Topics in L inguistics (2017), 18(1), pp. 17 - 31 10.1515/topling - 2017 - 0002 Vowel reduction patterns of early Spanish - English bilinguals rece iving continuous L1 and L2 input Emily Byers Florida International University, United States of America Indiana University, United States of America Abstract This study investigates the production of three morphophonetic variations of schwa in American English: the plural allomorph { - s} as in watch e s , the possessive allomorph { - s} as in Sash a ’s , and word - finally as in Russi a. T he production of these three allomorphs were examined in Miami’ s English monolingual and early Spanish - English bilingual populations . Our purpose was to determine how native - like early Spanish - Engl ish bilinguals’ sp ectral qualities and reduced vowel durations were compared to Miami English monolinguals during a reading task. Results indicate that e arly bilinguals’ reduced vowels followed the same overall pattern as monolinguals, but had different acoustic properties. Key words b ilingualism, phonetics, sociolinguistics, individual differences, vowel reduction, American English Introduction years in a multilingual environment can This paper examines schwa production in maintain cat egorical distinctions between monolingual and Spanish - English bilingual schwa in the plural and possessive speakers living in Miami, Florida (United morphemes as well as word - finally States) across three phonological (Flemm ing and Johnson, 2007). If Miami environm ents where vowel qualities are English monolinguals exhibit signs of categorically differentiated : before the blurring these phonetically discrete plural morpheme { - s} as in wa tch e s categories (Byers, 2012; Flemming, 2009; [w ɑ ʧə z] ; before the possessive morpheme Fl emmin g and Johnson, 2007), t his feature { - s} as in Sash a ’s [s ɑ ʃə z] ; and word - finally suggests that one feature of “ Miami as in Russi a [r ʌʃə ] (Flemming and Johnson, English ” is targetles sness with regard to 2007). Presently, linguists are exploring vowel reduction 1 (Carter and Lynch, 2015; whether the unique linguistic environment Enzinna, 2015). While other dialects of in Miami, Florida , where speakers’ native English may maintain or blur schwa values language (L1) is prevalent and shares due to other factors (including speaking prestige with L2 English in many domains, rate, nasalization etc.), this feature should has given rise to a new American accent be considered one feature of an emerging that, while Spanish - influenced, contains accent among other co - occurring features. phonetic features that make the accent Our second research question asks identifiable as “Miami English .” Our whether early sequential Spanish - English primary research question adds to the bilinguals tend to categ orically contemporary body of phonetic studies dif ferentiate vowel qualities of schwa that are cataloguing the features of Miami English by exploring whether monolingual 1 See Kondo, 1994 for a description of targetless English speakers who have lived many schwa. 17 Topics in L inguistics (2017), 18(1), pp. 17 - 31 along the same phonetic parameters as word - final position, as often occurs in Miami - based monolinguals. If early nativized loanwords (e.g. “ yog a ” [jog ə ] ). Spanish - English bilinguals do approximate (Fokes and Bond, 1993; Fourakis, 1991; English monolingual phonetic values of Hammond, 1997) . schwa , how stable are these categories ? To compare Sp anish - English bilingual 1.2 Phonetics of schwa and English monolingual vowel qualities, In some instances, the acoustic phonetic we collected a mini - corpus of speech production of schwa results in centralized samples from Eng lish monolinguals who articulations with some degree of have lived in Miami for a considerable reliability along a spectral and/or temporal period of time . This corpus was collected dimension. Spectral qualities correspond to have phonetic measurements of the to dark bands of energy found in the ambient English dialect. We were acoustic output of vowels (Peterson & interested to learn whether Miami English Barney, 1952). In acoustic phonetics, these monolinguals would differentiate phonetic energy bands are known as “ formants ” variations of schwa along the same pattern (Boersma, 2002; Lennes, 2011). The as monolingual spea kers of American formant frequencies most relevant for English , who have limited contact with vowel analysis are fundamental frequency Spanish - accented English . We then directly (F0), vowel height (F1), anterior/posterior compare d early sequential Spanish - English dimension of the vowel (F2), and overall bilinguals’ spectral and temporal vowel position of the vowel in the oral cav ity (F2 - qualities of schwa to those of Miami F1) (Barlow et al., 2013; Erickson, 2002). English monolingual values. We aim to Based on phonological descriptions of contribute to the ongoing debate on h ow schwa, one would expect formant values subtle phonetic variations are in fluenced to approximate the centralized values by continuous L1 & L2 input over the codified by Olive et al. in their acoustic lifespan (Flege, 2012). analysis of American Eng lish (1993), where The remainder of the introduction is 500/ 1500/2500 Hz are the “ typical ” F1, structured as follows: Section 1 .1 outlines F2, and F3 spe ctral qualities of schwa for a the role of vowel reduction as a prototypical adult male (Johnson, 2003). phonological feature of the American - However, a complete neutralization of English accent . Sec ond, sec tion 1. 2 the phonological features [+/ - back] and identifies quantifiable phonetic variables [+/ - high] is seldom evident at the of schwa . Lastly, s ection 1.3 review s the phonetic level (Bro wman and Goldstein, limited existing literature regarding early 1992) . Where coarticulatory effects from Spanish - English bilinguals’ vowel neighboring consonants are observed, reduction patterns when speaking L2 schwa may be produced as little more than English . a burst of air serving the purpose of preserving the prosody of running speech 1 . 1 Vowel reduction in American English (Shockey, 2008). C o articulatory effec ts due Vowel reducti on is a crucial component for to surrounding consonants have also been achieving native - like stress patterns in shown to heavily influence both the American English (Flege and Bohn, 1989 ; spectral and temporal characteristics of Gut, 2007 ) . From a phonological schwa (Browman a nd Goldstein, 1992; perspective, v owel r educ tion is a p rocess Byers, 2012; Flemming, 2009; Flemming where by unstressed vowels undergo and Johnson, 2007; Gahl et al., 2012; neutralization of multiple feature c ontrasts Kondo, 1994; Van Be rgam, 1994 ) . For to form a central vowel ( Chomsky and example, schwa has been observed to be Halle, 1968; Burzio, 2007; Flemming, higher and more fronted before labial 2009). C andidates for vowel reduc tion in consonants and more posterior before American English occur most often when velar consonants (Stevens and Ho use, they are adjacent to the tonic syllable. For 1964; Kondo, 1994). C oarticulatory effects example, th e pre - tonic vowel in are also not unique to American English, “ mit o chondria ” [ma ɪɾ ə k ɑ ́nd ɹ i ə ] is reduced as s tudies have found evidence of to a schwa. Likewise, the post - tonic neighbo u ring consonants altering schwa syllable in “ photography ” [f ə t ɑ ́g ɹ ə fi] is values in Swedish (Lindblom, 1963) and reduced, or centralized. A third German (Hertrich and Ackermann, 1995). environment where phonological vowel Prior research has demonstrated that reduction occurs in American English is in schwa can also be susceptible to vowel 18 Topics in L inguistics (2017), 18(1), pp. 17 - 31 harmony effects – particularly wi th regard formants schwa in the plural morpheme in to movement observed on the anterior - “ roses. ” Discrete differences in vowel posterior plane corresponding to F2 height between schwas in possessi ve and frequencies. Back vowels /u, o, ɔ / in plural morphemes ha ve led some to preceding syllables predictably move suggest narrower transcription be schwa posteriorly from the central vowel implemented to differentiate schwa space. Similarly, full front vowels in pronunciations. For instance, Flemming neighbo u ring syllables have been shown and Johnson (2007) proposed that schwa to create more anterior schwas in adjacent in the plural morpheme be transcribed (or syllables (Van Bergam, 1994; Fowler and at least regarded) as high er [ ɨ ], while the Brancazio , 2000 ). possessive morpheme should remain More nuanced observations regarding transcribed as [ ə ] . To reit erate, these possible coarticulatory effects on schwa differences can be heard in words that are have indicated that schwa may be transcribed identically but are phonetically “ targetless ” for on e (or more) acoustic discriminable, as in the example “ Rosa ’ s dimension s (Kondo, 1994; Van Bergam, ros es. ” 1994; Van Oostendorp, 1995; Flemming, Another phonological environment 2009). That is, schwa may be centralized where unstressed vowels are reliably along either height or anterior/posterior centralized is word - final position . While dimension s, but centralization along one there are perceivable differences in word - dimension does not i mply centralizat ion final schwa compared to schwa in the along another. plural and possessive morphemes , Prior observations of “ targetlessnes s ” minimal pairs exist that demonst rate for spectral qualities of schwa have certain words have centralized