The role of the native language in foreign accent learning

This is a response to John E. Booth's (Reference Booth2015) article, ‘The fossilization of non-current English pronunciation in German EFL teaching’, published in English Today. Booth makes a number of claims in his paper, but the focus here is on his main claim that German pronunciation of English is based on an archaic accent of British English.

The vowel of interest in Booth's paper is the near-low, front vowel /æ/. It is the vowel found in English words such as ‘pat’ and ‘bat’ (see http://dictionary.cambridge.org for a transcription and audio model for Standard British and Standard American pronunciation as this vowel is present in both varieties). In German, such English words are often (mis)pronounced as ‘pet’ and ‘bet’, in other words, with a higher open-mid, front vowel /ε/. Bohn and Flege (Reference Bohn and Flege1992: 155) noted that this substitution ‘is well known to the extent of being a popular stereotype’. Booth claims that the use of /ε/ for /æ/ is not due to native-language interference: ‘There is, arguably, no question of first language (L1) interference, since the short “a” is a common enough vowel in German, found in many basic words such as Katze (cat), Mann (man) [ … ]’ (Booth, Reference Booth2015: 16). Instead, his claim is that Germans base their English pronunciation on the prestige accent of English: Received Pronunciation (RP), from the World War II era circa 1944–1953, in which the pronunciation of Dan Chaucer, for example, is /dεn tʃɔːsə/.

The above claims are problematic. German does not have a /æ/ vowel, or a short ‘a’, as Booth claims. This was established in a study conducted by Bohn and Flege (Reference Bohn and Flege1992). Bohn and Flege sought to test a model of second language acquisition which predicts different patterns of performance for sounds that are similar in the native language (L1) and the second language (L2) (e.g., /i, ɪ, ε/ in English and German) and sounds that are absent in the L1 but present in the L2 (e.g., /æ/ for German learners of English). Briefly, the model posits that it is more difficult to acquire native-like production and perception of similar sounds than to acquire a new sound (see also Flege, Reference Flege, Heubner and Ferguson1991, Reference Flege and Strange1995). To explore this hypothesis, Bohn and Flege established that /i, ɪ, ε/ are indeed similar vowels in German and English whilst /æ/ is absent firstly in a review of the literature that they conducted on German and English vowels and secondly in an acoustic comparison of vowels in English and German in which native speakers of German and English produced words with the vowels of interest. Native German-speaking participants produced the words bieten, bitten, betten, batten and bäten containing /i, ɪ, ε, a, εː/ respectively in the carrier sentence ‘Ich sage ____’. It should be noted that ‘batten’ is not a real word in German, but it is phonotactically legal, similar to the existing word ‘hatten’. Native English (American) speaking participants produced the words beat, bit, bet, and bat containing /i, ɪ, ε, æ/ respectively in the carrier sentence ‘I will say ____’. Acoustic measures including vowel duration, fundamental frequency and frequency of the first three formants were measured for each vowel of interest. For the acoustic comparisons, Bohn and Flege established /i, ɪ, ε/ as similar vowels in German and English but found that ‘[t]he English vowel /æ/ [ … ] is a new sound for native Germans’ (p. 139). This contradicts Booth's assertion, that /æ/ (or short ‘a’ as Booth calls it) is in both English and German.

Bohn and Flege's study continued with a comparison of vowel production by native and non-native speakers of English for /i, ɪ, ε, æ/ using beat, bit, bet, and bat in the carrier sentence ‘I will say ____’ produced by native speakers of American English. They also had two groups of German speakers of English in which one group had spent considerable time in an English-speaking country (average of 7.5 years) and the other group had spent less time in an English-speaking country (average of less than 0.6 years). As would be expected for a phonemic contrast, they found two distinct vowel spaces for /æ/ and /ε/ for the native speakers of English in an acoustic analysis. For the less-experienced non-native speakers, there was an almost complete overlap of vowel space for /æ/ and /ε/. The more experienced non-native speakers showed a pattern closer to the native speakers, with a clearer separation of the vowels /æ/ and /ε/. Bohn and Flege used this evidence to support the hypothesis that with more exposure, production of new vowels can become more native-like. However, in a subsequent intelligibility experiment for vowel production in the same groups, Bohn and Flege found that English native speaker raters were at chance level or little better than chance for discrimination between bat and bet as produced by the experienced speakers of English (mean discrimination accuracy = 65.9%) and the less-experienced speakers of English (mean discrimination accuracy = 50.9%). It seems clear that German speakers conflate /æ/ and /ε/ in production in early stages of English acquisition, though with more experience they are able to produce two vowels that differ acoustically even if English native speakers cannot perceive the difference.

The source for the production problem is likely to be rooted in perceptual problems for the /æ/ - /ε/ contrast; see Bohn and Flege (Reference Bohn and Flege1990). In that study, Bohn and Flege asked German learners of English to listen to the words beat, bit, bet, and bat in the carrier sentence ‘I will say ____’ as spoken by two native speakers of American English. They were asked to circle the German word on an answer sheet that most closely represented the word they heard, e.g., biet, bitt, bett, bäht and bert or ‘N’ if none of the words adequately represented the word. For ‘bet’, the German participants circled ‘bett’ most often and rated their confidence as high for their decision. For ‘bat’, German participants circled ‘bert’ and ‘bäht’ most often, but their confidence ratings for these decisions were lower than for the ‘bet’ stimuli. The ‘N’ response, which would indicate that none of the choices was a match, was rarely circled. Here it is clear that Germans perceive a difference between the /æ/ in ‘bat’ and /ε/ in ‘bet’, but they perceive the new sound /æ/ as similar to some existing German vowel, though it is clear that they are unsure of what this vowel is, i.e., whether it is the vowel in ‘bert’ or ‘bäht’. This mirrors their production problems as shown by Bohn and Flege (Reference Bohn and Flege1992): Germans, at least those with more English experience, produced different vowels for /æ/ and /ε/, though these vowels differed acoustically from native production, and intelligibility ratings showed that native English speakers had difficulty perceiving the difference, indicating that for Germans, the acoustic and articulatory criteria for /æ/ and /ε/ are different than for native English speakers.

A number of studies have investigated other problematic sounds for German speakers of English, including not only the /æ/ vowel, but also the dental fricatives /θ/ and /ð/, which are absent in German. Hanulikova and Weber (Reference Hanulikova, Weber, Dziubalska–Kołaczyk, Wrembel and Kul2010) showed that Germans frequently substitute /s/ for voiceless /θ/ and in their acoustic exploration of productions of /s/ found that the /θ/-substitution /s/ differed from intended /s/, showing that German speakers of English produce phonetically different sounds for /θ/ and /s/, though they are perceptually similar sounds to native speakers of English, with both being perceived as /s/. German speakers of English also have difficulty with the /v/-/w/ contrast in production and perception (Ankerstein & Morschett, Reference Ankerstein and Morschett2013). For German learners, /w/ is a new yet unproblematic sound, but /v/ is problematic as it is phonetically different in German and English. Germans tend to produce the English /v/ as they would in German, i.e., with less friction, which is then perceived as an approximant /w/ by English native speakers (Hamann & Sennema, Reference Hamann and Sennema2005; Iverson et al., Reference Iverson, Ekanayake, Hamann, Sennema and Evans2008). The diphthongs /eɪ/ as in ‘bait’ and British /əʊ/ or American /oʊ/ as in ‘boat’ also differ from the German monophthong vowels /eː/ and /oː/, respectively (see Russ, Reference Russ2010). Germans commonly have trouble with the diphthongs in English, producing them instead as monophthongs as they would in German (Eckert & Barry, Reference Eckert and Barry2005). In these cases, the source of the difficulty of the acquisition of the sounds is the L1, which affects not only the production of the new and similar sounds, but also the perception of new and similar sounds.

The role of the L1 in perception and production of new and similar vowels is a far more parsimonious explanation of features of German accented English. It can not only account for difficulties with dental fricatives, the /v/-/w/ distinction and problems with diphthongal /eɪ/ and /əʊ/ or /oʊ/, but also the /æ/ vowel. The proposal that Germans aim for a fossilized accent cannot account for these other common segmental features of German accented English. If Germans use an archaic RP target for /æ/, why do they not do the same for /eɪ/ and /əʊ/? A far more probable and parsimonious explanation is that L2 pronunciation is affected by the L1.