Springe direkt zu Inhalt

DLL | Gender comprehension in Heritage Spanish speakers: effects of lexical knowledge and vowel production

26.05.2026 | 16:00 c.t. - 18:00
DLL Laura Colantoni

DLL Laura Colantoni

Laura Colantoni (University of Toronto

(reporting on collaborative work with Ana T. Pérez-Leroux, Danielle Thomas, & Crystal Chen)

In this talk, I present the results of a project investigating the interaction between phonetic variability and vocabulary knowledge in the comprehension of Spanish grammatical gender. We analyze word-final unstressed Spanish vowels /a, o/ produced by early and late Spanish–English bilinguals to determine the extent to which variability in vowel production due to contact with English impacts the learning and representation of Spanish gender, as proposed by the Modular Interaction Hypothesis [1, 2]. Spanish gender is quasi-transparent, with over 60% of nouns ending in /o/ and /a/ aligning with masculine and feminine gender, respectively [3]. These vowels are known to merge and centralize in heritage Spanish–English speakers
[4, 5]. Moreover, in some Spanish varieties (e.g., Mexican and Andean), word-final vowels may devoice following voiceless obstruents [6, 7]. Thus, contact-induced phonetic variability and variety-specific phonetic features may conspire to obscure the acoustic input, making phonology-based gender attribution [8] more difficult for heritage speakers. If variability in
vowel production predicts gender accuracy, speakers who merge vowels in production should diverge in gender performance.
Adult Heritage Speakers of Spanish (HSs; n = 18; residing in North America before age 11) and adult Spanish immigrants (ASIs; n = 19; who moved to Canada after age 14) completed two tasks. Production data were collected through a picture-naming task with 244 target words ending in unstressed /a/ or /o/. Aperiodic (non-analyzable) vowels were identified, and F1–F2 values were extracted at the vowel midpoint from voiced tokens; Pillai scores were then computed to quantify vowel overlap [9–12]. Gender comprehension was assessed using an eye-tracking task implemented with a Tobii Pro Fusion system at 120 Hz. Participants viewed three images accompanied by an unspliced auditory prompt. Stimuli were controlled for gender, animacy, noun morphology, and determiner type (definite vs. possessive), and were counterbalanced across participants. Analyses focused on the proportion of gaze fixations to the target image within a time window spanning the word ending plus 300 ms. We concentrate on gender-transparent nouns preceded by possesive determiners (e.g., mi pala ‘my shovel’ vs. mi palo ‘my stick’), where the final vowel provides the only gender cue.
Results show that although the proportion of devoiced vowels was higher in the HS group, between-group differences were not statistically significant. Pillai scores revealed substantial within-group variability, with most HS participants exhibiting partial or complete overlap of the /a/ and /o/ vowel spaces (i.e., Pillai scores below 0.5). Eye-tracking results revealed no main effect of group, but a significant group × alternation interaction, indicating that HSs fixated less on vowel-alternating nouns compared to ASIs. Due to the wide range of individual variation observed in both tasks, correlations between vowel overlap measures and eye-tracking outcomes were not significant. Ongoing analyses explore the role of a third factor—lexical performance. Although these initial results do not fully support the Modular Interaction Hypothesis, the substantial within-group variability observed among HSs suggests that highly variable phonetic input, combined with limited formal education in the heritage language, may affect the acquisition and processing of grammatical categories.

Zeit & Ort

26.05.2026 | 16:00 c.t. - 18:00

L115 Seminarzentrum
(Otto-von-Simson-Straße 26 / Mensa)