Date of Degree

9-2018

Document Type

Dissertation

Degree Name

Ph.D.

Program

Linguistics

Advisor

Gita Martohardjono

Committee Members

Richard G. Schwartz

Janet D. Fodor

Subject Categories

Linguistics | Psycholinguistics and Neurolinguistics

Keywords

heritage language, bilingualism, sentence processing, cross-linguistic priming, cross-linguistic influence

Abstract

This study examines real-time heritage language syntactic processing and tests the hypothesis that some commonly observed properties of heritage languages—apparent instability in grammatical knowledge and divergence from monolingual grammatical norms—can be attributed to cross-linguistic influence from the socially dominant language during online processing. To test this hypothesis, a novel cross-linguistic structural priming experiment based on self-paced listening was conducted with a group of heritage Spanish speakers and late Spanish-English bilinguals to test whether exposure to preposition stranding in English—a feature of core syntax that does not exist in Spanish—could facilitate processing of (ungrammatical) preposition stranding in a subsequently encountered Spanish sentence. Results were subjected to group-level and individual differences analyses with mixed-effects modeling to determine whether any measurable priming effects were influenced by individual differences in exposure, use, and proficiency for Spanish and English.

The results indicate that exposure to preposition stranding in English primed the comprehension of structurally-parallel, but illicit, Spanish sentences for some heritage speakers and all late bilinguals. Heritage speakers who had lower fluency in Spanish than English did not show a priming effect, but they processed Spanish preposition-stranded sentences the fastest and gave higher acceptability ratings, suggesting that preposition stranding may be a feature of their Spanish. An analysis of within-language cumulative priming also revealed that repeated exposure to Spanish preposition-stranded sentences facilitated processing for some late bilinguals. No cumulative priming effect was found among heritage speakers and, for both groups, repeated exposure to Spanish preposition-stranded sentences did not modulate the cross-linguistic priming effect. These findings suggest that while some late bilinguals implicitly learned to process Spanish preposition stranding over the course of the experiment (Loebell & Bock, 2003), the cross-linguistic priming effect is most compatible with the structural priming account based on residual activation of abstract syntactic representations shared between Spanish and English (Hartsuiker, Pickering, & Veltkamp, 2004).

An analysis of baseline syntactic processing for a separate set of complex grammatical Spanish sentences also showed that heritage speakers and late bilinguals processed these sentences similarly. Like the processing results for Spanish preposition-stranded sentences, heritage speakers showed a processing advantage over late bilinguals. Heritage speakers who were more fluent in Spanish than English also patterned like late bilinguals (showing slower response times) and some late bilinguals but no heritage speakers showed evidence of cumulative priming. Acceptability judgment results also aligned with baseline syntactic processing and cross-linguistic priming results in that heritage speakers who had higher fluency in Spanish than English patterned like late bilinguals. While these results suggest that Spanish preposition stranding may not be entirely ungrammatical for heritage speakers who have significantly higher fluency in English than Spanish, heritage speakers who had higher fluency in Spanish than English were qualitatively identical to late bilinguals in terms of syntactic processing for grammatical and ungrammatical complex sentences, cross-linguistic priming, and grammatical representations.

Taken together, the results of the present study suggest that relative fluency in the heritage and dominant language is the most important predictor of heritage language syntactic processing and grammatical representations. These results also provide some of the first evidence that core syntactic processing in the first-learned language is susceptible to influence from a later-learned language for simultaneous, early sequential, and late bilinguals. Together, the findings of this study lend empirical support to the central concept of Putnam and Sánchez’s (2013) model of heritage language grammar and provide indirect evidence that some heritage language characteristics that are regularly observed in heritage language studies using offline measures may be due in part to real-time influence from the dominant language during heritage language processing.

This work is embargoed and will be available for download on Monday, September 30, 2019

Graduate Center users:
To read this work, log in to your GC ILL account and place a thesis request.

Non-GC Users:
See the GC’s lending policies to learn more.

Share

COinS