The Phonological Basis of Foreign Accent,' a Hypothesis*

• TESOl QUARTERLY Vol. 15. No. A December 1981

The Phonological Basis of Foreign Accent,' A Hypothesis*

James Emil Flege

Foreign accent is often thought to be the result of an age-related diminution in the ability to learn to pronounce languages. Existing studies of L2 pronunciation, however, do not seem to support the claim that there is some fundamental difference between children and adults in phonetic learning ability. The continued presence of foreign accent may instead be a consequence of the establishment of stable phonologi• cal representations for sounds and words in the native language. Lan• guage learners who perceive sounds in the target language to be phonologically identical to native-language sounds (despite possible phonetic differences between the two languages) may base whatever phonetic learning that does occur during the acquisition process on an acoustic model provided by pairs of similar sounds in two languages, rather than on a single language-specific acoustic model as in first-lan• guage acquisition. Thus an adult or child learner of a foreign language may retain the same kind of phonetic learning ability evident in early childhood and yet still speak with an accent because phonological translation provides a two-language source of phonetic input that may ultimately limit progress in learning to pronounce a foreign language.

It has been observed that adults learning a foreign language "never seem capable of ridding themselves entirely of foreign accent" (Scovel 1969:245) whereas most normally developing children "learn to recognize and pronounce the sounds of ... their speech community so well that ... their speech lacks any trace of the foreign accent of people who learn the language later" (Ferguson and Garnica 1975: 154). The observation that children and adults differ fundamentally in their ability to learn to pro• nounce a foreign language has led to the formulation of a critical period hypothesis. The continued preselice of foreign accent in the speech of post• pubescent language learners has often been specifically linked to neuro• physiological maturation and the establishment of cerebral lateralization for language functions (Penfield and Roberts 1966, Lenneberg 1967, Scovel 1969). According to a critical period hypothesis for second language learn• ing, adults usually cannot learn to speak a foreign language without accent because the central nervous system undergoes some permanent reorganiza- • The author would like to thank Roy Koenigsknecht for conmlents on a previous version of this paper. This work was supported by NINCDS grant NS07IOO to the Department of Communicative Disorders at Northwestern University. Mr. Flege is NIH Post-doctoral Fellow in the Department of Communicative Dis• orders at Northwestern University. He has published in Language and Sp'!ech and Lang/lage Leaming on second language acquisition. 443 444 TESOL Quarterly tion at ahout puberty, therehy distinguishing adults and adolescents frorn younger language learners. In a recent phonetic study of English pronunciation hy non-native' speakers, Flege (1980) noted a direct inAuence of phonetic characteristics of Arahic on the English stops produced by Saudi Arabians. Even the more experienced of two adult speaker groups (.3 years resiuence in America) continued to produce stop consonants much as if they were Arabic sounds. There was, however, some evidence of phonetic learning. The more ex• perienced Saudis produced a durational contrast between word-nnal Ip-b/, It-d/, and Ik-gj. However, the magnitude of the durational contrast pro• duced by the Saudis was much smaller than the one produced by Ameri• cans, seemingly because of the absence of a similar duration contrast be• tween word-final stops in the Saudis' native language. ~;[oreover, it was noted that experienced Saudi speakers of English more nearly approximated native English pronunciation than did newly arrived Saudis. For both groups, however, it seemed that phonetic implementation of the stop voic• ing contrast was characteristic neither of Arabic nor of English. Instead, values for the phonetic parameters fell between those measured for stops in Arabic and stops in English. Findings such as this raise two questions about how adults learn a second language. First, why did the Saudis not simply use Arabic sounds when producing English words if, in fact, they had passed a critical period for learning to pronounce foreign languages? Second, since the Saudis did modify their pronunciation of English stops, why wasn't their modification complete? In other words, why did the Saudis' production of English stops seem to represent a compromise between the phonetic characteristics of Arabic and those of English? It is unlikely that small phonetic differences such as those noted between the Saudis and Americans would in themselves be sufficient to cause the perception of segmental substitutions by American listeners. Yet such pho• netic differences are of interest because they may contribute to what is perceived as a foreign accent. ~loreo\'er, an understanding of factors that tend to limit the extent of progress in foreign language pronunciation at any level of analysis may offer insight into why language learners often seem to show a distinct limit on the extent to which L2 pronunciation ordinarily improves (Selinker 1972. Nemser 1971).

In this article "Oe shall consider foreign lanCTuacreo ~ pronunciation from several perspectives. First, we will discuss some of the dimensions that may form the acoustic basis of foreiCTn.::> accent. Second, we will review evidence concerning the claim that children and adults differ fundamentally in terms of phonetic learning ability. And. third. we will propose an alternative to the critical period h~'pothesis to account for the continued presence of a foreign accent. The phonological translation hypothesis proposed here starts Foreign Accent 445

\rith the assumption that neither physiolol.!ical maturation nor neurological fcprg:;U1izationrenders an adult incapable of speaking a foreign language \\'ithout an accent.

1. Foreign Accent Perception of a foreign accent derives from differences in pronunciation of a language hy nati\"e and non-native speakers. The most readily apparent b;lSis for a foreign accent are mispronunciations that lead to the perception of ;1 segmental sound substitution, such as in French-accented I sink so or .\r;lbic-accented I put my car in the barking lot. One recent study indicated th,lt the frequency with which segmental substitutions were noted in short excerpts of speech produced by non-native speakers was highly correlated \rith native-speaker judgments of accentedness (Brennan, Ryan, and Daw• son 1975). However, this does not necessarily mean that perception of a foreign .1ccent is hased just on overtly detectable mispronunciations of sound seg• ments. Listeners are more likely to base a judgment of foreign accent on some combination of segmental, subsegmental, and suprasegmental differ• ences which distinguish the speech of native from that of non-native speak• ers. Some mispronunciations which depart from target-language phonetic norms may merely sound distorted, such as the underaspirated jptkj that can often be heard in the English produced by native speakers of Romance languages. An important question is whether every audible acoustic differ• enC'e is weighted equally when a foreign accent is perceived. One study (Flege and Hammond 1980) suggests that at least not all segmental sound substitutions produce a similar effect. \lonolingual English speakers were asked to try to read English sentences with a Spanish accent. Their imita• tions of Spanish-accented Eng1ish included a number of the sound substitu• tions one might associate with a stereotypic Spanish accent. Some sound substitutions were produced much more frequently than others, suggesting that they may figure prominently in what Americans perceive as a Spanish .\ccent in English.

2. Cross-language phonetic differences Children not only learn to produce sounds intelligibly in their native language, they also learn to produce them according to the language• specific phonetic norms of the surrounding speech community. The mono• lingual child's attainment of accent-free speech in the course of normal speech development is by no means a trivial achievement, for human languages manifest considerable variation at the phonetic level. Language• specific phonetic characteristics mmt be acquired by the child if sjhe is to sound like a native speaker. At the same time, the many phonetic differ• ences which distinguish h\"o languages represent a potential source of foreign accent in the speech of a second language learner since interference 446 TESOL Quarterly during second language acquisition appears to be prominent at the level of phonetic implementation (Flege and Port 1981). Some phonetic differences between languages may he localized at tlw level of phonetic segments. The fact that phonetically similar sounds ill two languages might be transcribed ,,'ith the same IPA symbol should not ohscure the fact that sounds may- be realized differently at the phonetic: level. For example, the Danish vo\ ••.el Iii is higher than its phonologica: counterpart in English. Both Hallsa and h":alabari, to take :mother example, have a voiced bilabial imposive in the phonetic surface hut the same sound is produced with a very different kind of voicing in the two lan• guages (Ladefoged 1980) . Vowels preceding a nasal consonant may be nasalized to a greater extent in some languages than others (Clumeck 1976) . Phonetic differences between languages that may potentially contribute to accentedness may be suprasegmental. For example, average speaking fundamental frequency may differ cross-linguistica!1y (Majewski, Hollein. and Zalewski 1972) as may the fundamental frequency variations which underlie similar intonation contours in two language (vViI1ems 1978). Cross-language differences may also be thought of as subsegmentaL such as the speech timing differences between languages which together affect the perceived rhythmic qualities of speech and may carry over from the native to the target language. For example, the extent to which vowels are lengthened at the end of an utterance may be much greater in some languages than others (Delattre 1966, Oller 1977). In languages where the voicing characteristic of a consonant influences the duration of a preceding vowel, the relative magnitude of this effect appears to differ substantial1y across languages (Chen 1970). Even the phonetic domain within which successive phonetic segments have a temporal influence on one another seems to vary across languages. In English, for example, a consonant affects only the duration of a preceding vowel, while in Japanese the• duration of a consonant seems to affect the duration of both preceding as well as following vowels (Port, AI-Ani, and Maeda 1980).

3. Child vs. adult phonetic learning Considering the many phonetic differences that might distinguish the learner's native language from a foreign language, one might wonder if any learner of a second language-child or adult-will manage to pro• nounce a foreign language completely without accent. At this point, Wt simply don't know if a speah:r can be bilingual at the phonetic level, or it native-like pronunciation at the level of phonetic implementation is evel. necessary for accent-free speech. If children are more capable than adult. •. of learning to correctly pronounce foreign languages, they will of courSt' be more likely than adults to escape the onus often associated with ac- r Foreign Accent 447

ccntedness.1 Existing empirical ('vidence, however, does not seem to sup• port the notion that there is a true qualitative difference in how well children as opposed to adults pronounce foreign languages, A foreign ac• cent can sometimes be detected in the speech of children (Vallette 196-1. .-\.sher and Garcia 1969). At the same time, some adults appear capable of producing a foreign language without accent (\Villiam 1980, Ncufeld 19S0 ) . Several studies have investigated how English native speakers rate the English produced by individuals who began learning English as a foreign language at various ages. Some such studies seem to indicate that the speech of someone who begins learning a foreign language at an early age will be perceived as less accented than that of someone who begins learning at a relatively later age (Asher and Garcia 1969, Fathman 1975, Oyama 1976). But at least one study indicates that older children and adults may be, at least initially, more successful than young children in pronounc• ing a foreign language (Snow and Hoefnagel-Hohle 1978). Still other studies indicate that older children and adults can imitate words in an unfamiliar foreign language better than young children, and that ability to discriminate unfamiliar sounds in a foreign language-seemingly a prerequisite for the development of correct pronunciation-may actually improve rather than diminish with age (Politzer and Weiss 1969, Snow and Hoefnagel-Hahle 1977, \,yinitz 1981). Finally, it appears that pronuncia• tion of a foreign language by both adolescents and adults will improve with additional exposure (Asher and Garcia 1969, Snow and Hoefnagel• Hahle 1977, 1978), at least within limits (Oyama 1976). Thus at present there does not appear to be evidence of a fundamental difference between children and adults in their ability to learn to pronounce a second language. There will admittedly be instances where a child's accent appears to diminish more rapidly or thoroughly than an adult's. However, such disparities might be due to differences in linguistic input (Burling 1981), cultural expectations (Hill 1970), or to the period of l In several experiments, listeners have been asked to evaluate the personality of

aevaluational speaker after experiments, listening to a short single samples talker has of speech. produced Unknown a single to speech the listeners sample in twice, such once in an accented guise and another time with no accent (Anisfeld, Bogo, and Lam• bert 1962, Arthur, Farrar. and Bradford 1974). In other experiments, speech samples produced by larger numbers of both native and non-native speakers have been evaluated (Ryan and Carranza 1975). The consistent finding of such studies seems to be that a person whose speech is accented will be rated less favorably (along subjective rating scales such as intelligent vs. stupid or kind 'Is. cruel) than speakers who talk without a foreign accent. Other studies have established the validity of using such rating scales to measure subjecth'e reactions to a complex phenomenon like accented•

ness.tude Measures estimation, of sensory accentedness modality resulting matching) from psychophysical show a high scaling correlation techniques with measures (magni• derived from rating scales. Such studies indicate that not only are native speakers with• out special phonetic training quite capable of distinguishing native from non-native speakers, but also that they can reliably distinguish between degrees of accentedness (Brennan, Ryan, and Dawson 1975, Ryan, Carranza, and l\Ioffie 1977) d. Giles 19(2), Unfortunately, for most foreign language learners, the greater the degree of perceived accent, the more unfavorably a speaker will be evaluated. 448 TESOL Quarterly time over which pronuTlciation continues to show ill1pro\TnH.'nt for learners of different ages (SIlOW and I-Ioefuagel-H()hle 1977) rather than IH.'uwphysi_ olngical matllfatioll or reorganization.

4. Simultaneous VS. successive learnin~

Perhaps the best way to learn two languages without accent is to learn both in early childhood. There is some tentative evidence from siIH~le sub• ject observation to suggest that young children who acquire two languages simultaneously hefore about age 3;0 may produce both languages with a nath'e-Iike accent (~rcLaughIin 1978), at least insofar as judged by t:1e standards of articulation that adults apply to child speech. For example, a two-year-old exposed to English and German produced the word ball (Ball) differently in those two languages. In English she said [ba U] and in German [baI], thereby approximately the dark word-final [1] of English and the light (1) of German (Leopold 1947). Another t\vo-year-old, who was learning English and Portuguese, produced voiceless stops with aspira• tion when speaking English but without aspiration in Portuguese (Major 1977). Given the scarcity of empirical evidence, it would be premature to assume this is typical of early childhood bilingualism. But if such a pattern is characteristic, it might provide some insight into why it seems young bilingual children are more apt than others to pronounce hoth languages with native-like accents. Perhaps such children learn the sounds occurring in the phonetic surface of their two languages independently, much as if each sound were a separate phoneme in an enriched, pan-language system.

5. The phonological translation hypothesis

It may perhaps be more useful to ask why anyone should speak a foreign language with an accent than to ponder the source of possible differences in degree of accent between children and adults. One possible approach to this issue would be to determine why children acquiring a second lan• guage before about age 3;0 may be more likely than older learners to speak the second language without accent. One possibility is that there is a critical period for phonetic learning and that it occurs long before puberty (Krashen 1973). If so, age rather than simultaneity of learning is the crucial factor. However, a more probable cause, in my estimation, is the important difference in phonological develop• ment between the young child who simultaneously learns two languages and the older child or adult who begins learning a second language after the establishment of the first. One important difference Illay be that only relatively mature speakers are apt to interpret sounds in a second language in terms of sounds found in another language. I would like to propose that a tendency by mature speakers to interpret sounds occuning in a foreign language in terms of sounds found in their native language may he a more important cause of foreign accent than Foreign Accent 449 allY limitation on phonetic learning imposed by neurophysiological matura• tion (Lenneherg 1967). It has been observed that before about the age of three hilingual children often do not fully realize they are speakin!! two different languages (McLaughlin 197f)). a confusion which is most reaclily ,Ipparent from language mixing at the lexical level (Redlinger and Park 19RO). ~Ioreover, young pre-literate children may not yet be capahle of reliably identifying phone-sized units in speech (~'lorais, Gary, Alegria, :111(1 Bertelson 1979). Somewhat older children, however, clearly do distin• guish between languages and seem to be aware of the similarity of sounds found in two languages (see Tervoort 1979). For example, a young American boy speaking French used English words spoken with a French ,Iccent when he didn't know the French equivalent (Valette 1964). English• speaking children have been observed trying to "speak" Spanish by produc• ing English words with a Spanish accent (Hernandez-Chavez, cited in Ervin- Tripp 1974). American children exposed to French for only a month showed in an informal experiment that they were able to produce English words with French sounds (Ervin-Tripp 1974). This kind of linguistic be• havior suggests that even children are capable of phonological translation (Catford 1965) between the sounds of two languages, a factor that may potentially influence their production of sounds in both languages. The phonological translation hypothesis can be illustrated by consider• ing how French native speakers produce English words. VOT (voice onset time) is a well known acoustic measure of the laryngeal timing differen~es which may serve to distinguish a voiceless stop (like Ip/) from its voiced homorganic cognate Ubi). A VOT difference between voiced and voice• less stops can be observed in the speech of English-learning children as early as age two (Macken and Barton 1980). Figure 1 displays VOT values reported by Caramazza for stops in isolated English words produced by monolingual and bilingual speakers of English and French. Children acquiring English as a first language will eventually learn to produce stops in a similar phonetic context with VOT values like those seen here for adult English monolinguals in Figure la (about 60-90 msec). Children learning French, in contrast, will eventually produce stops with the rela• tively short VOT values (0-30 msec) typical of French, such as those seen here for French Canadian adults in Figure 1d. Since corresponding French and English stops al'e realized differently at the level of phonetic implementation, how will a French Canadian produce English stops? Previous research and common experience tell us that the English produced by French Canadians is likely to be accented. :\Ioreover, there is reason to think that subsegmental phonetic differences such as the VOT difference between English and French stops may be detected by listeners experiencing a foreign accent (Flege and Hammond 1981). Thus we might expect the VOT values in English \vords produced 450 TESOL Quarterly

FIGURE 1

~fl'an \'OT in msec of'ptk . in English and French words produced by (A) IIHmolinL!ual EnL!lish speakers; (B) French Canadian speakers of En.dish: (C) hilinL!lIal French ClI1adians: (D) monolingual French Canadialls: and (E) Illonlllill!.!lIal French speakers ill Franct'. w .- I Ipl~--- ~ IB u PLACE OF ARTICULATIONI 0 b ENGLISH WORDS / 80 r [~

This ngure is based on data reported by Caramazza, Yeni-Komshian, Zurif, and Carbone 1973, and Caramazza and Yeni-Komshian 197-1.

by French Canadians to either match or resemble the VOT values mea• sured in phonologically similar French words. In fact, as seen in Figure lb, it appears that French Canadian speakers produce English words with VOT values between those typical of English and French. Two aspects of these data are important to the present discussion. First, the French Canadians produced English words with shorter VOT values than monolingual English speakers. And second, although the En• glish stops' of French Canadians were not completely English-like, they were nonetheless different from the stops produced in French \\lords by these same speakers. These data can be interpreted in two radically different ways. One can infer that the French Canadians, who began learning English at about six, did not possess the same phonetic learning ability as somewhat younger children acquiring English as a first language. But one can instead interpret these data to mean that the French Canadians did not differ from young Foreign Accent 451 children in terms of phonetic learning ability. The French-Canadian hilinguals studied by Caralllazza et a!. may have produced English stops with smaller VOT values than the English monolingual precisely hecause the capacity of older children and adults for l{>arning to pronounce ]an• g:uages remains essentially unchanged from early childhood. It seems reasonable to think that the French-speaking child who is exposed to English will consider English /p/, for example, to be the same as the Ip/ found in French words despite VOT and other acoustic differences between French and English stops. Phonologists since Trubetz• koy have recognized that a listener interprets sounds in a foreign language in terms of phonological categories of the native language (see Lisker 1978). One important consequence of phonological translation may be that a French-speaking child (or adult) learning English as n second language will base production of English Ipl on the acoustic phonetic model pro• vided by voiceless bilabial stops occurring in both English and French words. Recall that the French Canadians pronounced Ipl differently in En• glish and French. If we assume that the acoustic model upon which they based production of English was just the acoustic substance of sounds in English words, we must necessarily conclude that their phonetic learning was only partially successful. This would of course imply that these lan• guage learners differed in some important way from somewhat younger children acquiring English as a first language. But if, on the other hand, the French Canadians based their production of English words on a com• posite of corresponding sounds (and other nonsegmental phonetic dimen• sions) occurring in both their languages, one might reasonably conclude that their phonetic learning was completely successful. The VOT of English stops produced by the French Canadians was intermediate between monolingual French and English values. This seem• ing compromise between French and English patterns of phonetic im• plementation may reflect a restructuring of the phonetic space so that it encompasses two languages (\,yilliams 1980). Such a convergence can be described in Piagetian terms as the accommodation of existing French patterns of stop production to the English pattern, together with an assimilation of the English pattern of stop production to the French Canadians' French phonetic pattern.

6. Testing the hypothesis It has been postulated here that children and adults possess the same general capability for learning to pronounce foreign languages and t~at. one important cause of foreign accent is phonological translation between languages by speakers who already speak a first language. According to the phonological translation hypothesis, an individual may be completely successful in his/her phonetic learning of a second language and yet still 452 TESOL Quarterly retain an accent hecall.';(, prol1l1nci:ltioll of tht' foreign Iallgtla~e is based on pairs of corr('spondin~ sounds (or non-segmental phonetic dimensions) found in the native and target language. The hqJothesis rf'sts on the as• sumption that hoth childrell and adult lan~ua~e l('al'l1('rs modify native• language patterns of phonetic implementation. and that superordinate acoustic models based on pairs of corresponding sounds or phonetic dimen• sions in two lan~l1ages serve as input for phonetic learning in second lan• gllage acquisition. This hypothesis seems to be consistent with the results of existing phonetic studies showing that the values of phonetic parameters measured in the speech of second language learners are often intermediate to those typical of monolingual speakers of the native and target language (Pinkerton 1972, Suomi 1976, Niemi 1979. Flege 1980. Port and Mitleb 1980, Elsendoorn 1980, 'Williams 1980, Flege and Port ImH. :\[itleb 1981). However, further investigation is clearly needed to evaluate the importance of phonological translation as a mechanism leading to foreign accent. The hypothesis as it is formulated leads to two predictions through which it is potentiaIIy falsifiable. First, it predicts that bilingualism is not possible at the phonetic level since it posits that pronunciation of a foreign language is, for most speakers, eventuaIIy shaped to conform to an acoustic model provided jointly by the native and foreign languages. An individual might learn to speak a foreign language without apparent accent, but a fine-grained acoustic analysis should reveal differences be• tween the language learner and native speakers of the target language along those parameters where phonetic differences exist between the native and target language. The expected pattern is for the language learner to produce foreign language sounds in such a \Vay that they are phonetically intermediate to those found in the native and target languages. Second, one would expect phonetic learning in a second language to affect pronunciation of a learner's first language. If the acoustic model provided by native language sounds can influence foreign language pro• nunciation, then foreign language sounds should, by virtue of their identi• fication with native language sounds, exert an influence in the opposite direction. The extent of such influence might be determined by such factors as the discriminabiJity of cross-language phonetic differences,:! an indi• vidual's ability to transform an auditory image abstracted from an acoustic model into a stable articulatory representation, or the amount of use and/or exposure to the foreign language (see Rosen 1979). Some support for this prediction is evident in Figure 1, which shows that the VOT values in French words produced by French-English bilinguals are slightly less French-like (i.e., longer, in the direction of English values) than are stops

~ \Ve have n'ot discussed here the perceptual modification which ma~' occur as the result of exposure to a foreign language. There is evidence to suggest that hoth dis• crimination (~[acKain, Best. and Strange 1980) and identification (Williams 1980) of foreign language speech sounds may change during second language learning. Such changes might well influence speech production. Foreign Accent 453 produced by French monulin.!.!;uals in either Canada or France. Sonw\\'hat more substantial support for this prediction is provided by \Villiarns' ( 1980) ~tudy of Puerto Rican children learning English. \Villiams found that nativE' Spanish-speaking children learned to produce English stops with increas• ingly large (i.e .. English-like) VOT values and also to produce Spanish stops more like English stops (i.e., \'lith more aspiration) than monolingual Spanish speakers. The hypothesis proposed here wiIi not, of course, provide an explana• tion for cross-language interference where translation pairs of sounds or non-segmental dimensions in two languages cannot be identified by theo• retical or empirical means. Nor does it attempt to account for the carry-over of phonological processes from the native to the target language. How• ever, it does provide an explanation for the important interference that exists at the level of phonetic implementation during second language ac• quisition (Flege and Port 1981). If it is substantiated by further research, the phonological translation hypothesis offered here may provide insight into why learners typically continue to show interference at the phonetic level when other aspects of their control of L2 are more nearly native-like.

REFERENCES Anisfeld, M., N. Bogo and W. Lambert. 1962. Evaluational reactions to ac• cented English speech. Journal of Abnormal and Social Psychology 65:223• 31. Arthur, B., D. Farrar and C. Bradford. 1974. Evaluation reactions of college students to dialect differences in the English of ~fexican-Americans. Lan• guage Gild Speech 17: 2,55-70. Asher, J. and R. Garcia. 1969. The optimal age to learn a foreign language. The Modern Language Joumal 53:334-41.

Brennan,ness E.,by magnitude E. Ryan and estimation\V. Dawson. and sensory 1975. modality Scaling of matching. apparentJournal accented• of Psycholillguistic Research 4:27-36. Burling, R. 1981. Sodal constraints on adult language learning. Paper presented to the New York Academy of Sciences, Ne\v York City, January, 1981. To appear in the conference proceedings. Caramazza, A., C. Yeni-Komshian, E. Zurif and E. Carbone. 1973. The acquisi• tion of a new phonological contrast: The case of stop consonants in French• English bilinguals. Journal of the Acoustical Society of America 54:421-28. Caramazza, A. and C. Yeni-Komshian. 1974. Voice onset time in two French dialects. Journal of Phonetics 2:239-45 .. Catford, J. 1965. A linguistic theory of translation. London: Oxford University Press. Chen, M. 1970. Vowel length variation as a function of the voicing of the following consonant environment. Phonetic:a 22: 129-59. Clumeck, H. 1976. Patterns of soft palate movements in six languages. Journal of Phonetics 4:337-51. Delattre, P. 1966. A comparison of syllable length conditioning among languages. I nternational Review of Applied Linguistics 4: 183-98. Elsendoorn, B. 1980. Vowel durations as a cue to a Dutch foreign accent in English CVC-words. Prngress Report of the Institute of Phonetics 5:64-86. University of Utrecht, Holland. Ervin-Tripp, S. 1974. Is second language learning like the first? TESOL Quar• terly 8: 111-27. 454 TESOL Quarterly

Fathman. A. 1975. The relatiunship between age and secund language produc• tive ability. Language Lcarning 25::24.5-.53. Ferguson. C. and O. Garnica. 1975. Theories of phunological development. In E. H. Lenneherg and E. Lenneberg (Eds.), Foundations IIf language de• vdopn/i'nt, 'lol. 1. 0:ew York: Acadl'mic Press. Flege. J. 1980. Phunetic approximation in secund bnguage acquisition. Lan• guage Lcami/lg 30: 117 -:34. Flege, J. and R. Hammond. 1980. Stereotypes in foreign accent-A preliminary analysis. Paper presented at the Southeast Conference on Linguistics, ylem• phis, April, 1980. Flege, J. and R. Hammond. 1981. Speakers' awareness l)f some nonsegmental aspects of foreign accent. In ~1. Henderson (Ed.), 1980 Mid-A.merica linguistics conference papers. Lawrence: Department of Linguistics, Uni• versity of Kansas. Flege, J. and R. Port. 1981. Cross-language phonetic interference: Arabic to English. Language and Speech 24: 125-146. Giles, H. 1972. The effects of stimulus mildness-broadness in the evaluation of accents. Language and Speech 15:262-69. Hill, J. 1970. Foreign accents, language acquisition, and cerebral dominance revisited. Language Learning 20:237-48. Krashen, S. 1973. Lateralization, language learning, and the critical period: Some new evidence. Language Learning 23:63-74. Ladefoged, P. 1980. What are linguistic sounds made of? Language 56:485-502.

Leopold,Lenneberg, W. E. 1947. 1967. SpeechBiological development foundntions of a of bilingual language. child,New Sound-learning York: Wiley. in the first two years, Vol. 2. Evanston: Northwestern University Press. Lisker, L. 1978. Speech across a linguistic boundary: Category naming and phonetic description. Haskins Laboratories Status Report on Speech Re• search SR-54: 105-11. MacKain, K., C. Best, and W. Strange. 1980. Native language effects on the perception of liquids. Journal of the Acoustical Society of America 27.S27 (A) . ~vlacken, M. and D. Barton. 1980. The acquisition of the voicing contrast in English: A study of voice onset time in word-initial stop consonants. Journal of Child Language 7:41-74. r-.Iajewski, W., H. Hollein and J. Zalewski. 1972. Speaking fundamental fre• quency of Polish adult males. Phonetica 25: 119-25. Major, R. 1977. Phonological differentiation of a bilingual child. Ohio State Unit;ersity "Working Papers in Linguistics 22:88-122. McLaughlin, B. 1978. Second language acquisition in childhood. Hillsdale, NJ: Lawrence Erlbaum Associates. Mitleb, F. 1981. Timing of English vowels spoken with an Arabic accent. Re• search in Phonetics 2: 193-226. Indiana University: Department of lin• guistics. Morais, J., L. Cary, J. Alegria, and P. Bertelson. 1979. Does awareness of speech as a sequence of phones arise spontaneously? Cognition 7: 323-331. Nemser, W. 1971. Approximative systems of foreign language learners. Inter• national Rer;-iew of Applied Linguistics 9: 115-123. Neufeld, G. 1980. On the adult's ability to acquire phonology. TESOL Quar• terly 14:285-98. Niemi, J. 1979. Noun and noun phrase stress: A phonetic study of English supplemented with an error analysis using Finnish speaker-hearers, 1. Pro• duction. Language and Speech 22:221-35. Oller, D. 1977. Syllable timing and Spanish, English, and Finnish. In H. Hollein and P. Hollein (Eds.), Current issues in the phonetic sciences. Amsterdam: John Benjamins. Oyama, S. 1976. A sensitive period for the acquisition of the non-native pho• nological system. Jaurnal af Psyc1lOlingui.'ific Research 5: 261-83. Foreign Accent 455

Penfield, W. and L. Roberts. 1966. Speech and brain-mechan~ms. New York: Atheneum. Pinkerton. S. 197:2. The leaming of English suprasegmental rules for stress and final syllables by Spanish speakers. Paper presented at the Mid-Americ.:a Linguistics Conference, October, 1973. l'lJlitzer, R. and L. \Veiss. 1969. De\'elopmental aspects of auditory discrimina• tion, echo response and recall. The Modem Language Journal 53:75-85. port, R. and F. Mitleb. 1980. Phonetic and phonological manifestations of the voicing contrast in Arabic-accented English. Research in Phonetics 1: 137 -66. Indiana University: Department of Linguistics. Port. R., S. AI-Ani and S. Maeda. 1980. Temporal compensation and universal phonetics. Phonetica 37 :235-52. Redlinger, W., and T. Park. 1980. Language mixing in young bilinguals. Journal of Child Language 7:337-352. Rosen, S. 1979. Range and frequency effects in consonant categorization. Jour• nal of Phonetics 4:393-402. Ryan, E. and :VI. Carranza. 1975. Evaluative reactions of adolescents toward speakers of standard English and Mexican American accented English. Journal of Personality and Social Psychology 31 :855-63. Ryan, E., M. Carranza and R. Moffie. 1977. Reactions toward varying degrees of accentedness in the speech of Spanish-English bilinguals. Language and Speech 20:267-73. Scovel, T. 1969. Foreign accent, language acquisition and cerebral dominance. Language Learning 19:245-54. Sclinker, L. 1972. Interlanguage. International Review of Applied Linguistics 10:209-231. Snow, C. and M. Hoefnagel-H6hle. 1977. Age differences in the pronunciation of foreign sounds. Language and Speech 20: 357 -65. Snow, C. and M. Hoefnagel-Hohle. 1978. The critical period for language acquisition: Evidence from second language learning. Child Development 49: 114-28. Suomi, K. 1976. English voiceless and voiced stops as produced by Finnish and native speakers. Jyviiskyla Contrastive Studies 2. Jyvaskyla Univer• sity, Department of English, Finland. Tervoort, B. 1979. Foreign language awareness in a five- to nine-year-old lexi• cographer. Journal of Child Language 6: 159-166. \' alette, R. 1964. Some reflections on second-language learning in young chil• dren. La/lguilge Learning 14:91-98. Willems. N. 1978. Pitch characteristics of production of a Dutch accent in English: A pilot study. Progress Report of the Institute of Phonetics 3:54-66. University of Utrecht, Holland. Williams, L. 1980. Phonetic variation as a function of second-language learn• ing. In G. Yeni-Komshian, J. Kavanagh, and C. Ferguson (Eds.), Child phonology, Voll/me 2: Perception. New York: Academic Press. Winitz, H. 1981. Two considerations in the comprehension of Ll and L2. Paper presented to the New York Academy of Sciences, New York, January, 1981. To appear in the conference proceedings.