Restricted (Penn State Only)
Steuck, Jonathan William
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
February 26, 2018
Committee Members:
  • Rena Torres Cacoullos, Dissertation Advisor
  • Rena Torres Cacoullos, Committee Chair
  • Karen Lynn Miller, Committee Member
  • Michael Travis Putnam, Committee Member
  • Melinda Fricke, Outside Member
  • codeswitching
  • prosody
  • syntax
  • variation
  • bilingualism
  • Intonation Unit
  • prosodic-syntactic interface
  • linguistics
  • sociolinguistics
  • cognitive psychology
  • functional linguistics
  • usage-based linguistics
  • New Mexican Spanish
  • Spanish in the US
  • corpus linguistics
  • Spanish
  • English
  • Prosodic sentence
  • Benefits of bilingualism
  • community norms
  • discourse analysis
  • variationist comparative method
  • prosody-syntax interface
  • Spanish of New Mexico
  • language in context
  • variationist linguistics
  • constraints of code-switching
This study examines the prosody of spontaneous code-switching, broadly defined as the alternation of languages in the same conversation. While recent findings from phonetic (e.g. Fricke, Kroll, & Dussias, 2016) and psycholinguistic (e.g. Tamargo, Kroff, & Dussias, 2016) research suggest that bilinguals may utilize certain features to anticipate an upcoming language switch, studies of prosodic patterns in production have been lacking. A large sample of multi-word code-switches (MWCS; n = 407) comprised of at least two words in English and Spanish is taken from the New Mexico Spanish-English Bilingual corpus (Torres Cacoullos & Travis, in prep), a collection of sociolinguistic interviews from members of a speech community where code-switching is a discourse mode (i.e. a community norm). The data are transcribed in Intonation Units (IUs) (see Du Bois, Schuetze-Coburn, Cummings, & Paolino, 1993), which enables the analysis of the linguistic properties of MWCS at the interface of prosody and syntax. This study utilizes a novel unit for the study of code-switching: The prosodic sentence (PS), defined using intonation (see Chafe, 1994). The PS is the basis for characterizing intra-sentential MWCS according to prosodic-syntactic factors: the prosodic position of MWCS (across-IUs versus IU-internally within a single IU), pause expression (pauses expressed at the beginning/end of an IU versus IU-internally), transitional continuity (the intonation contour of the IU preceding a target IU), length measures (number of IUs, words, seconds), and switch direction (which captures the language a speaker code-switches into within the PS). To provide accountability in analysis, bilingual PSs containing MWCS (n = 323) are systematically compared with two sets of PSs. The first is constituted by otherwise unilingual PSs containing a noun of other-language origin (“LOLI-nouns”; Spanish-origin (n = 78), English-origin (n = 216)). The second is the benchmark of entirely unilingual PSs (n = 584) with a lexical Spanish (n = 231) or English (n = 238) noun (“nouns”) (n total nouns = 763). Overall, clear prosodic-syntactic properties of spontaneous MWCS emerge. First, an analysis of prosodic position indicates a strong speaker preference for code-switching across prosodic boundaries (i.e. IU boundaries) rather than IU-internally. However, when it comes to LOLI-nouns, the preferred prosodic position—at the very beginning or end of the IU rather than in IU-medial position—is mirrored in the prosodic position of nouns in unilingual PSs. Second, with respect to pause expression, more unfilled pauses are produced at the beginning of IUs containing an IU-internal MWCS, whereas pauses preceding MWCS across-IUs tend to be filled pauses such as uh and eh. The same asymmetry between unfilled and filled pauses is observed within IUs immediately preceding an IU-internal MWCS (although IU-internal pauses are infrequent compared to IU-initial pauses). While the rate and type of pause expression varies according to the prosodic position of MWCS, helping to differentiate between each prosodic position, pause expression does not vary to a similar degree in the unilingual PSs. Thus, it appears that when code-switching, speakers tend to prosodically separate the two languages through pauses, regardless of whether the code-switch is positioned across-IUs or IU-internally. Third, truncation was examined as a possible indicator of relatively heavier cognitive load, as a test of the hypothesis that code-switching is cognitively costly. On the one hand, IU-internal MWCS are found to be more often preceded by a truncated IU than are MWCS across-IUs, which may indicate that IU-internal MWCS is more demanding than across-IU MWCS. However, contrary to the hypothesis, MWCS occur on average no more after truncation than continuing intonation as compared to unilingual IUs in the same PS, indicating that code-switching is no less fluid than speech produced in just one language. Fourth, a first-time quantification of various length measures of bilingual and unilingual PSs reveals differences, although the greater length of bilingual PSs may to some extent reflect the operational definition of bilingual PSs as containing at least two contiguous words in each language. Finally, when switch direction is considered, the prosodic-syntactic factors examined vary according to the language a speaker code-switches into. In contrast, the prosodic features of unilingual PSs in English versus Spanish do not differ for these New Mexican bilingual speakers. This result provides additional evidence for particular prosodic patterns in MWCS. Together, the findings highlight the unique prosodic-syntactic signature of MWCS as compared to unilingual stretches where MWCS are absent, part of the norms for combining two languages in bilingual communities.