L2 Production of American English Vowels in Function Words by Spanish L1 Speakers

Abstract

The present study investigates the production of American English (hereafter English) vowels in both function and lexical words by Spanish native speakers (n=62) and English native speakers (n=30) in recordings from the Speech Accent Archive (Weinberger & Kunath, 2011). Vowels contained in English function words are often reduced (Jurafsky et al., 1998), while those in Spanish are less so (Hualde, 2005). Thus, the Spanish speaker learning English must learn not only a new vowel inventory but also the process of reducing these vowels in function words. The majority of research on L2 speech has focused on vowels in lexical items, a literature gap that the current study addresses. The recordings were analyzed acoustically in Praat (Boersma & Weenink, 2016), where the first two formant values of each vowel token were extracted from the vowel midpoint. Formant values were normalized with the “Nearey Intrinsic” method in phonTools (Barreda, 2015).  As age of onset of acquisition (AOA) is known to affect L2 pronunciation, the Spanish L1 speakers were split into two groups based on AOA. Both learner groups’ productions were compared to those of native speakers with a MANOVA run for each vowel, with normalized formant values as the dependent variables, and L1 as the independent variable. Pillai scores from the MANOVAs were used as a measure of vowel overlap (Hay et al., 2006). Pairwise comparisons were also run between vowels within groups to determine the extent of vowel neutralization in function words. In function words, the low AOA group patterned closely with the native speaker group, showing complete neutralization of some vowels and partial neutralization of others. The high AOA group showed partial neutralization of vowels in function words, which did not happen in lexical items. Results are discussed in the context of L2 speech learning theories and general phonetic theory.