Some Knowledge Children Don’t Lack Andrea Gualmini University of Maryland at College Park∗

1. Quantifiers and Negation in Child English Children’s interpretation of quantified expressions is the subject of a vast literature. In recent years, considerable attention has also been devoted to the interaction between quantifiers and negation. In particular, Musolino (1998) reported a series of experimental investigations on children’s interpretation of sentences like the following: (1) Every horse didn’t jump over the fence. (2) The detective didn’t find some guys. (3) The Smurf didn’t catch two birds. Before we illustrate the findings of Musolino’s investigations of child English, let us briefly summarize the relevant properties of adult English. A common feature to the sentences in (1) - (3) is that adults’ preferred interpretation involves some grammatical operation that ‘reverses’ the scope relations represented in the overt syntax.1 For example, consider the two possible readings of sentence (1), represented in (4). (4) a. ∀(x) (horse(x) → ¬ jumped over the fence(x)) b. ¬ ∀(x) (horse(x) → jumped over the fence(x)) On the isomorphic interpretation, (4)a, sentence (1) means that none of the horses jumped over the fence. On the non-isomorphic interpretation, (4)b, ∗

I wish to thank Martin Hackl, Norbert Hornstein, Luisa Meroni and, especially, Stephen Crain and Julien Musolino for helpful discussion. I also thank Amanda Gardner, Beth Rabbin, Nadia Shihab and Andrea Zukowski for their help in conducting the experiment. Finally, I thank the staff, teachers and children at the Center for Young Children at the University of Maryland at College Park, where the experiment was conducted. The reader is referred to Gualmini (2003) for a more extensive report of the present study. 1. In the present study we will ignore the fact that different quantifiers could receive scope in different ways. For an excellent overview, see Reinhart (1997).

© 2003 Andrea Gualmini. BUCLD 27 Proceedings, ed. Barbara Beachley et al., 276-287. Somerville, MA: Cascadilla Press.

277 sentence (1) is true in contexts in which some of the horses jumped over the fence, as long as not all of them did. It is usually reported that adult speakers of English have a preference for the interpretation in (4)b, in which negation takes scope over the universal quantifier every, despite the fact that negation follows the universal quantifier in the surface syntax. Let us now consider (2).2 Adult speakers of English accept (2) if the detective found some but not all of the guys present in the context. As suggested by the acceptability of (5) as a paraphrase of (2), one can describe adults’ interpretation of (2) by saying that the indefinite noun phrase some guys is interpreted as having scope outside of negation. (5) There are some guys that the detective didn’t find. More precisely, one can represent the interpretation of (2) as (6). (6) ∃(x) [guys(x) & ¬ the detective found (x)] Again, the relative order of the existential quantifier and negation in (6) is opposite of that of the indefinite some and negation in (2).3 To investigate children’s interpretation of quantifiers in negative sentences, Musolino (1998) conducted a series of experiments employing the Truth Value Judgment task. The Truth Value Judgment task is an experimental technique that allows one to investigate whether a specific interpretation of a target sentence is licensed by the child’s grammar (Crain and McKee, 1985; Crain and Thornton, 1998). The first relevant experiment by Musolino (1998) was concerned with sentences like (1), repeated below as (7). (7) Every horse didn’t jump over the fence. The children who participated in the experiment were asked to evaluate (7) as a description of a story whose final outcome is represented in Figure 1. Notice that the context in Figure 1 makes (7) true on its non-isomorphic reading (i.e., (4)b) because not every horse jumped over the fence. By contrast, the context in Figure 1 makes (7) false on its isomorphic reading (i.e., (4)a), because some of the horses jumped over the fence.

2. Throughout the paper, we will limit our attention to the stressed variant of the indefinite some. 3. It is unclear whether (2) can receive an isomorphic interpretation. However, it bears observing that both the isomorphic and the non-isomorphic reading are available if an indefinite like two is used, e.g., The detective didn’t find two guys.

278

Figure 1 The finding of the Musolino study was that five-year-old children consistently rejected (7) on the grounds that some of the horses had indeed jumped over the fence. One can describe children’s behavior by saying that they interpreted the target sentence on its isomorphic interpretation, assigning wide scope to the universal quantifier every. Another experiment conducted by Musolino (1998) was designed to assess children’s interpretation of negative sentences containing the indefinite some. This experiment used sentences like (2), repeated below as (8). (8) The detective didn’t find some guys. Sentence (8) was presented as a description of a story in which the detective had only found two of the four guys participating in the story (see Figure 2). The context employed by Musolino (1998) was designed to make (8) true on its adult interpretation (i.e., the non-isomorphic interpretation paraphrased in (9)) and false on its non-adult interpretation (i.e., the isomorphic interpretation paraphrased in (10)). (9) There are some guys that the detective didn’t find. (10) It is not the case that the detective found some guys.

278

279

Figure 2 The finding was that children as old as 5;9 rejected the target sentence, whereas the adult controls consistently accepted it. Many of the children interviewed by Musolino (1998) pointed out that (8) was incorrect because the detective had indeed found some guys (i.e., the two guys that appear on the top-right corner in Figure 2). Apparently, children’s rejection of (8) is indicative of the interpretation in (10), the isomorphic interpretation. A similar picture emerges from another experiment conducted by Musolino (1998) assessing children’s interpretation of sentences like The Smurf didn’t catch two birds. Attempting to provide a unified description of children’s interpretation of quantifiers in negative sentences, Musolino (1998) proposed the Observation of Isomorphism, which states that “when syntactic scope and semantic scope do not coincide [for adults - AG], children’s interpretations correlate with the interpretations determined by syntactic scope” (Musolino, 1998; p. 145).4 The Observation of Isomorphism is not presented by Musolino (1998) as a learning principle. Instead, the Observation of Isomorphism is taken to be a descriptive generalization that is derived from other properties of the Language Acquisition Device or from idiosyncratic properties of the particular quantifier involved. As a consequence, children’s adherence to the isomorphic interpretation of sentences containing negation and a given quantifier could in principle receive a different explanation for each quantifier. For instance, Musolino, Crain and Thornton (2000) explain young children’s interpretation of indefinites in the scope of negation as resulting from children’s developing knowledge of the 4. The Observation of Isomorphism is also invoked by Musolino (1998) to describe children’s interpretation of negative polarity items in questions (see Thornton (1994) for the original study and Gualmini (2003) for a review of children’s knowledge of polarity phenomena).

280 distinction between some and any and children’s preference for more restrictive grammars, i.e., grammars that initially license interpretations that are true in the narrowest set of circumstances (see Crain, Ni and Conway, 1994).5 The Observation of Isomorphism has been recently subjected to further scrutiny (for extensive review see Musolino, 2001). One of the aims of current research is to determine whether the Observation of Isomorphism describes a difference between children’s and adults’ linguistic competence or a difference between children’s and adults’ linguistic performance (see Musolino, 2000; 2001). In a series of experimental investigations, Musolino and Lidz (2002a,b) showed that the use of a positive lead-in dramatically improves children’s performance with negative sentences containing the universal quantifier. Recall that Musolino (1998) found that young children preferred the isomorphic interpretation in (12)a for sentences like (11), whereas adults preferred the nonisomorphic interpretation in (12)b. (11) Every horse didn’t jump over the fence. (12) a. ∀(x) (horse(x) → ¬ jumped over the fence(x)) b. ¬ ∀(x) (horse(x) → jumped over the fence(x)) Interestingly, Musolino and Lidz (2002b) show that the use of a positive lead-in, as illustrated in (13), led children to access the non-isomorphic interpretation of the target sentence much more often than had been observed in earlier research. (13) Every horse jumped over the log and/but every horse didn’t jump over the fence. Children’s improved performance with sentences like (13) was interpreted by Musolino and Lidz (2002a,b) as showing that the observed differences between children and adults reside in the performance component. In particular, Musolino and Lidz (2002a,b) argue that children’s non-adult performance with negative sentences containing the universal quantifier every results from their developing pragmatic knowledge. On this view, children do not differ from adults in their grammatical competence. Rather, children differ from adults in how their grammatical competence is put to use, possibly because of processing limitations or limited pragmatic competence. The experimental responses collected by Musolino and Lidz (2002a,b) are certainly at odds with an explanation that posits differences between children’s and adults’ grammatical competence. This should not be taken as evidence that the observed differences lie in the linguistic performance component, however. A third alternative should be considered, in view of the learnability problems 5. An alternative account of children’s interpretation of indefinites in negative sentences was proposed by Krämer (2000), who argued that children’s difficulty is limited to specific uses of indefinites.

280

281 that would arise if children and adults did not share the same processing mechanisms. The null hypothesis is that the observed differences follow from non-linguistic factors (see Crain and Wexler, 1999). For instance, one should evaluate whether the experimental responses documented in the literature could be due to unwelcome features of experimental design. 2. The Pragmatics of Negative Sentences Negative sentences provide a rich domain for linguistic research. Drawing upon the principles of pragmatic theory, numerous psycholinguists have observed that negative sentences are ordinarily more difficult to process than affirmative sentences (see De Villiers and Tager Flusberg, 1975; Wason, 1965). Several studies have also discussed the conditions under which the difficulty posed by negative sentences is mitigated. This research led to the conclusion that negative sentences are accompanied by specific felicity conditions. Consider (14), due to Wason (1972). (14) 5 is not an even number. The sentence in (14) is true. Still, (14) is more difficult for people to evaluate than (15), despite the fact that (14) and (15) apparently express the same proposition. (15) 5 is an odd number. Interestingly, the difficulty associated with (14) is considerably mitigated if the sentence is preceded by a positive statement, as in (16) (see Musolino and Lidz, 2002b). (16) 4 is an even number, and/but 5 is not an even number. Similar remarks about the difficulty associated with negative sentences were offered by De Villiers and Tager Flusberg (1975). Consider (17). (17) I didn’t drive to work. As pointed out by De Villiers and Tager Flusberg (1975; p. 279), the statement in (17) “is more plausible, and consequently easier to comprehend, if it is made by someone who normally drives rather than by someone who commutes by train.” According to De Villiers and Tager Flusberg (1975), this property of negative sentences results from the fact that “negative statements are generally used to point out discrepancies between a listener’s presumed expectations and the facts” (De Villiers and Tager Flusberg, 1975; p. 279).6 The observation that 6. See also Givon (1978) and Wason (1965).

282 negative sentences are associated with a specific set of felicity conditions led De Villiers and Tager Flusberg (1975) to conduct an experimental investigation. In the experiment, children were asked to complete negative questions (i.e., ‘This is not a ___?’) in contexts in which the use of a negative sentence was or was not plausible (e.g., in the plausible context, the experimenter had pointed to various instances of a particular object). The results show that children as young as two respond significantly faster when the negative question is presented in a plausible context. De Villiers and Tager Flusberg (1975) interpret this finding as evidence that very young children know that the felicitous use of negative sentences is governed by specific conditions of discourse felicity. To sum up, previous studies show that subjects experience a difficulty in processing negative sentences, at least in certain contexts. The same studies, however, suggest that the difficulty associated with negative sentences can be mitigated in various ways (e.g., if the target sentence is preceded by a positive lead-in or if it is used to point out the fact that an expectation went unfulfilled). The exact ‘nature’ of the difficulty associated with negative sentences remains to be determined. This issue was extensively addressed by Horn (1989), who argued that the facts above do not highlight a property of negative sentences per se, but rather reflect a consequence of the way negative sentences are ordinarily used. In order to support this claim, Horn (1989) shows how the infelicity of negative sentences in ordinary contexts can be explained by the interplay of pragmatic principles. In our view, the arguments discussed by Horn (1989) underscore the pragmatic nature of the phenomenon under investigation. For the purposes of the experimentalist, however, this should not undermine the importance of the phenomenon itself. Be it a semantic or a pragmatic phenomenon, the difficulty posed by the unmotivated use of negative sentences remains. As a consequence, in conducting experiments with child or adult subjects one should ensure that subjects’ responses are not vitiated by the infelicitous use of negative sentences. More specifically, before we attempt to determine whether linguistic competence or linguistic performance is responsible for children’s non-adult responses, one has to ensure that children’s non-adult responses persist in studies that satisfy the felicity conditions associated with negative sentences. 3. Experimental Design This section presents the findings of an experiment employing the Truth Value Judgment task designed to investigate the role of felicity conditions in children’s interpretation of the indefinite some in negative sentences (Crain and McKee, 1985; Crain and Thornton, 1998).7 In a Truth Value Judgment task, one experimenter acts out a short story in front of the child, using props and toys. 7. The experiment focuses on the interaction between negation and some because this kind of interaction resisted some of the contextual manipulations that lead to improved performance with other quantifiers (see Musolino, 2001).

282

283 The story constitutes the context against which the child evaluates the target sentence, which is uttered by a puppet that is manipulated by a second experimenter. The acceptance of the target sentence is interpreted as indicating that children can assign the target sentence an interpretation that makes it true in the context under consideration. By contrast, the rejection of the target sentence in a context suggests that the child’s grammar does not license any interpretation that makes the sentence true in the context under consideration. To investigate the role of felicity conditions in children’s interpretation of negative sentences, we focused on one particular way to mitigate the difficulty associated with negative sentences, inspired by the study by De Villiers and Tager Flusberg (1975). In particular, we controlled for children’s expectation about the final outcome of each story. Thirty children participated in the experiment. The children were divided in two groups. One group of 15 children (age: 4;00;20 to 5;5;18 – mean: 4;10) was asked to evaluate a negative sentence that truthfully described the final outcome of the story and expressed a mismatch between the final outcome and an expectation built up during the story. The second group of 15 children (age 4;01;27 to 5;07;18 – mean age: 4;11) was asked to evaluate a negative sentence that also truthfully described the final outcome of the story, but did not express a mismatch between the final outcome and an expectation that was established during the story. Let us illustrate a typical trial. Consider the trial in (18), which is a modified version of Story 1 of Experiment 3 conducted by Musolino (1998). (18) This is a story about a firefighter who is going to play hide and seek with four dwarves. While the firefighter counts, the dwarves look for a spot to hide. When the firefighter has finished counting, he starts looking for the dwarves. Initially, the firefighter cannot find any of the dwarves and he is ready to give up, but then he decides to try harder. He finds one dwarf who was hiding behind a barrel and he asks the dwarf: “Ok, am I done now?” and the dwarf says: “No! there’s three more dwarves for you to find.” The firefighter spots a second dwarf who was hiding inside the barrel, and he asks him: “Ok, am I done now?” but the dwarf says: “No! there’s two more dwarves for you to find.” The firefighter starts looking again, but then he says: “You know guys, those two dwarves did a very good job, I cannot find them. I must give up.” At this point, one group of children was presented with the sentence in (19) and the second group was asked to evaluate (20). (19) This was a story about a firefighter playing hide and seek with four dwarves and I know what happened. The firefighter didn’t find some dwarves. (20) This was a story about a firefighter playing hide and seek with four dwarves and I know what happened. The firefighter didn’t miss some dwarves.

284

Before we report the results, let us consider the extent to which the two target sentences differ. The target sentences in (19) and (20) do not differ in truthvalue, since both sentences are true in the scenario under consideration.8 Sentence (19) is true because there are two dwarves that the firefighter did not find, and (20) is true because there are two dwarves that the firefighter did not miss, namely the dwarves that the firefighter did find. The two sentences differ with respect to their felicity, however. The story conveys the expectation that the firefighter would find all the dwarves, and it does not build the expectation that the firefighter would miss all of them. As a consequence, (19) is felicitous because it expresses a mismatch between what happened and what was expected to happen (i.e., the number of dwarves that the firefighter found and the number of dwarves that he was expected to find). By contrast, (20) is infelicitous because there is no immediate mismatch between what happened and what was expected to happen (i.e., the number of dwarves that the firefighter missed and the number of dwarves that he was supposed to find). This simply follows from the lack of any expectation that the firefighter would miss all the dwarves. As a consequence of this difference, the experimental hypothesis was that children would respond in a fully adult-like fashion to sentences like (19), and that possible cases of non-adult responses would be limited to ones like (20). This is exactly what was found. Children accepted sentences like (19) in 54 out of 60 trials (90%), and they accepted sentences like (20) only in 30 out of 60 trials (50%).9 The difference between the two conditions runs deeper than the numbers suggest, moreover. In order to further assess children’s judgments, children who accepted the target sentences were asked to motivate their answers by pointing to the characters that “made the puppet right.” Interestingly, when children accepted sentences like (19), they consistently pointed out that the puppet was right because the firefighter had failed to find two of the dwarves. By contrast, some of the children who gave the adult (affirmative) response to (20) could not motivate their answer by pointing to the dwarves that the 8. The optimal design of the Truth Value Judgment task requires a context that falsifies the adult interpretation of the target sentence and verifies the nonadult interpretation under investigation. As discussed by Musolino (1998), however, this ideal state of affairs cannot be obtained in the present case, since the non-isomorphic reading of a sentence like The detective didn’t find some guys is true in a superset of the circumstances in which its isomorphic reading is true. Given this state of affairs and the desire to minimize the differences between our experiment and previous studies, our experiments employ the same design as Experiment 3 in Musolino (1998): the context makes the isomorphic reading false and the non-isomorphic reading true. 9. Eighteen native speakers of English participated as adult controls. Four subjects accepted sentences like (19) in 13 out of 16 trials (81%), and the remaining fourteen subjects accepted sentences like (20) only in 27 out of 56 trials (48%).

284

285 firefighter had not missed, i.e., the dwarves that the firefighter had found. Despite their correct judgment, it is possible that these children’s grammatical competence was offset by the infelicity of the linguistic material presented to them, and they may have accepted the target sentence without accessing its adult interpretation (see Grimshaw and Rosen, 1990). 4. Conclusions The experimental findings of the present study reveal children’s adult-like interpretation of negative sentences containing the indefinite some when the felicity conditions associated with the target (negative) sentences are satisfied. As we observed, previous research has emphasized the descriptive character of the Observation of Isomorphism. The experimental findings of the present study suggest that, as it pertains to the interpretation of some in negative sentences, the Observation of Isomorphism simply describes the way children (and adults) respond to infelicitous sentences. Thus, the Observation of Isomorphism does not describe either a difference between children’s and adults’ competence or their performance systems. Rather, the difference is between children’s and adults’ ability to deal with infelicitous linguistic material. The present paper focused on the difference between children and adults in interpreting negative sentences containing the indefinite some. This was a pretext, however. Our goal was not to adjudicate between alternative explanations of children’s non-adult interpretation of some in negative sentences. Our goal was to determine whether children’s non-adult interpretation of some in negative sentences, as documented by previous research, demands a grammatical explanation. We have argued that this is not the case, and we have credited 4-year-olds with full knowledge of the grammatical properties of the indefinite some. This does not exclude the possibility that children might fail to distinguish between some and any at earlier stages of language development. Our results, however, show that it would be unwarranted to argue that 4-yearolds have incomplete knowledge of the linguistic properties of some and, more importantly, it would be unwarranted to argue that children’s interpretation of some in negative sentences is dictated by syntactic scope. The interpretive properties of the lexical items some and any are subject to cross-linguistic variation, and it would not be surprising if child language turned out to differ from the target adult language in this respect. However, it would be surprising if semantic scope in child language turned out to be securely tied to syntactic scope, since most (if not all) natural languages make use of covert operations that affect syntactic scope (for similar remarks see Musolino and Lidz, 2002b). The present findings also highlight children’s pragmatic competence in that they reveal children’s sensitivity to the conditions under which a negative sentence is ordinarily used. As a consequence, the findings also bear on much current research on children’s pragmatic competence which focuses on children’s interpretation of scalar terms like or (see Gualmini (2001) for a review and an alternative view). Further research is needed to determine how the

286 present findings might help in evaluating current models of children’s pragmatic knowledge. References Crain, Stephen and Cecile McKee 1985. “The acquisition of structural restrictions on anaphora.” Proceedings of NELS 15, 94-110. Crain, Stephen, Weijia Ni and Laura Conway 1994. “Learning, parsing and modularity.” In C. Clifton, L. Frazier, & K. Rayner (Eds.), Perspectives on sentence processing, 443-467. LEA, Hillsdale, NJ. Crain, Stephen and Rosalind Thornton 1998. Investigations in Universal Grammar. A Guide to Experiments on the Acquisition of and Semantics. The MIT Press, Cambridge, MA. Crain, Stephen and Ken Wexler 1999. “Methodology in the study of language acquisition: A modular approach.” In W. C. Ritchie and T. K. Bhatia (Eds.) ,Handbook of Language Acquisition, 387-425. Academic Press, San Diego. De Villiers, Jill and Helen Tager Flusberg 1975. “Some facts one simply cannot deny,” Journal of Child Language, 2, 279-286. Givon, Talmy 1978. “Negation in Language: Pragmatics, Function, Ontology.” In Peter Cole (ed.) Syntax and Semantics 9: Pragmatics, 69-112. Academic Press, New York. Grimshaw, Jane, and Sara T. Rosen 1990. “Knowledge and Obedience. The developmental status of the binding theory.” Linguistic Inquiry, 21, 187-222. Gualmini, Andrea 2001. “The Unbearable Lightness of Scalar Implicatures.” Doctoral Research Paper, University of Maryland at College Park. Gualmini, Andrea 2003. (in press). “Children Don’t Lack Some Knowledge.” University of Maryland Working Papers in Linguistics, 12, University of Maryland at College Park. Horn, Laurence 1989. A Natural History of Negation. University of Chicago Press, Chicago. Krämer, Irene 2000. Interpreting Indefinites. An Experimental Study of Children’s Language comprehension. PhD Dissertation, MPI Series. Ponsen & Looijen, Wageningen. Musolino, Julien 1998. Universal Grammar and the Acquisition of Semantic Knowledge: an Experimental Investigation into the Acquisition of Quantifier-Negation Interaction in English. Unpublished Ph.D. Dissertation, University of Maryland at College Park. Musolino, Julien 2000. “Universal quantification and the competence/performance distinction.” Paper presented at the 25th Boston University Conference on Language Development. Musolino, Julien 2001. “Structure and Meaning in the Acquisition of Scope.” Ms. Indiana University. Musolino, Julien, Stephen Crain and Rosalind Thornton 2000. “Navigating negative quantificational space.” Linguistics, 38, 1-32. Musolino, Julien and Jeffrey Lidz 2002a. “Preschool Logic: Truth and Felicity in the Acquisition of Quantification.” Proceedings of the 26th Boston University Conference on Language Development, 406-416. Musolino, Julien and Jeffrey Lidz 2002b. “Why children aren’t universally successful with quantification.” Ms., Indiana University and Northwestern University.

286

287 Reinhart, Tanya 1997. "Quantifier-Scope: “How labor is divided between QR and choice functions.” Linguistics and Philosophy, 20, 335-397. Thornton, Rosalind 1994. “Children’s Negative Questions: a Production/Comprehension Asymmetry.” Proceedings of ESCOL, 306-317. Wason, P. C. 1965. “The context of plausible denial,” Journal of Verbal Learning and Verbal Behaviour, 4, 7-11. Wason, P. 1972. “In Real Life Negatives Are False,” Logique et Analyse, 15, 17-38.