Erik Arnold and Laura Catharine Smith, Germanic and Slavic Languages
The role of accent in second language (L2) production and perception has been acknowledged by researchers, but generally only in the sense of limiting their subjects to a specific dialect region to avoid any dialect influence on their results. Some studies have, however, investigated the role of first language accent in L2 production, i.e., pronunciation (Smith & Baker, 2010; Baker & Smith, 2010; O’Brien & Smith, 2010). These studies demonstrate that first language (L1) accent can indeed affect L2 production. The role of the accent to which learners are exposed has also been shown to impact L2 production (Smith, 2010; Escudero & Boersma, 2004). While Smith and several others have looked into the role of first language accent in L2 production, little research has explicitly examined the role of L1 accent in the perception of L2 sounds. As part of a larger project on the effect of L1 accent on L2 sound acquisition, this study seeks to fill the gap on perceptual research while also addressing the theoretical impact of accent on L2 perception in general. In this study, we examine how a native English speaker’s accent (e.g., Scottish, Northern Irish, London, American English (AE), etc.) impacts his or her perception (i.e., identification) of German vowels.
To answer these questions, we are administering an identification survey delivered online via Qualtrics. Those from the UK who were recorded by Dr. Smith during March-April 2014 as part of the pronunciation phase of the study are the primary subjects who will complete the identification task. Native speakers of American English at BYU who have learned German are also being tested Fall 2014 to expand the number of varieties examined by this study. Before taking the survey, subjects must complete the consent form and fill out a background survey collecting information about their first language accent as well as details regarding their German language learning experience (e.g., years of study, time spent abroad, etc.). Next, subjects hear German words presented via headphones and for each word they hear, they see a list of words on the computer screen containing the 15 German monophthongs (vowels such as [i] in feel, [ɪ] in fit, and [e] in bake). They then select which word from the list they think they heard. Questions are randomized and appear one at a time on the screen. The survey consists of 120 tokens (15 vowels x 4 environments x 2 speakers). The four environments the vowels appear in are b_te, b_ne, b_re, and b_le. We expect the consonants following the vowels to influence vowel production. In fact, these consonants were selected because /r/ and /l/ are particularly troublesome markers of dialect in some of the UK varieties. In this way, we can see not only if vowels are difficult to perceive, but if speakers from some dialect areas have more problems perceiving the vowels in the /l/-/r/ environments than speakers from other dialect areas. For instance, Scots English uses a “dark-l” as in ball in every word position, and some Southern English varieties lack “r” after vowels in many positions. Tokens used in the perceptual task were recorded by a male and female native speaker of Standard German. Responses to the perceptual tasks will be tabulated and saved to an Excel spreadsheet in terms of both accuracy and patterns of responses. This will allow us to identify pattern of errors in misidentification for the various dialects. These results will then be analyzed using statistical tests such as ANOVAs and mixed model analyses to determine the effect of dialect, self-reported strength of accent, experience with the language, and time in the target country, among other factors.
The general accuracy rates were calculated for each vowel and each environment. Unfortunately, only students from Scotland responded to the survey, and not as many students responded as expected. So accuracy rates only represent Scottish speakers. Accuracy rates were highest for the vowels /i/ and /ɛ/. Accuracy rates were lowest for /Y/ and /u/. The rates were highest in the b_ne environment and lowest in the b_re environment. There were a few significant patterns of misidentification. The /e/ vowel was commonly misidentified for /ɛ:/. /ø:/ was commonly misidentified for /y/ and /Y/ was misidentified for /oe/. Other trends were that long vowels were misidentified for long vowels and short vowels were misidentified for short vowels.
The top identified vowels by each of the four non-native groups were vowels found in the English vowel inventory e.g., /i:/, /ɛ/, /a:/, and /a/. This suggests that learners are able to perceive vowels they are familiar with rather “new” sounds found in the L2 inventory. The results of the accuracy rates of the vowels indicate that the German rounded vowels are more difficult to perceive than the unrounded vowels. Such a finding suggests that German learners should give special emphasis on their ability to pronounce and perceive rounded vowels near the beginning of their learning, especially with the short rounded vowels /ʏ/ and /oe/. The biggest limitation to this study is the lack of participants. It is difficult to compare cross dialectally if there are not enough participants from different dialects. It is unclear why the b_re environment is more difficult for learners to perceive, but this has been seen in several previous studies.
This project contributes to the growing research in second language acquisition as it investigated the role of L1 dialect in learning L2 sounds and common patterns of misidentification. These findings will also help second language teachers better understand the specific problems learners from various regions have perceiving L2 sounds.