes a second possible adult semantic interpretation of one in these contexts. 3
If so, the instructions are in fact
ambiguous and we need also to determine which if either of the two interpretations is favored or disfavored in this context.
In the within-category condition, whether or not you interpret one in the question part of the prompt as ‘horse’ or ‘thing’,
there is an appropriate matching horse to select. This is not the case in the across-category condition in (6). If you interpret
one in the question as ‘rhinoceros’, there is no appropriate object (no other blickish rhinoceros) present in the toy array to
select (Rulf, 2004). This problem is arguably exacerbated given the use of not just ‘‘one’’ but ‘‘another one’’ in the
prompt. The only way to successfully extend a novel adjective in the across-category condition is by interpreting one
more generally, such as ‘thing’.
To test whether this purportedly pragmatically odd use of one in the prompt prevented children from extending
adjectives across categories, we conducted two experiments. In the first experiment, we tested the adult interpretation
of the pronoun one. We wanted to determine whether they would allow a generic interpretation of one as ‘thing’ in the
across-category condition in (6). If adults disallow or strongly disprefer the generic interpretation and fail to select a
property-matched object in the across-category condition, it is not unlikely that 3-year olds also failed to extend novel
adjectives across categories for the same reason. Note, there is independent evidence from Lidz et al. (2003) and
Radford (1990a, 1990b) that children as young as 18 months interpret one as adults do. It is, of course, possible that in
this domain there exists U-shaped learning where 18-month olds have the adult interpretation of one, 3-year olds have
lost it, and 4-year olds have regained it. That is, we cannot deduce from 18-month-old competence regarding one that
K. Hiramatsu et al. / Lingua 120 (2010) 1209–1218 1211
1 Generative linguistics has much to say about linguistic anaphoric interpretation and syntactic analysis. Whether one is interpreted as ‘horse’ or
‘thing’ is not determined by X-bar theory but rather by pragmatics. However, we are not presenting a comprehensive theory of pragmatic conditions
under which exophoric interpretations are adopted. We would need a full theory of lexical semantics, pragmatics, and syntax to articulate precise
syntactic and ‘‘real-world’’ conditions under which anaphoric and exophoric interpretations are adopted, and this falls outside the scope of this paper. 2
Klibanoff and Waxman (1998, 1999) describe the procedure in the following way: ‘‘Children were shown a target object . . . and then asked to
choose between two test objects’’. We interpreted this to mean that the researchers showed one object first followed by two other objects, as opposed
to showing all three objects simultaneously. 3
There are countless other generic interpretations possible, such as ‘‘something that moves’’ or ‘‘something with legs’’. We will be focusing our
discussion on just two possibilities: ‘‘animal’’ and ‘‘thing’’.
there must also be identical competence at 3 years old. We do however take such identity to be the null hypothesis
(under the so-called Continuity Hypothesis).
In Experiment 2, we revised the prompt by removing the allegedly pragmatically odd pronoun one and used a new
syntactic frame involving the phrase ‘‘all over’’. We tested 3- and 4-year olds, as well as adults, on whether they are
able to extend novel adjectives in the across-category condition.
3. Experiment 1
3.1. Participants
Twenty-four adult native speakers of English participated in the study for course credit. One other adult subject
completed the task but was excluded based on comments afterward that she had responded as if she were a child.
3.2. Procedure
The experimenter showed subjects 16 sets of pictures, 1 set at a time. After presenting each set of pictures, the
experimenter pointed to the target picture and gave the following prompt.
(7) This is a(n) ________ one. This one is ________. Is there another ________ one?
We omitted the intensifier ‘‘very’’ that Waxman and Markow (1998) used in the first part of the prompt since ‘‘very
blickish’’ is potentially confusing in that the intensifier ‘‘very’’ modifies an adjective bearing the -ish suffix, meaning ‘not
very’ or ‘somewhat’. We also changed the question wording. We thought that the existential question ‘‘Is there another
one?’’ gave the subject the option of responding ‘‘no’’ more clearly than asking ‘‘Can you give another one?’’, which may
presuppose in this context, the existence of another (an interpretation for which we will present anecdotal evidence
below).
Participants were asked to respond to the prompt by pointing. Because we were interested in the adult interpretation
of one and not whether adults can extend novel adjectives, we used familiar color terms: red, orange, yellow, green,
blue, purple, white, and black. 4
Subjects were tested individually and responses were recorded by the experimenter.
3.3. Materials
Each picture set contained five drawings of common objects. Two test items were of the same category as the target
item and two test items were of a different category. The conditions differed on which test item(s) matched the color of
the target item. In the control case, there was no item of the same color. In the within-category condition, a test item of
the same category had the same color as the target. In the across-category condition, a test item of a different category
had the same color as the target. In the within/across-category condition, two test items (one from the same category,
one from a different category) had the same color as the target. The prompt in (7) was used throughout all four
conditions. A description of the picture sets used in each condition is shown in Table 1.
The picture of the target object was centered at the top of the page and the four test objects appeared below it. The
color-matched test items were balanced for location and appeared in all four positions. There were four examples of
each condition for a total of 16 sets of pictures. The picture sets were randomized and the presentation order was
counterbalanced across participants.
3.4. Coding
For each condition, we calculated the proportion of color-matched test items that were selected by each subject. On
the within/across-condition, we counted the number of color-matched test items from a different category.
K. Hiramatsu et al. / Lingua 120 (2010) 1209–12181212
4 This experiment is designed to probe just the adult interpretations of the pronoun one so the methodology (use of real words, change in prompt,
number of test items) is different from Waxman’s in several aspects. We did not test adult subjects using Waxman’s exact methodology and leave this
for future research.
3.5. Results
Subjects responded as expected on the control and within-category conditions. On the control condition, all 24
subjects responded ‘‘No’’, by hypothesis because there were no color-matched test items. On the within-category
condition, all 24 subjects responded ‘‘Yes’’ and pointed to the color-matched test items from the same category. For the
across-category condition, only 6 of the 24 subjects chose the color-matched test item on all four picture sets. An
additional two subjects chose the color-matched test item on one of the four picture sets. The remaining 16 subjects
answered ‘‘No’’. For the within/across-category condition, the vast majority (22 subjects) chose the test item from the
same category. Two subjects each simultaneously pointed to both the within-category and across-category color-
matched test items on most/all of the trials. 5
For each subject, we calculated the proportion for choosing a color-matched test item. Recall that on the within/
across-condition, we counted the number of color-matched items from a different category. The mean proportions are
shown in Table 2.
The difference in means by condition was statistically significant by a repeated measures one-way ANOVA
(F(3,69) = 87.58, p < .001). A pairwise comparison reveals statistically significant differences for all pairs except for the Within/Across and Control comparison ( p = .16). For the within- as compared to the across-category conditions,
subjects were more likely to choose the color-matched item in the within-category condition than in the across-category
condition. For the across-category as compared to the within/across-conditions, subjects were more likely to choose the
color-matched item from a different category in the across-category condition than in the within/across-condition.
3.6. Discussion
The results strongly suggest that adults prefer to interpret one as synonymous with the nominal name of the target
item rather than interpreting one generically as ‘thing’. For example, in the across-category condition in Table 1, most
adults seem to have interpreted ‘‘another yellow one’’ as ‘another yellow toothbrush’ and not as ‘another yellow thing’.
If there was nothing pragmatically odd about the use of ‘another one’ in the prompt, and if one could be interpreted as
‘thing’, we might expect some adults to select the color-matched test object from a different category in the across- and
within/across-category conditions. However, most subjects chose the color-matched test item from the same category.
K. Hiramatsu et al. / Lingua 120 (2010) 1209–1218 1213
Table 2
Experiment 1: mean proportions of property-matched objects chosen (standard deviations are in parentheses).
Condition Mean proportion
Within 1.0 (0)
Across 0.27 (0.44)
Within/across 0.08 (0.28)
Control 0 (0)
Table 1
Four conditions for Experiment 1.
Condition Example stimuli
Control Target item: yellow starfish
Test items: brown starfish, white starfish, blue mask, green mask
Within-category Target item: red butterfly
Test items: black die, yellow die, blue butterfly, red butterfly
Across-category Target item: yellow toothbrush
Test items: blue book, yellow book, red toothbrush, brown toothbrush
Within/across-category Target item: white shirt
Test items: white walrus, pink walrus, yellow shirt, white shirt
5 One of these subjects pointed to both matching test items on three of the four trials. The other did so on all four trials.
Given this preference, it is likely that adults would not succeed in the across-category condition if given the
instructions Waxman used in the child experiments.
If children have the same knowledge as adults regarding the lexical semantics and pragmatics of one, then they too
should prefer to interpret one as synonymous with the target item (‘rhinoceros’) and not as the generic category
(‘thing’). They too would then fail in the across-category condition, as they in fact did. In our view, even though it was
only anecdotally reported, we find it extremely interesting that Waxman and Markow (1998) reported that 21-month
olds in their across-category condition experiments
‘‘. . . could not find a satisfactory match for the target object. When the experimenter requested an object during the test phase, some infants looked around the room and under the table, as if searching for another test object. . . These observations suggest that infants may have been disturbed by the absence of a test object from the same
basic level kind as the target’’ (1326).
We are still left with six adults from Experiment 1, who selected a color-matched item in the across-category
condition, as well as the 4-year olds from Klibanoff and Waxman (1999), who selected a color-matched item in their
forced-choice across-category condition. Both groups were, we conjecture, trying (and able) to accommodate the task,
something that requires pragmatic competence and complex contextual interpretation of the experimenter’s intentions.
Klibanoff and Waxman (1999) offer this possibility to account for 4-year olds extending novel nouns to object properties.
In one study, children were first shown a ‘‘blicket’’ (a spotted dog) and were then asked to select ‘‘another blicket’’ from
two test items: a spotted snake and a solid-colored snake. Three-year olds performed at chance on this task but 4-year olds
preferred the property-matched item. Four-year olds know that nouns refer to object categories but there is no appropriate
choice in this forced-choice task. Klibanoff and Waxman conclude that 4-year olds, but not 3-year olds, consider
‘‘alternate interpretations when the preferred interpretation is not available’’ (1999:366). They also cite Prasada (1997)
who encountered the same phenomenon with adults. The same hypothesis might be advanced here also.
4. Experiment 2
In the second experiment, we tested whether both 3- and 4-year olds are able to extend novel adjectives across
categories if the prompt is revised to exclude the pronoun one.
Mintz and Gleitman (2002) and Mintz (2005) discuss the complex pragmatic conditions necessary to felicitously
use the pronoun one as well as the noun thing. For example, Mintz and Gleitman (2002) highlight the importance of
having an established domain of reference prior to using one. In their prompts, the puppet always referred to training
objects as ‘‘things’’ before using ‘‘one’’ to refer to it. Mintz (2005) points out that using both thing and one in the same
prompt may complicate interpretation because the use of one presupposes that the speaker has ‘‘a category in mind’’
while the use of thing does not. Referring to a familiar object as thing or referring to an object as one without first
providing a referent is potentially infelicitous. 6
For these reasons, in the prompts for Experiment 2, we avoided using either potentially complicating term and
instead used the phrase ‘‘all over’’ to force an adjectival interpretation of the novel word.
4.1. Subjects
Sixteen 3-year olds and sixteen 4-year olds (mean ages 3;7 and 4;6) participated in the study. 7
Thirty-six adults
participated as control subjects. All were native speakers of English.
4.2. Procedure
We used a picture selection task similar to those used in Waxman’s studies. We introduced a dragon puppet who
sometimes says funny things and explained to the participants that we needed their help in trying to understand what
K. Hiramatsu et al. / Lingua 120 (2010) 1209–12181214
6 We will not have a comprehensive account of the full interpretation of a lexical item without a formalized theory of the lexicon, semantics,
syntax, pragmatics and the interaction among them. We will not be presenting such a theory since it is outside the scope of this paper. We will also
not be experimentally testing how children interpret one or thing. 7
Three children (two 4-year olds and one 3-year old) could not attend to the task and were excluded.
the dragon meant. The experimenter showed participants six sets of pictures, one set at a time, and for each set, asked
them to respond to the following prompt.
(8) See what’s in this picture? The dragon says it’s ________ all over.
Look at these two pictures. Is there something here that’s ________?
4.3. Stimuli
Each picture set consisted of three black-and-white drawings of common objects or animals. Of the two test items,
one had the same pattern as the target item while the other had a contrastive pattern. An example picture set of each
condition is presented in Figs. 1 and 2.
Following Mintz and Gleitman (2002), we used the novel adjectives bisk, rup, drin, prall, stoof , zav that lacked
suffixes such as -ish. We avoided nonce words such as blickish because it is possible that the inherent meaning of the
suffix and its combinatorial properties may influence the interpretation of the novel adjective.
Participants were randomly assigned to the within- or across-category condition. Each condition consisted of six
trials, which were randomly ordered. The presentation order was counterbalanced across participants.
4.4. Coding
For each participant, we calculated the proportion of property-matched objects selected.
4.5. Results
As expected, all of the adult control subjects chose the property-matched objects in both conditions. Among the
child participants, there was no significant difference among the mean proportions by age (F (1,28) = 1.04, p = .32) or
K. Hiramatsu et al. / Lingua 120 (2010) 1209–1218 1215
Fig. 1. Experiment 2: example picture set for within-category condition.
condition (F (1,28) = 1.04, p = .32), nor any interaction effect (F (1,28) = 1.04, p = .32). The proportion of property-
matched objects was statistically different from chance for only the 4-year olds in the across-category condition,
t(7) = 4.58, p < .01 (Table 3). To examine individual patterns, we calculated the number of children who chose a majority of property-matched
objects (at least four property-matched objects) on the six trials (Table 4).
Five out of eight children in both age groups selected at least four matching objects in the within-category
condition. However, on the across-category condition, we see a difference between the two age groups. All but one 4-
year old chose a matching object on the majority of trials, while only half of the 3-year olds did.
K. Hiramatsu et al. / Lingua 120 (2010) 1209–12181216
Fig. 2. Experiment 2: example picture set for across-category condition.
Table 3
Experiment 2: mean proportion of property-matched objects chosen (standard deviations are in parentheses).
Within-category Across-category
Three-year olds 0.65 (0.30) 0.65 (0.37)
Four-year olds 0.65 (0.35) 0.88 (0.23)
Table 4
Experiment 2: number of children choosing at least four property-matched objects.
Within-category Across-category
Three-year olds 5 4
Four-year olds 5 7
4.6. Discussion
By removing the pragmatically odd use of the pronoun one in the prompt, 3-year olds, like 4-year olds, were able to
extend novel adjectives across categories. This suggests that previous task failure by 3-year olds might have been due
to an adult-like interpretation of the pronoun rather than an inability to extend novel adjectives.
Furthermore, by not using thing or one in the prompt at all, 3-year olds, for the first time to the best of our knowledge,
were shown to be able to extend across categories with just one example of the novel adjective. Recall that previous
studies (Klibanoff and Waxman, 2000; Mintz and Gleitman, 2002; Mintz, 2005; Waxman and Booth, 2000; Waxman and
Klibanoff, 2000) used other ‘‘richer’’ input such as comparisons or multiple pretrial examples. Sandhofer and Smith
(2007) and Mintz (2005) discuss the types of linguistic cues that children use to interpret adjectives and how the
taxonomic specification of the noun used in the syntactic frames is relevant. They compare taxonomically specified nouns
(‘‘a stoof horsie’’, ‘‘a stoof toy’’) and taxonomically underspecified nouns (‘‘a stoof one’’, ‘‘a stoof thing’’). The syntactic
frame we used in this experiment involved neither specified nor unspecified nouns and children succeeded in interpreting
the novel adjective. The role of the phrase ‘‘all over’’ as a linguistic cue for adjective learning merits further investigation.
Given learnability theory, it is a welcome result if children can be shown to extend novel adjectives across categories.
To the best of our knowledge, adjectives in human language are universally unrestricted in their semantic applicability.
That is, we know of no adjective that can modify ‘‘horse’’ but cannot modify ‘‘rhinoceros’’, hence we assume semantic
composition involving adjectives is entirely unrestricted. Importantly, note in this regard, that expressions like ‘‘honest
politician’’ and ‘‘green idea’’ are assigned a meaning by an English speaker (but not by a French monolingual). Such
expressions are not meaningless and native knowers of the I-language ‘‘English’’ in fact assign an intersective reading to
such constructions and interpret them as such. Indeed it is our (largely universal, but in part language-particular)
knowledge of lexical semantics coupled with our universal syntactic knowledge along with our knowledge of universal
compositional semantic principles that give rise to precisely these (sometimes bizarre) semantic interpretations.
We are left, however, with an unexplained difference in individual results between 3- and 4-year olds on our across-
category condition. Four-year olds are slightly better at choosing the property-matched object than 3-year olds are.
One possible explanation, which we leave for future research, is that both age groups (3-year olds and 4-year olds)
have the adult syntactic and semantic knowledge regarding adjectives but pragmatic factors were interfering, this time
for the 3-year olds. In the first part of our prompt, the object introduced is characterized as being ‘‘stoof all over’’.
However, the question part of the prompt (‘‘Is there something here that’s stoof?’’) does not include the phrase ‘‘all
over’’. Even though the phrase ‘‘all over’’ is missing, it is still arguably implied that the child should look for
something that is ‘‘stoof all over’’, not something else that is ‘‘merely stoof yet not stoof all over’’. Our inadvertent
omission of the phrase unnecessarily complicates the task since the child must recognize this implicit inference (or
pragmatic infelicity) in order to understand our intended instructions.
5. Conclusion
Waxman and Markow (1998) express the need for extreme caution when trying to determine, from observed behavior,
the relative development of a child’s co-developing lexicon, phonetics, phonology, morphology, syntax, semantics,
pragmatics, as well as perceptual and conceptual development. We agree with this important point, and as we suggested
here, other factors may have had an impact on the earlier results. Until the complex human behaviors displayed in such
experiments are correctly explained in terms of the coordination of the many and disparate human faculties recruited in
and underlying the performance of such complex tasks, the notion of ‘‘extension’’ itself remains to be clarified.
Acknowledgements
For extremely helpful comments and discussion, we are grateful to the following people: Pam Beddor, Julie Boland,
Tim Chou, Susan Gelman, Rick Lewis, Jeffrey Lidz, Carmel O’Shannessy, Acrisio Pires, Marilyn Shatz, and three
anonymous reviewers. We are especially grateful to Sandra Waxman for her generosity which allowed two of us (Rulf
and Epstein, Hiramatsu being unavailable at the time) the very fortunate opportunity to openly and amicably discuss
and debate many of the complex issues addressed here. We also thank audience members at the 2007 Michigan
Linguistic Society conference and those who attended a 2005 colloquium at the University of Michigan Linguistics
Department, where much earlier versions of some of this material were presented. We are also very grateful to Johan
K. Hiramatsu et al. / Lingua 120 (2010) 1209–1218 1217
Rooryck and Neil Smith for their interest in and consideration of our research. None of the aforementioned people
necessarily agree with any of our ideas or analyses presented here.
References
Baker, C.L., 1978. Introduction to Generative Transformational Syntax. Prentice Hall, Englewood Cliffs, NJ.
Crain, S., 1991. Language acquisition in the absence of experience. Brain and Behavioral Sciences 14, 597–612.
Crain, S., Lillo-Martin, D., 1999. An Introduction to Linguistic Theory and Language Acquisition. Blackwell Publishers, Oxford.
Crain, S., Thornton, R., 1998. Investigations in Universal Grammar. MIT Press, Cambridge, MA.
Epstein, S., Flynn, S., Martohardjono, G., 1996/1997. Second language acquisition: theoretical and experimental issues in contemporary research.
Behavioral and Brain Sciences 19, 677–714.
Hamburger, H., Crain, S., 1982. Relative acquisition. In: Kuczaj, S. (Ed.), Language Development: Syntax and Semantics. Lawrence Erlbaum,
Hillsdale, NJ, pp. 245–274.
Hamburger, H., Crain, S., 1984. Acquisition of cognitive compiling. Cognition 17, 85–136.
Hornstein, N., Lightfoot, D., 1981. Explanation in Linguistics: The Logical Problem of Language Acquisition. Longman, New York.
Klibanoff, R.S., Waxman, S.R., 1998. Preschoolers’ acquisition of novel adjectives and the role of basic-level kind. In: Greenhill, A., Hughes, M.,
Littlefield, H., Walsh, H. (Eds.), Proceedings of the 22nd Annual Boston University Conference on Language Development. Cascadilla Press,
Somerville, MA, pp. 442–453.
Klibanoff, R.S., Waxman, S.R., 1999. Syntactic cues to word meaning: initial expectations and the development of flexibility. In: Greenhill, A.,
Littlefield, H., Tano, C. (Eds.), Proceedings of the 23rd Annual Boston University Conference on Language Development. Cascadilla Press,
Somerville, MA, pp. 361–372.
Klibanoff, R.S., Waxman, S.R., 2000. Basic level object categories support the acquisition of novel adjectives: evidence from preschool-aged
children. Child Development 71, 649–659.
Lidz, J., Waxman, S., Freedman, J., 2003. What infants know about syntax but couldn’t have learned: experimental evidence for syntactic structure at
18 months. Cognition 89, B65–B73.
Mintz, T.H., 2005. Linguistic and conceptual influences on adjective acquisition in 24- and 36-month-olds. Developmental Psychology 41, 17–29.
Mintz, T.H., Gleitman, L.R., 2002. Adjectives really do modify nouns: the incremental and restricted nature of early adjective acquisition. Cognition
84, 267–293.
Prasada, S., 1997. Young children’s use of structural cues in learning the meanings of novel adjectives. Unpublished manuscript. Dartmouth College,
Hanover, NH.
Radford, A., 1981. Transformational Syntax: A Student’s Guide to Chomsky’s Extended Standard Theory. Cambridge University Press, Cambridge.
Radford, A., 1988. Transformational Grammar: A First Course. Cambridge University Press, Cambridge.
Radford, A., 1990a. The syntax of nominal arguments in early child English. Language Acquisition 1, 195–223.
Radford, A., 1990b. Syntactic Theory and the Acquisition of English Syntax: The Nature of Early Child Grammars of English. Blackwell Publishers,
Oxford.
Rulf, K.E., 2004. The Syntax, Semantics, and Early Acquisition of One. Undergraduate Honors Thesis. Department of Linguistics, University of
Michigan, Ann Arbor, MI.
Sandhofer, C., Smith, L.B., 2007. Learning adjectives in the real world: how learning nouns impedes learning adjectives. Language Learning and
Development 3, 233–267.
Waxman, S.R., 2002. Not by perception alone: conceptual and semantic factors underlying children’s extension of novel adjectives. In: Skarabela,
B., Fish, S., Do, A.H.-J. (Eds.), Proceedings of the 26th Annual Boston University Conference on Language Development. Cascadilla Press,
Somerville, MA, pp. 746–757.
Waxman, S.R., Booth, A.E., 2000. Distinguishing count nouns from adjectives: evidence from ‘four-month-olds’ novelty-preference and word-
extension. In: Howell, S.C., Fish, S.A., Keith-Lucas, T. (Eds.), Proceedings of the 24th Annual Boston University Conference on Language
Development. Cascadilla Press, Somerville, MA, pp.773–784.
Waxman, S.R., Klibanoff, R.S., 2000. The role of comparison in the extension of novel adjectives. Developmental Psychology 36, 571–581.
Waxman, S.R., Markow, D.B., 1998. Object properties and object kind: twenty-one-month-old infants’ extension of novel adjectives. Child
Development 69, 1313–1328.
K. Hiramatsu et al. / Lingua 120 (2010) 1209–12181218
