Contextual variation in the acoustic and perceptual similarity of North German and American English vowels

Abstract

Strange et al. [J. Acoust. Soc. Am. 115, 1791-1807 (2004)] reported that North German (NG) front-rounded vowels in hVp syllables were acoustically intermediate between front and back American English (AE) vowels. However, AE listeners perceptually assimilated them as poor exemplars of back AE vowels. In this study, speaker- and context-independent cross-language discriminant analyses of NG and AE vowels produced in CVC syllables (C=labial, alveolar, velar stops) in sentences showed that NG front-rounded vowels fell within AE back-vowel distributions, due to the "fronting" of AE back vowels in alveolar/velar contexts. NG [I, e, epsilon, inverted c] were located relatively "higher" in acoustic vowel space than their AE counterparts and varied in cross-language similarity across consonantal contexts. In a perceptual assimilation task, naive listeners classified NG vowels in terms of native AE categories and rated their goodness on a 7-point scale (very foreign to very English sounding). Both front- and back-rounded NG vowels were perceptually assimilated overwhelmingly to back AE categories and judged equally good exemplars. Perceptual assimilation patterns did not vary with context, and were not always predictable from acoustic similarity. These findings suggest that listeners adopt a context-independent strategy when judging the cross-language similarity of vowels produced and presented in continuous speech contexts.

The perception of English front vowels by North Holland and Flemish listeners: Acoustic similarity predicts and explains cross-linguistic and L2 perception

“…These predictions fitted well with the observed data ( Table 5). As such, our results are not in line with those of Strange et al (2004Strange et al ( , 2005, who have argued that acoustic similarity is not always a good predictor of cross-language speech perception. By contrast, they confirm the predictions made by the L2LP model (Escudero & Boersma, 2004;Escudero, 2005Escudero, , 2009), according to which the acoustic similarity between L1 and L2 sounds explains the patterns found in non-native vowel perception.…”

“…This indicates that, in contrast to previous findings (Strange et al, 2004(Strange et al, , 2005, the acoustic (and hence auditory) properties are important in determining similarity across languages. It is worth mentioning that our LDA models only included a subset of the vowel inventories of English and Dutch, while Strange et al (2004Strange et al ( , 2005 included whole inventories, which could account for the higher predictive success in our model. However, Gilichinskaya and Strange (2010), who used the eight monophthongs of New Yorkian English, report that the results of their LDA predicted Russian listener's classifications for seven out of the eight American English vowels.…”

“…Following previous studies on the effect of acoustic similarity on non-native vowel perception (Gilichinskaya & Strange, 2010;Strange et al, 2004Strange et al, , 2005, we will use Linear Discriminant Analysis to examine whether dialectal acoustic differences predict different perceptual assimilation patterns.…”

“…We followed the procedure reported in Strange et al (2004Strange et al ( , 2005 and Gilichinskaya and Strange (2010). Two different native discriminant analysis models, one for North Holland and one for Flemish, 4 were trained on the F1, F2, F3 (in Bark) and duration values of the Dutch vowels /i, e, a7 and >/.…”

“…They examined the phonetic similarity between the first three formants of North German (NG) and American English (AE) vowels using linear discriminant analysis (Klecka, 1980) and found that the perceptual assimilation could not always be predicted from acoustic similarity. However, a more recent study by Gilichinskaya and Strange (2010), who used the same discriminant analysis technique used in Strange et al (2004, Strange, Bohn, Nishi, andTrent, 2005), showed that acoustic similarity was a good predictor of cross-language assimilation patterns of American English vowels by Russian listeners, especially for point vowels.…”