Priming prepositional-phrase attachment ambiguities in English and Spanish-speaking children and adults during language comprehension

Abstract

To better understand language acquisition, researchers have explored the nature and development of children’s mental representations of language. However, to this day, children’s language development is still unclear. Thus, this thesis is aimed at answering the research question: what is the development and nature of children’s syntactic representations and – given that language experience plays an important role in language acquisition (Matthews, Lieven, Theankston, & Tomasello, 2005; Abbot-Smith, Lieven, & Tomasello, 2001) – how are they affected by language experience? To address this question, we used a structural priming paradigm, an experimental method that has been informative about the development and nature of children’s language representations (including specific syntactic structures), and the role of language experience. Structural priming occurs when processing a prime sentence facilitates processing a subsequent target sentence with the same structure. Priming effects in the absence of verb repetition between prime and target (i.e., lexically-independent) have been interpreted as evidence of abstract syntactic representations. Priming effects may be enhanced in the presence of verb repetition (i.e., lexically-dependent; the lexical boost) (Pickering & Branigan, 1998), which has been interpreted as evidence for a lexically-specific component in syntactic representations.

Until now, most developmental structural priming research has focussed on studying a limited number of structures during language production in English-speaking children. This is not ideal, as researchers need to understand: (1) whether current theories of language development, and specifically the role of language experience in development, hold across languages (2) if the mechanisms involved in structural priming in language production are the same as in language comprehension; and (3) cross-linguistic differences in how particular syntactic structures are acquired.

This thesis therefore uses a structural priming paradigm to study the influence of recent language experience on the high-attached (HA) and low-attached (LA) analyses of prepositional phrases (PPs) in globally ambiguous sentences during language comprehension (e.g., The girl is touching the dog with the banana) in 4- to 6-year-old (Younger group) and 8- to 10-year-old (Older group) English- and Spanish-speaking children and adults. PP attachment ambiguities have been successfully used to understand language comprehension in children and adults (e.g., Snedeker & Trueswell, 2004), and have been shown to have different cross-linguistic preferences in adults (e.g., Carreiras & Clifton Jr., 1999), therefore allowing direct cross-linguistic comparisons of PP attachment preferences in the absence and presence of a prime in children at earlier and later stages of development, and adults.

Thus, to answer our main research questions about the nature and development of children’s syntactic representations we investigate: (a) How does PP attachment develop over time during language comprehension? (b) How does recent experience of syntactic repetition with and without verb repetition affect PP attachment preferences in children and adults? and (c) Do children and adults show effects of cumulative experience?

Studies 1 and 2 used a child-appropriate web-based sentence-picture matching task to investigate the comprehension of PP preferences. We measured whether children chose a picture corresponding to the LA versus HA analysis when they heard a sentence with an ambiguous PP attachment, in the absence of a prime (i.e., baseline condition) or after hearing/matching a prime sentence that forced either an LA or HA analysis and that had the same or a different verb as the target sentence. We investigated whether English- and Spanish-speaking children showed existing and stable abstract representations of the HA and LA analyses when processing globally ambiguous PP sentences. We compared PP attachment preferences during language comprehension between 4- to 6-year-olds and 8- to 10-year-olds, as well as between 4- to 10-year-old children and adults. To better understand how children’s language representations were strengthened by experience, we also examined whether the individual language experiences with the HA and LA analyses accumulated within one session in children and adults.

In the baseline condition, English-speaking participants showed an overall preference for HA, however, children showed a stronger HA preference than adults. The Older group showed cumulative effects of experience, by showing an increase in their HA responses as the experiment progressed.

In the presence of a prime, English-speaking participants showed reliable sensitivity to recent language experience. At a session level, they showed a decrease in their overall HA preferences when they were exposed to HA and LA primes, compared to when they were not (i.e., baseline). At a trial level, they also showed reliable lexically-dependent priming effects, but not lexically-independent priming effects. This suggested that English-speaking children’s abstract representations may be more fragile than the adults’ and that more power may be needed to detect reliable lexically-independent priming effects. Although participants did not show evidence of a lexical boost, verb repetition between prime and target pairs facilitated the processing of the HA and LA analyses, as suggested by previous studies (Pickering & Branigan, 1998). Participants did not show significant cumulative effects of experience.

In the baseline condition, Spanish-speaking participants showed an overall preference for HA, however, children showed a stronger HA preference than adults. In the presence of a prime, Spanish-speaking children and adults showed sensitivity to recent language experience at a session level by showing an overall decrease in their HA responses when they were exposed to HA and LA primes, compared to when they were not (i.e., baseline). At a trial level, and surprisingly, they showed reliable priming effects when there was no verb repetition between prime and target pairs, but not when prime and target pairs involved the same verb (hence, there was no lexical boost). The Older group showed cumulative effects of experience only when the verb was repeated between prime and target sentences, by showing a decrease in their HA responses as the experiment progressed.

Overall, our findings with English- and Spanish-speaking children are more compatible with early abstraction accounts of syntactic development, which argue that lexically-dependent priming effects should not differ between children and adults.

Study 3 investigated cross-linguistic differences in the representations of the HA and LA analyses between English- and Spanish-speaking participants. In the baseline condition, English- and Spanish-speaking children and adults processed globally ambiguous PP sentences in a similar way. However, in the presence of a prime, only the Older children showed a significant difference in the overall PP preferences by language. This developmentally mediated cross-linguistic difference suggested that, when exposed to LA and HA structures within a session, by the age of 8, it is easier for the English-speaking children to access the LA interpretation for globally ambiguous PP sentences than their Spanish-speaking counterparts. However, by the time English- and Spanish-speaking children reach adulthood, their interpretation of globally ambiguous PP sentences is influenced by recent language experience to the same extent.

Study 4 focussed on answering the question about whether syntactic repetition affected the processing of temporarily ambiguous sentences during language comprehension. We specifically examined whether the processing of an unambiguous prime sentence could eliminate the garden-path effect or facilitate revision mechanisms in temporarily ambiguous target sentences (e.g., Put the duck on the sock on the star) in English-speaking adults. Our findings showed that prior processing of an unambiguous prime facilitated garden-path recovery in the ambiguous target sentences.

Taken together, our findings provide novel insight into the nature and development of the language representations of English- and Spanish-speaking children and adults. We have provided evidence that English- and Spanish-speaking children and adults showed reliable sensitivity to recent language experience during language comprehension and that their representations of the LA analysis were not entirely bound to lexically specific items but showed some level of abstraction. However, we did not find the same patterns of structural priming effects in English- and Spanish-speaking children. This suggests that, to better understand language development, future research should continue to study cross-linguistic differences in structural priming paradigms in children. We also found that when exposed to LA and HA structures within a session, the overall PP preferences may differ between English- and Spanish-speaking 8- to 10-year-olds, However, when native speakers of English and Spanish reach adulthood, recent language experience influences their interpretation of globally ambiguous PP sentences to the same extent.