In press, Language Acquisition

Children’s computation of implicatures

Anna Papafragoua and Niki Tantaloub

a
Institute for Research in Cognitive Science, University of Pennsylvania,

3401 Walnut St., Suite 400A, Philadelphia PA 19104 (address for correspondence)

Email: anna4@linc.cis.upenn.edu

Tel: 215 898 0360

Fax: 215 573 9247

b
Department of Cognitive Science, Johns Hopkins University, 3400 N. Charles St.,

Baltimore MD 21218

Email: tantalou@vonneumann.cog.jhu.edu

1
 In press, Language Acquisition

Children’s computation of implicatures

Keywords: implicature; pragmatics; semantics; scalar inference; quantifiers

1. Introduction

This article focuses on a familiar kind of conversational implicature known as scalar

implicature (SI):

(1) A: Do you like California wines?

B: I like some of them.

Implicature: B doesn’t like all California wines.

Even though some is semantically compatible with all, it is used in (1) to communicate ‘some

but not all’. According to the traditional Gricean account of such examples, given that B

could have used a more informative term (all), and since it would have been relevant to use

all if it were true, A is entitled to infer that B is not, in fact, in a position to offer a statement

containing all - most probably because such a statement is not true. Similar interpretations

arise with logical connectives (“A or B” -> not A and B), modals (“possibly x” -> not certainly

x), and a variety of other terms which can be seen to fall on an informational scale (see

Grice, 1989; Horn, 1972, 1984; for varying perspectives, see Gazdar, 1979; Carston, 1988,

1990; Sperber and Wilson, 1986; Levinson, 2000; Chierchia, 2001).

Several studies show that preschool children have difficulties computing SIs. Chierchia,

Crain, Guasti, Gualmini and Meroni (2001) and Gualmini, Crain, Meroni, Chierchia and

* We thank two anonymous reviewers and the Squibs editors for very useful comments. This research was
partly supported by NIH grant # F32MH066020-02 to the first author.

2
 In press, Language Acquisition

Guasti (2001) discovered that, in scenarios which made a statement containing the stronger

term on a scale true (e.g. “Every boy chose a skate-board and a bike”), five-year-old children

- unlike adults – failed to reject a statement containing a weaker scalar term (e.g. “Every boy

chose a skate-board or a bike”). In a related set of studies, Noveck (2001) found that 7-9-

year-old children are more likely than adults to accept statements such as “Some giraffes

have long necks”, again presumably because they fail to generate the implicature Not all

giraffes have long necks (cf. also Smith, 1980; Braine and Rumain, 1981).

More recent work by Papafragou and Musolino (2003) showed that children’s

performance with implicatures improves under certain conditions. Their experiments

included a training phase during which children were introduced to a puppet which

occasionally said things which were ‘silly’ (i.e. true but infelicitous). In the main phase,

children were asked to judge descriptions of acted-out stories given by the puppet: in one

critical trial, a bunny completed a puzzle as part of a contest but the puppet said that “the

bunny did some of the puzzle”. The introduction of training and of clear informativeness

expectations in the critical trials made children more likely to judge underinformative

statements as bad descriptions of what happened. Nevertheless, even in these contexts,

children generally showed sensitivity to SIs only about half of the time1 (cf. also Papafragou,

2003a).

So far, the conclusion that children have limited sensitivity to scalar inferences has

mainly relied on studies of logical expressions (e.g. quantifiers, connectives, modals) ordered

in terms of entailment. However, it has long been known that scalar inferences can be

1
Certain scalar expressions seem to give rise to higher success with SIs: numerals are one such case
(Papafragou and Musolino, 2003), the degree modifier half is another (Papafragou, 2003b). There are good
theoretical reasons for treating such ‘exact’ numerical and degree modifiers differently from other scalars,
which may explain children’s exceptional performance (see Carston, 1988; Horn, 1992; Papafragou, 2003b
for discussion).

3
 In press, Language Acquisition

induced by any kind of salient contextual ordering, including stable (encyclopedic) or

arbitrary (ad hoc) partial orderings - cf. (2) and (3) respectively (Hirschberg, 1985; cf.

Fauconnier, 1975):

(2) A: Have you read ‘A Beautiful Mind’?

B: I’ve read chapter 1.

Implicature: B hasn’t read the whole book.

(3) A: Did you get an autograph from the Jacksons?

B: I got one from Janet.

Implicature: B didn’t get an autograph from Michael.

A psycholinguistic account of ordinary implicature needs to extend to cases such as (2) and

(3) alongside the more familiar cases of entailment scales. At present, however, nothing is

known about the derivation of such inferences by young children.

Furthermore, studies which document children’s limited awareness of SIs have primarily

relied on judgments of the acceptability2 of ‘weak’ scalar expressions in contexts in which a

stronger term is warranted. However, these tasks are different from the actual circumstances

in which SIs are computed during naturalistic conversations in several respects. First,

experimental conditions do not make it clear whether (or why) SIs should be considered as

part of what the speaker actually intended to communicate. In ordinary cases of intentional

communication such as (1), the speaker intends the addressee to conclude that he doesn’t

like all California wines (and he also intends the addressee to recover this intention on the

basis of what is said; cf. Grice, 1989). But in experimental designs used so far, the

2
Gualmini et al. (2001) and Chierchia et al. (2001) take scalar inferences to contribute to truth conditions
(following Chierchia, 2001). They therefore interpret their subjects’ rejections of the weaker statements as
truth-value judgments. We cannot go into the details of this theoretical approach here (see Horn, 2003 for a
defense of the Gricean, non-truth-conditional treatment of SIs). But we take our discussion of judgment
tasks to apply also to truth-value judgment tasks, since they are partly designed in this case to detect effects
of pragmatic inferencing.

4
 In press, Language Acquisition

computation of SIs was not similarly constrained by the speaker’s intention. In some

experiments (Papafragou and Musolino, 2003), the speaker uttered an underinformative

statement probably out of incompetence and may not have noticed it carries the potential

for conveying a SI (cf. the ‘silly’ puppet); in others (Noveck, 2001), underinformative

statements were presented out of context and therefore invited participants to reconstruct a

possible context in which they could be uttered by an actual communicator. In short,

previous tasks measured children’s sensitivity to potential implicatures in an effort to

approximate their performance with actual (communicated) implicatures.

Second, experimental scenarios in previous tasks did not raise specific enough

expectations of informativeness, so as to motivate the computation of (even a potential) SI.

In an exchange such as (1), A’s question sets up relatively clear expectations of cognitive

gains (i.e. to obtain information about B’s opinion of California wines in general) which B’s

response then fails to meet. But such expectations were not consistently provided in test

situations (e.g. the puppet was simply asked to report “what happened”; Gualmini et al.,

2001; Papafragou and Musolino, 2003). In fact, when expectations of cognitive gains were

made clearer, and hence a stronger alternative became more salient, children’s peformance

improved (Papafragou and Musolino, 2003).

Finally, previous tasks typically involve situations in which an utterance containing a

scalar term (e.g. “Some of the Xs Ved”) semantically conveys a true proposition (e.g. Some

and possibly all of the Xs Ved) but carries a (potential) implicature which is false (Not all of the Xs

Ved). In order to perform correctly in these tasks (i.e. reject the statement), hearers should

take the implicature (rather than simply the proposition expressed) as the basis for their

assent/dissent with the original statement. To do so, participants had to estimate the

experimenter’s goals in setting up the task. This can be subtle: adults - who are otherwise

5
 In press, Language Acquisition

able to compute SIs - when presented with underinformative statements (e.g. “Some

airplanes have wings”) without supporting linguistic or extra-linguistic context agree with the

statements about half of the time (Noveck, 2001).3 More importantly, such judgments are

removed from naturalistic conversations, where what is said and what is implicated are not

normally pitted against each other but are taken to jointly contribute to what is meant by the

speaker. Putting together the above observations, we conclude that the family of judgment

tasks, however useful as an initial tool in exploring awareness of SIs, may in fact

underestimate preschoolers’ ability to compute implicatures ‘in the wild’.

In this paper, we provide new experimental evidence on early implicature

comprehension. Our investigation focuses on preschoolers’ performance with SIs that are

licensed by the familiar quantificational scale <all, some>, as well as by encyclopedic and ad

hoc partial orderings. Unlike previous studies which have relied on truth value/pragmatic

judgments, we introduce a new method which directly tests for the computation of SIs. This

method reproduces, to the extent possible, the textbook environments for SIs given in (1)-

(3). Our study has two main goals: first, to collect data about certain kinds of SIs which have

so far escaped experimental attention; second, to explore possible reasons for failures in

early implicature-calculation in children. Specifically, if earlier reports of children’s failures

with SIs are (at least in part) due to the specific character of judgment tasks, there is an open

possibility that children will be successful with scalar inferences from quantificational,

encyclopedic and ad hoc scalar orderings if a different method is used; in that case, there

might be no difference in children’s performance across the three types of scalar inference.

3
The training task in Papafragou and Musolino (2003) was designed to overcome this difficulty by

6
 In press, Language Acquisition

2. The experiment

2.1 Participants

Thirty Greek-speaking children participated in the study. They ranged in age from 4 years 1

month to 6 years 1 month (mean age: 5 years 3 months). All children were recruited in a

daycare center in Athens.

2.2 Materials and procedure

Children were presented with a set of animals and were told that they would play a game.

Each animal would be assigned a certain job. If at the end of the game the animal performed

the job, children should give the animal a prize, if not, the animal should get nothing.

Children were randomly assigned to one of three conditions, which corresponded to a

certain scale type: quantificational, encyclopedic and ad hoc.

The quantificational condition included the familiar <all, some> scale (Gr <oli, meriki >).

In a typical trial, an elephant was given a set of four paper stars and was told by the

experimenter that he had to color them. The elephant then went into a dollhouse to do the

coloring in quiet. After a while, he came out of the house and the following conversation

took place:

(4) Experimenter: Did you color the stars?

Elephant: I colored some. 4

showing that true but infelicitous statements should be rejected.

4
Examples are translated from Greek throughout. In Greek, as in English, all conditions included
contrastive stress on the scalar expression in the animal’s response.

7
 In press, Language Acquisition

After hearing the animal’s response, children had to decide whether the animal should

receive a prize or not and justify their response. We hypothesized that, if children were able

to compute SIs, they should refuse to award a prize to the elephant; furthermore, their

justifications should reflect their sensitivity to the presence of the implicature. If children

ignored SIs (e.g. if, in the above example, they interpreted some as being compatible with all),

they should be more generous in awarding prizes.

The encyclopedic condition included a set of orderings which were licensed by world

knowledge (and supported by the visual context). For example, a bear had to eat a sandwich

which consisted of bread, cheese and ham. She decided to go into a nearby dollhouse so as

not to litter the place with crumbs. When she came out of the house, the experimenter

asked:

(5) Experimenter: Did you eat the sandwich?

Bear: I ate the cheese.

Finally, the ad hoc condition introduced a range of circumstantial, context-specific

orderings. In one of the scenarios, a cow was assigned the task of wrapping two gifts, a toy

parrot and a doll. Since she was too embarrassed to wrap them up in front of everyone, she

decided to go into the dollhouse. When the cow re-appeared, the experimenter asked:

(6) Experimenter: Did you wrap the gifts?

Cow: I wrapped the parrot.

This task offers a fairly straightforward means of evaluating children’s pragmatic

sophistication by making a certain behavior (here, the refusal to give a reward) contingent on

the spontaneous computation of an implicature. The experimental scenarios resemble

naturalistic communicative circumstances in which implicatures are actually computed: SIs

are part of what the animals intended to convey and their recovery is set off by the fact that

8
 In press, Language Acquisition

the animals failed to observe the required level of informativeness in answering the

experimenter’s question. Notice that the choice of the weaker alternatives in (4)-(6) has a

natural motivation: animals which were unable or unwilling to complete their task chose to

report their partial progress (and only imply that the task was not completed) in the hope of

getting at least some reward.5 For these reasons, the present method is an improvement on

pragmatic judgment tasks previously used as a means of assessing early implicature

calculation.

Children also received a number of control trials (which were identical across conditions

and did not involve scalar expressions). In control items, the animal characters always

performed the action they had been assigned. Control items ensured that children could give

positive (alongside negative) responses when asked whether an animal should be rewarded.

A full list of experimental items is given in the Appendix. 6

Ten children participated in each condition. In the quantificational group, children

ranged from 4 years 1 month to 6 years 1 month (age mean: 5 years 3 months); in the

encyclopedic group, children’s age ranged from 4 years 10 months to 6 years (mean: 5 years

4 months); in the ad hoc group, children ranged from 4 years 11 months to 6 years (mean: 5

years 2 months). In each condition, subjects received four control trials and four test trials

administered in a pseudo-random order. Within each condition, order of presentation was

counterbalanced among subjects.

5
In the words of Larry Horn (2003), “it’s safer to implicate the bad news rather than to assert it as part of
what is said”.
6
One could argue that children in the test trials might choose not to award a prize simply because animals
did not respond with a simple ‘Yes’. We have two reasons to doubt that this could be true. First, we took
care to ensure that animals never answered the experimenter’s ‘Yes/No’ question with a simple ‘Yes’ or
‘No’ but always offered a complicated/long response. In test trials that response was equivalent to a ‘No’
while in control trials it was equivalent to a ‘Yes’. Second, we have pilot evidence from similar
experiments in English which shows that children can successfully reward the animals if completion of the
desired action is part of an entailment (“Did you color some of the stars?” – “I colored all of them”). Both

9
 In press, Language Acquisition

2.3 Results and discussion

The overall result from the test trials is that children overwhelmingly refused to give a

prize to the animal. Specifically, children correctly withheld the prize 77.5% of the time in

the quantificational cases, 70% in the encyclopedic cases and 90% in the ad hoc cases. There

was no reliable difference among these means (F(2, 27) = 0.72, p=0.49). Overall, children’s

performance in test trials was reliably different from chance responses (p<.0001).

In control trials, children were successful 97.5% in the quantificational condition, 100%

in the encyclopedic condition, and 92.5% in the ad hoc condition. Again no significant

difference was found among these means (F(2, 27) = 2.1, p = 0.14). Children’s performance

in control trials was significantly different from chance (p<.0001). Results from the test and

control trials are summarized in Table 1.7

Table 1

Proportion of correct responses

Condition Test trials Control trials

Quantifier 77.5% 97.5%

Encyclopedic 70% 100%

Ad hoc 90% 92.5%

these pieces of evidence show that children do not treat all indirect responses to ‘Yes-No’ questions as
being equivalent to a ‘No’.
7
Statistical comparison showed that children were better on control than on test trials (p=.02). A closer
look at incorrect responses in test trials reveals that certain children consistently failed the task, while
others consistently passed. Specifically, ‘failers’ include 4 children who failed all 4 test items and 2
children who failed 3 out of 4; ‘passers’ include 3 children who passed 3 out of 4 test items and 21 children
who passed all 4. Children who respond incorrectly in test items may be genuinely unable to detect SIs;

10
 In press, Language Acquisition

After providing their response, children were asked to justify their answer. In cases

where children decided that a prize should be awarded, they justified this answer by stating

that the animal had completed the action (or “had done its job”, “had done what we told it

to”, etc.). More interestingly, in cases where children refused to award a prize, they always

did so for the right reason, namely because they had inferred from the use of a ‘weaker’

scalar term that a stronger term did not apply. Specifically, in the quantificational case,

children in the majority (72%) of cases justified their negative responses by invoking the

strong quantifier all: to use an example mentioned in the previous section, children refused

to give a prize to an elephant who colored some of the stars because he didn’t color all of

them. In the encyclopedic case, children’s justifications were more varied, with the most

popular one involving the use of only (which surfaced 43% of the time). Finally, in the ad hoc

condition, variability in children’s justifications increased (since the scalar ordering in these

cases was neither as stable nor as transparent as in the previous two conditions), but their

form again showed evidence for the successful retrieval of a scalar implicature. A full list of

the types of justification offered by children on test trials is given in Table 2.

alternatively, they may be overly lenient, awarding prizes even for incomplete actions. Our data do not

11
 In press, Language Acquisition

Table 2

Children’s justifications for negative responses on test trials

Conditions

Quantificational Encyclopedic Ad hoc

‘He didn’t do all the _’.8 72% 25% 11%

‘He only did _’. - 43% 17%

‘He didn’t do the rest (too)’. - 7% 19%

‘He didn’t do _ [other part] (too)’. - 3.5% 25%

‘He didn’t do it (right)’. 3% 11% 14%

‘He did some/a few’. 12.5% 3.5% -

Other 12.5% 7% 12.5%

3. General discussion

These results provide compelling evidence for children’s early ability to compute implicated

aspects of the speaker’s meaning. Specifically, we show that, in contexts which approximate

naturalistic conversations, children are capable of assessing informativeness expectations

built during a talk exchange and derive SIs when these expectations are not met by the

allow us to adjudicate between these two possibilities at this point.

8
Some justifications included a full verb rather than do (‘He didn’t clean all of _’, etc.). We present all
justifications schematically here for ease of exposition.

12
 In press, Language Acquisition

speaker’s conversational contribution. The ability to assess expected levels of

informativeness involves making spontaneous comparisons between a given linguistic

stimulus (e.g. some) and other possible stimuli the speaker could have used in order to

produce the cognitive effects s/he intended (e.g. all.). This comparative ability is

unambiguously demonstrated in subjects’ justifications for their negative responses which

regularly make reference to stronger alternatives (“He shouldn’t get a prize because he didn’t

do ALL of X”). The massive success of children in our study contrasts with results from

previous studies which have reported children’s difficulties with SIs on the basis of truth-

value/appropriateness judgment tasks and, in that sense, confirms the importance of task

demands for children’s performance with SIs (cf. Papafragou and Musolino, 2003). 9

The structure of critical trials in our experiment included salient expectations of cognitive

gains (set up by the experimenter’s question), and highly accessible stronger alternatives to

the weak scalars used by the animals. It would be worth investigating children’s pragmatic

sensitivity in circumstances where the cost of computing SIs becomes higher (e.g. where

alternatives become less accessible by changing the structure of the experimenter’s question,

or are altogether inferred from context). At present, the extent to which children can

correctly attend to informativeness expectations - and monitor linguistic alternatives - in

different contexts remains an open question. Nevertheless, other evidence suggests that the

ability to consider contrastive alternates is not only within young learners’ reach but is, in

fact, very active in language acquisition. For instance, it is well-known that observations of

lexical contrast are instrumental at very early stages of word learning, where the fact that an

adult used a novel word a rather than a (related) known word b in a certain context can be

9
Based on our results, one may expect that children should stop short of computing SIs when
informativeness expectations are satisfied by the lower scalar term. For some evidence in this direction, see
Papafragou (2003a).

13
 In press, Language Acquisition

used by young learners to restrict the denotation of a to an appropriate non-b range (Carey,

1978; Clark, 1987).

A striking aspect of the present results is the finding that children’s computations of

informativeness are sensitive not only to logical entailment but also to stable (‘encyclopedic’)

or arbitrary (‘ad hoc’) orderings. This is important because a large number of SIs in everyday

conversation are of this more idiosyncratic variety. More generally, these findings raise the

question of how children come to grasp a variety of other, non-scalar implicatures whose

computation relies on idiosyncratic assumptions tied to specific contexts (cf. A: “Do you

want to go to the movies?” - B: “I’m tired” → Implicature: B doesn’t want to go to the movies). The

calculation of such inferences is a core task of the utterance interpretation device, and an

account of how it is achieved is crucial for understanding how this device functions and

grows.

Finally, evidence of children’s comprehension of conversational implicatures may

provide some insights into the architecture of the utterance interpretation system in adults. It

has recently been proposed that certain types of scalar implicature belong to a class of

default inferences which are supported by the presence of stable, contrastive alternates in the

lexicon (Levinson, 2000). Such generalized conversational implicatures (GCIs) include the

familiar quantificational, modal, etc. scales and are assumed to arise by default, context-

independently, whenever an appropriate ‘weak’ scalar element is present. These inferences

are differentiated from particularized implicatures which depend on more circumstantial scalar

orderings and hence rely heavily on context (cf. our encyclopedic and ad hoc scales). Even

though this line of reasoning was not directly concerned with language development, one

might expect that generalized SIs would prove easier for children to derive than

particularized (context-specific) ones, other things being equal. For instance, it might seem

14
 In press, Language Acquisition

plausible that the inference from some to not all, for the same reasons that it becomes

generalized for adults, should be less computationally intensive for children compared to

more ad hoc SIs. However, the results reported in this paper do not offer evidence for such

a split.10 On the contrary, the present findings can be used to support a unified view of

implicature, according to which all implicatures are ‘particularized’ and rely on context-

specific computations of the speaker’s communicative intentions.

Appendix

A1. Test items (translated from Greek)

Condition Animal’s task Experimenter’s question Animal’s response

Quantifier Tiger has to eat 4 oranges ‘Did you eat the oranges?’ ‘I ate some’.

Elephant has to color 4 stars. ‘Did you color the stars?’ ‘I colored some’.

Horse has to clean 4 toys. ‘Did you clean the toys?’ ‘I cleaned some’.

Pig has to feed 4 frogs. ‘Did you feed the frogs?’ ‘I fed some’.

Encyclopedic Bear has to eat a sandwich. ‘Did you eat the sandwich?’ ‘I ate the cheese’.

Frog has to paint a house. ‘Did you paint the house?’ ‘I painted the roof’.

Bunny has to clean a wagon. ‘Did you clean the wagon?’ ‘I cleaned the wheels’.

Elephant has to wash himself. ‘Did you wash yourself?’ ‘I washed my ears’.(lit.)

10
For further arguments against GCI theory, see Carston (1995), Hirschberg (1985), Geurts (1998).

15
 In press, Language Acquisition

Ad hoc Giraffe has to put 4 animals in a row

(starting with cow). ‘Did you put the animals in a row?’ ‘I put the cow’. (lit.)

Cow has to wrap 2 gifts

(a toy parrot and a doll). ‘Did you wrap the gifts?’ ‘I wrapped the parrot’.

Dog has to clean a merry-go-round

(which has purple seats and one

yellow seat). ‘Did you clean the merry-go-round?’ ‘I cleaned the yellow seat’.

Lion has to feed snake

(its food: a lollipop and a sausage). ‘Did you give the snake its food?’ ‘I gave it the lollipop’.

A2. Control items (translated from Greek)

Condition Animal’s task Experimenter’s question Animal’s response

All conditions Bear has to drink some milk ‘Did you drink the milk?’ ‘I drank it and liked it’.

Worm has to read a book. ‘Did you read the book?’ ‘I read it. It was a nice story’.

Frog has to fix a broken chair. ‘Did you fix the chair?’ ‘I fixed it but it was hard’.

Starfish has to put flower in vase. ‘Did you put the flower in ‘In the beginning I couldn’t

the vase?’ find a vase but in the end I

found it and put the flower in’.

References

Braine M. and B. Rumain (1981). Children’s comprehension of ‘or’: Evidence for a sequence of

competencies. Journal of Experimental Child Psychology 31: 46-70.

16
 In press, Language Acquisition

Carey S. (1978). The child as word learner. In M. Halle, J. Bresnan and G. Miller (eds.), Linguistic theory

and psychological reality, 264-293. Cambridge, MA: MIT Press.

Carston R. (1990). Quanity maxims and generalised implicature. UCL Working Papers in Linguistics 2: 1-

31. Reprinted in Lingua 96 (1995): 213-244.

Carston R. (1998). Informativeness, relevance, and scalar implicature. In R. Carston and S. Uchida (eds.),

Relevance theory: Applications and implications. Amsterdam: Benjamins.

Carston R. (2002). Thoughts and utterances: The pragmatics of explicit communication. Oxford:

Blackwell.

Chierchia G. (2001). Scalar implicatures, polarity phenomena, and the syntax/pragmatics interface. University

of Milan-Bicocca ms.

Chierchia G., S. Crain, M. T. Guasti, A. Gualmini and L. Meroni (2001). The acquisition of disjunction:

Evidence for a grammatical view of scalar implicatures. In A. H.-J. Do et al. (eds.), BUCLD 25

Proceedings, 157-168. Somerville, MA: Cascadilla Press.

Clark E. (1987). The principle of contrast: A constraint on acquisition. In B. McWhinney (ed.),

Mechanisms of language acquisition, 1-33. Hillsdale, NJ: Erlbaum.

Fauconnier G. (1975). Pragmatic scales and logical structure. Linguistic Inquiry 6: 353-375.

Gazdar G. (1979). Pragmatics. New York: Academic Press.

Geurts B. (1998). Scalars. In P. Ludewig and B. Geurts (eds.), Lexicalische Semantik aus kognitiver sicht, 95-

117. Tübingen: Narr Verlag.

Grice P. (1989). Studies in the way of words. Cambridge, MA: Harvard University Press.

Gualmini A., S. Crain, L. Meroni, G. Chierchia and M. T. Guasti (2001). At the Semantics/Pragmatics

Interface in Child Language. Proceedings of SALT XI. Cornell University, Ithaca, NY.

Hirschberg J. (1985). A theory of scalar implicature. Doctoral diss., University of Pennsylvania.

Horn L. (1972). On the semantic properties of the logical operators in English. Doctoral diss. UCLA

Distributed by IULC, Indiana University.

Horn L. (1984). Toward a new taxonomy of pragmatic inference: Q- and R-based implicature. In D. Shiffrin

(ed.), Meaning, form, and use in context, 11-42. Washington, DC: Georgetown University Press.

17
 In press, Language Acquisition

Horn L. (1992). The said and the unsaid. In C. Barker and D. Dowty (eds.), Proceedings of SALT II, 163-192.

Columbus, OH: Ohio State University Linguistics Department.

Horn L. (2003). Grisotto Milanese. Talk delivered at the Workshop on Scalar Implicature and Polarity,

University of Milan-Bicocca, June 18-21.

Levinson S. (2000). Presumptive meanings. Cambridge, MA: MIT Press.

Noveck I. (2001). When children are more logical than adults: Experimental investigations of scalar

implicature. Cognition 78: 165-188.

Papafragou A. (2003a). Aspectuality and scalar structure. To appear, Proceedings from the 27th Annual Boston

University Conference on Language Development. Somerville, MA: Cascadilla Press.

Papafragou A. (2003b). Scalar implicatures in language acquisition: Some evidence from Modern Greek. To

appear, Proceedings from the 38th Annual Meeting of the Chicago Linguistics Society. Chicago: University

of Chicago Press.

Papafragou A. and J. Musolino (2003). Scalar implicatures: Experiments at the semantics-pragmatics interface.

Cognition 86: 253-282.

Sadock J. (1984). Whither radical pragmatics? In D. Shiffrin (ed.), Georgetown University Round Table on

Language and Linguistics, 139-149. Washington, DC: Georgetown University Press.

Smith C. (1980). Quantifiers and question-answering in young children. Journal of Experimental Child

Psychology 30: 191-205.

Sperber D. and D. Wilson (1986). Relevance: Communication and cognition. Cambridge, MA: Harvard

University Press. 2nd ed. 1995.

18
In press, Language Acquisition

19
 Filename: Children's_computation_of_implicatures.doc
Directory: F:\Scales
Template: C:\Documents and Settings\anna4\Application
Data\Microsoft\Templates\Normal.dot
Title: Spontaneous computation of implicatures by young children
Subject:
Author: anna4
Keywords:
Comments:
Creation Date: 22/4/2003 12:28 PM
Change Number: 253
Last Saved On: 1/10/2003 5:12 PM
Last Saved By: Anna Papafragou
Total Editing Time: 1.797 Minutes
Last Printed On: 15/6/2004 5:55 PM
As of Last Complete Printing
Number of Pages: 19
Number of Words: 3.867 (approx.)
Number of Characters: 22.431 (approx.)

View publication stats