Intervention effects in the acquisition of raising: Evidence from English and Spanish

ABSTRACT

The present study is designed to investigate whether children’s difficulties with subject-to-subject raising (StSR) are due to intervention effects. We examine English-speaking children’s comprehension of StSR with seem and Spanish-speaking children’s comprehension of StSR with parecer ‘seem,’ a configuration never before tested in this language. Spanish parecer is ambiguous between a functional verb, which does not select an experiencer argument, and a lexical verb, which requires an overt experiencer (e.g.. In the first part of this study, we consider the hypothesis that the experiencer argument of seem may induce intervention effects even when it is not overtly produced and find support for this claim—English-speaking children perform poorly on StSR both when the intervening experiencer is overtly expressed and when it is implicit; Spanish-speaking children, on the other hand, only perform poorly on StSR with lexical parecer but do well on StSR with functional parecer. These results are in line with intervention accounts. The second part of this study aims to investigate whether these intervention effects are rooted in children’s grammatical or processing deficits. Results from a verbal processing task suggest that for a group of children—those who perform at chance or above on the StSR task—comprehension of sentences with an intervening experiencer is modulated by processing capacity. However, for those who consistently obtain a non-adult-like interpretation of StSR, processing capacity does not positively correlate with their performance. We hypothesize that, for this group, the difficulty is instead grammar-based.

ACKNOWLEDGMENTS

I am greatly indebted to Nina Hyams, Misha Becker, Fernanda Ferreira, Jesse Harris, Jeffrey Lidz, Carson Schu#x308;tze, William Snyder, Dominique Sportiche, María Luisa Zubizarreta, as well as two anonymous reviewers for their valuable help and comments at various stages of this research. I would also like to thank the research assistants as well as the UCLA Language Acquisition Lab for help testing and recruiting. Finally, I am grateful to the day care center directors, parents, and children who participated in this study, and the support of UCLA and the National Science Foundation (any findings or conclusions are my own and may not reflect the views of the NSF). Earlier versions of this paper were presented at the 2016 LSA Meeting and BUCLD 41.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ Becker does not provide a breakdown of the results by verb.

² A number of arguments in favor of a maturational approach are outlined in Orfitelli (2012): The acquisition of other constructions that involve A-movement past an argument, such as nonactional passives, is observed across a wide range of languages at approximately the same age, between 6 and 7 (Hirsch & Wexler 2006); identical twins are better correlated in the age at which they comprehend verbal passives than fraternal twins (Ganger et al. 2005); and acquisition is not predicted by environmental factors such as parental socioeconomic status or level of education (Hirsch et al. 2006).

³ We found strong interspeaker variation with respect to the judgments provided in (5).

⁴ A reviewer asks whether adjectival seem (e.g., John seems nice) also projects an experiencer argument. There are considerations suggesting that adjectival seem, in contrast to verbal seem, may not involve movement. Following May (1977), Diesing (1992) shows that when taking a stage-level predicate, verbal seem allows both existential and generic readings of bare plural subjects; adjectival seem only allows the generic reading. She suggests this difference is due to structural differences: Namely, verbal seem involves raising, while adjectival seem involves a control configuration. If this is true, young children may be expected to produce adjectival seem early on but develop verbal seem much later. This has been confirmed by a corpus study investigating the spontaneous use of seem (Hirsch 2011). While this analysis could certainly use refinement, it suggests an interesting set of future investigations centered around children’s comprehension of raising with adjectival seem.

⁵ Although Gibson’s DLT makes no mention of implicit arguments, it is conceivable that the processor would also treat them as discourse objects that must be stored and accessed in memory. However, the acquisition hypothesis we are testing, PIE, predicts intervention only with overt DPs (Choe 2012: 36–37; Choe & Deen 2016:123).

⁶ Note that Gibson’s memory-based approach to the dependency-length effect is not at odds with other processing approaches. Any theory in which dependency construction is length sensitive would predict a contrast between the sentences in (6) as well (e.g., Hawkins 1994; McElree 2000; McElree et al. 2003; O’Grady 1997; O’Grady et al. 2000; O’Grady et al. 2003).

⁷ See Cinque (2004) and Haegeman (2006) for a similar analysis of Italian sembrare.

⁸ F-parecer passes the standard diagnostics for raising: (i) the possibility of having a subject of a sentential idiom in matrix subject position without losing its idiomatic reading; (ii) the impossibility of embedding “parecer XP” under causatives also suggests that raising has taken place (Aissen 1974); (iii) scope ambiguities involving raised quantificational subjects. This is not surprising if we assume that modals are raising verbs (see Wurmbrand 1999, 2001; Wurmbrand & Bobaljik 1999).

⁹ See Mateu (2016) for a more in-depth discussion of these differences.

¹⁰ At the time of the corpus study, there were 2,800,324 adult utterances in English and 270,411 in Spanish.

¹¹ The data values regarding children’s use of the unraised/raised construction with overt experiencers are too small to be able to reject the null hypothesis.

¹² A reviewer appropriately asks whether the fact that Spanish has the ser/estar ‘be’ opposition, but English does not, could account for their better performance. Specifically, because ser gris ‘be gray’ is unambiguously interpreted as an individual-level predicate, but ‘be gray’ could be interpreted as a stage-level or individual-level predicate (despite the addition of “definitely”), this could have aided the Spanish-speaking children. However, this hypothesis cannot account for English-speaking children’s adult-like performance with “unraised seem,” which was followed by the same “ambiguous” predicate.

¹³ Under the Similarity-based Interference processing account (Lewis & Vasishth 2005; van Dyke & Lewis 2003; Gordon, Hendrick & Levine, 2002), a distractor would be a noun that shares features with the filler noun.

¹⁴ Note that Spanish-speaking children’s performance on F-parecer with a verbal complement (M = 5.5/6) was not any worse than that of F-parecer with an adjectival complement (M = 5.31/6). This constitutes potential counterevidence for the idea that English-speaking children’s worse performance on raised seem (followed by a verbal complement) is attributable to the difference in the complement (AdjP, vP, or TP).

¹⁵ Note that Hirsch (2011), Hirsch et al. (2008), and Orfitelli (2012) also found poor performance on StSR seem with a covert experiencer and that in none of these studies were the same children presented with both covert and overt experiencer sentences. However, a salient experiencer was always present either in the story or as the commenting puppet.

Additional information

Funding

This work was supported by the UCLA Dissertation Year Fellowship; National Science Foundation [BCS-1451589].