There are many factors influencing the intelligibility of speech for non-native listeners. One of which is called the Interlanguage Speech Intelligibility Benefit (ISIB) suggested by Bent and Bradlow [1]. They conducted an experiment asking native Chinese, Korean and English listeners and listeners with other L1 backgrounds to perform sentence recognition task, using recordings of native Chinese, Korean and English talkers reading simple English materials. They found that for non-native listeners, the talkers who shared with them the same native language were more intelligible than or at least as intelligible as native English talkers. Bent and Bradlow attributed the ISIB to the overall shared phonetic and phonological knowledge between non-native talkers and the non-native listeners. For example, when a non-native talker produced vowels of his/her L1 (e.g. Chinese) as substitutes for vowels of L2 (e.g. English), native English listeners might be misled while non-native listeners (Chinese) would be more likely to understand the intended vowels given the shared phonological and phonetic knowledge of the Mandarin sound system. Many studies have investigated the ISIB with mixed results. For example, Hayes-Harb et al [2] asked Mandarin listeners and native English listeners to perform in a word identification task, using minimal pairs in English with final-voicing contrasts as material. They found ISIB for Mandarin listeners, especially those with limited English proficiency. They also attributed ISIB to the shared phonological system between Mandarin listeners and talkers. · On the contrary, Munro et al [3] asked Cantonese, Japanese, Mandarin and English listeners to rate the English speech produced by Cantonese, Japanese, Polish and Spanish speakers in terms of intelligibility and accentedness. They found ISIB for Japanese listeners only, but not for Cantonese listeners. Similarly, using lecture recordings in English spoken by speakers of Mandarin, Japanese, Spanish and American English, Major et al [4] found ISIB for Spanish listeners only, but not for Japanese and Mandarin listeners. The discrepant results suggest that the mechanisms behind ISIB are complex and thus in need of further exploration. Specifically, if the presence of ISIB is attributable to the shared phonological system between the talkers and the listeners, then ISIB should be extendable to anyone who shares the same knowledge, i.e., bilingual L2 learners as well. If Mandarin listeners could understand Mandarin talkers speaking English better than native English listeners do because of the shared phonological knowledge, then native English listeners who have learned Mandarin as a second language (i.e., sharing the same Mandarin knowledge) should also understand Mandarin talkers better than monolingual native English listeners do as well. However, so far, there is no study investigating this possibility. The current study is a preliminary investigation to examine whether ISIB found for Mandarin listeners can also be extended to English-Mandarin bilingual L2 learners. Mandarin-accented English is characterized by these features: 1) devoicing of word-final stops; 2) insufficient contrast between /i:/ and //; and 3) non-distinction between // and /
/ [5, 6, 7]. We compared listeners’ identification of minimal pairs in English involving these contrasts to evaluate the ISIB for Mandarin listeners and EnglishMandarin bilingual L2 learners. 2. METHOD 2.1. Subjects Three listener groups participated in this study: 9 native Mandarin speakers (NM); 9 native English speakers who have learned Mandarin as a second language (BI) and 9 native monolingual English speakers (NE). The 9 female NM listeners had a mean age of 21. They came from various regions in China and spoke Mandarin as their first language. They were students at CUHK. The 9 BI listeners (5 females, 4 males) with a mean age of 24 were recruited from advanced courses in Mandarin at the Yale-China Chinese Language Center at CUHK. 7 out of the 9 BI listeners were exchange students from the USA. The 9 NE listeners (6 females, 3 males) with a mean age of 25 were recruited through word of mouth in Hong Kong. One was from the UK, one was from Australia, two were from Canada, and 5 were from the USA. None of them understood Mandarin. All listeners reported no history of speech or hearing problems. They volunteered to participate in this study. 2.2. Materials A list of 36 English monosyllabic words in CVC structure was used: deep, dip, peace, piss, sheep, ship, beat, bit, dead, dad, pet, pat, bet, bat, bed, bad, pick, pig, peck, peg, back, bag, buck, bug, cop, cob, cap, cab, cup, cub, rip, rib, bid, bud, but, bead. These 36 words form 20 minimal pairs. There are four minimal pairs for each of the five contrasts: vowels: /i:/ – //, // – /
/ and word-final stops: /p/-/b/, /t/-/d/, /k/-/g/. Seven female native Mandarin speakers produced the 36 target words in a short carrier phrase three times. Because of time limit, only materials from three female speakers were used for the listening experiment. These three speakers were chosen because of the heavy Mandarin accent in their English. One token of each target word was taken as the materials for the word identification experiment, resulting in 120 word tokens in total (20 word pairs × 2 words × 3 speakers). The target words were excised from the carrier phrase, i.e., only the target words would be played to the listeners for identification. 2.3. Procedure The subjects listened to the recordings in a semi-random order over a headphone in a quiet room. Care was taken to ensure that no minimal pair would appear in a row. For each recording, the subjects were asked to choose from two given words on an answer sheet (e.g. cob, cop). They were allowed to listen to the recordings repeatedly if necessary, and most listeners took advantage of that. In addition, they were asked to give a confidence rating ranging from 0 (no confidence) to 7 (full confidence) for each answer. Upon completion, the subjects were asked to fill in a questionnaire about their language backgrounds.