Normal Language Acquisition
The aims of the lab's research on normal language acquisition are 1) to investigate the nature of linguistic representations in children and adults, and 2) ) to delineate those aspects of language that are, on some level, innate from those that are not. I am particularly interested in the acquisition of functional categories as it relates to the acquisition of grammar (morphology and syntax). Languages that have been studied in the lab include English, German, Turkish, Armenian, Russian, Hebrew, and Mandarin.
In much of this work, we use longitudinal transcripts of normal children's spontaneous speech to refine and test acquisitional theories. In this research, we document the patterns of acquisition (the order in which constructions are acquired, the aspects of morphosyntax which are acquired simultaneously, etc.) and the types of errors children do and do not make. Examples of transcript-based work includes research that demonstrates that children innately distinguish between lexical categories and homologous functional categories (i.e., distinguishing between nouns and pronouns, lexical verbs and auxiliary verbs, adjectives and determiners, etc.), but not between semantic subtypes of lexical categories such as action verbs and experiential verbs (Stromswold, 1990a, 1994a, to appear-b, under review-a). This finding suggests that children may be predisposed to detect divisions among words that are based on formal properties but not divisions that are based on semantic properties of words. Together with results from adult sentence processing, speech errors, aphasia research, and functional neuroimaging, these results suggest that the lexical/functional distinction is a core distinction in language (Stromswold, 1994a). In other research, we have shown that children always obey certain aspects of tense-marking, whereas they systematically violate other aspects (Stromswold, 1990ab). For example, once children have acquired tense markers (e.g., -ed and auxiliaries), they tense questions exactly once (i.e., they do not say what does Kermit eats? or what Kermit eat?). However, they sometimes tense the lexical verb rather than use do-support (i.e., they say both what does Kermit eat? and *what Kermit eats?). These results suggest that the "tense once" requirement is a reflection of a basic (universal) linguistic principle whereas do-support is a recently devised and idiosyncratic 'hack' that English uses to get around English's competing requirements that verbs must raise in questions and negatives but only auxiliary verbs can raise. This finding is consistent with Grimshaw's (1997) recent work in Optimality Theory.
In crosslinguistic work (Stromswold, 1996d, to appear-a, Stromswold & Zimmermann, 1998, 1999), we have found that the distinction between sentential negation (e.g., I do not like fish) and anaphoric negation (e.g., No, I like fish) is a core distinction that children obey from the earliest stages of acquisition. Contrary to the claims of a number of researchers, these results suggest that children do not go through an early stage during which they produced negative-initial sentential negatives. These results also suggest that, although all languages allow for sentential negation, different languages do so in different ways (Zimmermann & Stromswold, under review).
Research on the acquisition of subject-auxiliary inversion (SAI) indicates that, contrary to Bellugi's (1968) oft-reported finding that children go through a stage during which they produce yes/no questions with SAI (e.g., can I eat?) but wh-questions that lack SAI (e.g., saying *what I can eat? rather than what can I eat?), such a stage does not exist for most children (Stromswold, 1990a, 1994c). Furthermore, SAI errors are almost exclusively restricted to adjunct questions (particularly why questions) and questions with negated auxiliaries (e.g., can't, won't, don't). It appears that the reason children fail to invert in adjunct questions reflects that they sometimes place adjunct wh-words in a position to the left of the Complementizer Phrase, a strategy that is not available for argument wh-words (Stromswold, to appear-a).
We have also tested and refined models of the acquisition and structure of syntactic constructions by studying the order in which children acquire complex constructions such as wh-questions (Stromswold, 1988a; 1995a), passives (Snyder & Stromswold, 1997; Stromswold, 1989d; Stromswold, Pinker & Kaplan, 1985a; Stromswold & Snyder, 1995), datives (Snyder & Stromswold, 1997; Stromswold, 1989d; Stromswold & Snyder, 1995), verb particle constructions (Snyder & Stromswold, 1997; Stromswold & Snyder, 1995), and small clause constructions (Snyder & Stromswold, 1997; Stromswold, 1989d; Stromswold & Snyder, 1995). Through this work, we have discovered that the order in which normal children acquire different types of questions, passives, datives, auxiliaries, negators, and tense-makers is remarkably uniform. Two variables (the age at which a child first uses multi-word utterances and a measure of overall rate of acquisition) appear to capture a great deal of the variance among children. If this holds true, this finding should prove to be a valuable research and clinical tool for studying acquisition in normal and impaired children.
In addition to transcript-based experiments, we use a number of other types of acquisitional experiments. We have developed a puppet game technique for obtaining grammaticality judgments from children as young as 3 years old. In this technique, which is a modification of Stephen Crain's truth-value judgment technique, children are taught to feed a dog puppet a bone if the dog speak correctly and a rock if the dog speaks incorrectly. By including sentences that differ syntactically but not semantically (e.g., what does Kermit eat? and *what Kermit eats?; do you know why Kermit can eat? and *do you know why can Kermit eat?), one can investigate children's grammars (Stromswold 1990a, 1990b; Stromswold & Batman-Ratyosyan, 1999, Batman-Ratyosyan & Stromswold, 1999).
We have also conducted imitation experiments (Stromswold & Batman-Ratyosyan, 1999, Batman-Ratyosyan & Stromswold, 1999) and act-out experiments (Stromswold, Pinker & Kaplan, 1985; Stromswold & Batman-Ratyosyan, under review). Lastly, we have used parent checklists and diaries to investigate lexical development (Batman-Ratyosyan & Stromswold, in preparation).
Some revelant papers
- Stromswold, K. 1995. The acquisition of subject and object wh-questions. Language Acquisition, 4, 5-48.
- Stromswold, K., & Snyder, W. 1995. Acquisition of datives, particles, and related constructions: Evidence for a parametricaccount. D. MacLaughlin & S. McEwen (eds.), Proceedings of the 19th Annual Boston University Conference on Language Development, Volume 2, 621-628. Somerville, MA: Cascadilla Press.
- Stromswold, K. 1996. Analyzing children's spontaneous speech. D. McDaniel, C. McKee, and H. Cairns, eds. Methods for assessing children's syntax, pp. 23-53. Cambridge, MA: MIT Press.
- Snyder, W., & Stromswold, K. 1997. The structure and acquisition of English dative constructions. Linguistic Inquiry 28, 281-317.
- Batman-Ratyosyan, N. & Stromswold, K. 1999. What Turkish acquisition tells us about underlying word order and scrambling. Proceedings of the 23th Annual University of Pennsylvania Linguistics Conference.
- Stromswold, K. & Zimmermann, K. 2000. The acquisition of nein and nicht and the VP-internal subject stage in German. Language Acquisition 8, 101-127.
- Stromswold, K. to appear. The acquisition of inversion and negation in English. Linguistic Inquiry.