You are on page 1of 18
The perception of the interdental fricatives [9] and [5] in English words by Hong Kong Cantonese-speaking ESL learners' . ‘Melissa Sze Man LAU & Cathy Sin Ping WONG sho ow fglnt® bog (Department of English ‘The Hong Kong Polytechnic University) (2) ‘This pape examines how Contonese-speaking ESL lamers perceive the English interdentl fsicatives [8] and 8} Sint test items, forty-eight of which were minimal pairs contrasting any two of the four segments: [8], {8}, {f] and [a] in word-inital and word-final postions. Sixty-five Cantonese-speaking ESI. leamers were asked to indicate whether the minimal pairs were the same or diferent. Results show thet subjects had dlficuities in differentiating [8] from [for [0] but not (8) som [4] This indicates that the voiceless interdentalfricative [8] and its voice counterpat (6) do not have the status in the interlenguage system of these learners 1. Introduction One of the characteristics of the spoken English produced by Hong Kong ESI. leamers is the substitution of[f] and [d forthe English interdental fitatives [0] and [6] (Bolton & Kwok 1990, Hung 2000) Hong Kong Cantonese speakers pronounce thing (Orcas ffrn] and hey [Sor] as (dor) For damiled version ofthis study, readers can afer tothe Fal yeu projectreport with an identical ite “The Perception of 8 and (6/in English words by Csetonese ESL. Lesners” (2001) authored by Melisa SM Lou, We would like t take this opportunity to express ou gratitude to ll who patcipated in this project We would also to thank Dr ili O'Grady forks comments on the presentation of the paper Any ror, of couse remain our A simple contrastve analysis of the two phonological systems ofthe leamers first language (Cantonese) andthe target language (English) seems to provide a ready answer to why such a feature surfaces in the inteslanguage system of ‘these Iearners-the two inferdentlfrcatives (8) and (6) are not found in Cantonese. However, this simplistic view may only explain why the production does not conform tothe tw target segments but cannot explain why they are replaced by the two specific native segments [f] and [4] Do Hong Kong Cantonese speakers misperceve (6) and (6) asf] and {4}, and so in production, they produce (f] and (4? This isan unanswered question because limited research has been done on the perception of the interdental fricatives [8] and [6} by Cantonese-speaking leamers ‘Most of the studies onthe aquisition of English by Cantonese speakers in Hong Kong focus on production, and ‘herefore no research has been conducted on the pereption ofthe inerdentalfictives [0] and [6]. It is necessary to investigate whether the production and perception of these twe sounds ae slated ‘The purpose of this study is to find out whether Cantonese speakers confuse the sounds [6] and [8] with [f] and [4] in perception 2. The English Interdental Fricatives Cantonese speakers are not the only group of learners who find difficulties with the English interdental fricatives. Ina number of studies on lamers of various language backgrounds, researchers find that these two segments always pose problems for leamers of English. Schunidt 1977) find that speakers of Egyptian Arabic use(s} and {z] for the English (0] and {8}. Similarly, Kentstowicz and Kisseberth (1979) note that fs] and [z] are used by the French speakers instead of [9] and [8] However, [t] and {dl are used by Serbo-Croatian speakers although both French and Serbo-Croatian lack the non-stident interdental fricatives [9] and 8) and thet [=], (2), [¥] and [el] are all present in both languages. According to Ritchie (1968), Russian speakers earning English typically substitute [t] for [0], and Japanese speakers substitute [s] for the same sound even though both Russian and Japanese have both [t] and [5] According to Bolton & Kwok (1990) and Hung (2000), substitution for the English (8) and (8] is very common ‘among Cantonese speaking ESI. learners. Bolton & Kwok (1990:153) find thatthe word thins [815s] is pronounced ‘8 [finks} by Cantonese speakers. Hung's (2000) study shows that while some of his subjects did produce an interdental, ficative [8] in [Oin} shin and [xloudtn] in clothing, others produced an [f] instead ofa [@} in all environments. He also points out that (6} is apparently not part ofthe learners’ system. He claims that neither isthe voiced interdentalfricative {0] partof the consonant system of most speakers of Hong Kong English Very often, (] is produced in word initial or intervocalic position in place of [8 for example, [4s] for dis, and forad} for Brother. ‘Table 1 below (based on Weinberger 1997: 269 with authors’ addition) surnmarizes the substitution pattems of [8] and [5] of leamers from different L1 background. The fac that leamers of diferent background employ different segments to replace the target [8] and [9] isan intriguing phenomenon for interlanguage phonologists. Weinberger (1997) refers to this phenomenon as “differential substitution” and he explores in-depth the possible phonol explanations for it 1 seems obvious from Table I below thatthe substitution pattem serves as good evidence for Ll transfer. However, Wenk (1979) notes a developmental pattem among the French learners with regard to the substitution of [D] ‘and [6], He found thatthe relatively advanced leamers of English generally had the highest proportion of correct [6] productions, the beginners had the highest proportion of [£] substitutions, and the most [+] substitutions were found ia the intermediete group. He points out that [f] is substituted for (8) by both L1 and L2 learners at the early stage due to the fuctthat the acoustic properties of[£] and [8] are very similar, and these two segments might be confused on strictly perceptual grounds Table 1 Differential substitution of (8) and (8) Ti air Sit | Smay Tapanese = z Kohmoto 1965, Ritchie 1968 French = z Weinreich 1968 ‘Cea A z Tames TORT Bayptian Arabic 3 z ‘Seki 1977 Rasa z a Weinrcich 1968 Serbo-Croatian t a TKanistowicz and Kissebers 1979 Sinhalese t q Wachael's 1975 Hangar, e ‘Altenberg-Vago 1983 Dut wi | Tames 1984 si Mandan Ti Fonda 1984 tt Cantonese T # Bolton © Rwok 1990, Hang 2000 Ts the substitution of {1 and [1] for [8] and 8} among speakers of Cantonese a feature of L1 transferor is ita developmental pattem or both? In other words, does the absence off} and [6} ead to their difficulty in producing these ‘two segments? Or is mispronunciation a result of misperception’ Are Cantonese-speaking ESL Jeamers able to perceive the distinction beoween (} and (f, as well as [and [6]? Perception and Production ‘The relationship between perception and production in second language ceqsition hs ong ben the fos of study fr many intehnguage phonologsts Leather & James (1996) observe that whether perception might lead production, or vce veri geting mere and more tention i he foreign lnguage teaching domain ‘onthe on hand, its sugested tht perception and production ae often mutually facilitative (Leather & Janes, 1986; Rocket, 1995) Forinstance, Yamada, Tokjr, Bradlow and Pisoni (1995) found tht rng Jopanese speakers to perceive English [and 1] improved their accuracy in prodvcing these sounds even without production tsning Morthews (1997) observed that Iapanese lamers who recived tring ony in the production of English contrasts ‘demonstrated significant improvement in their ability to discriminate the segments involved Phonetialy, the sent is an unaspirated [rather than a fuly-voiced(d} Voicing is ot contastive in Cantonese On the other hend, Leather and James (1996) continues to suggest that there is no constant, single interrelation between perception and production but they are developmentally diffrentited Nemser (1971) cared out some experiments to test Hungarian-English leamers’ perception and production sbiity, The results show that there is syzumetry between perception and production. The speakers of Hungarian tended to perceive English interdenals [0] and 6] as lebia catives[] and [v] respectively, whereas production results show that [f] and [are totally excluded They always produce [0] and [6] a apical stops [7] The result of dissimilar percoptual and productive pattems supports Nemser's (1971) argument that inteingul identifications from native language and target language are essential to only some parts of interlanguage Broselow and Park (1995) proposed the Split Parameter Seting Hypothesis which suggests that lamers do ‘ove through developmental stages in which perception end preduction are ‘split, based on the different “parameter settings of L1 and L2 Fits, the LI seting govern both perception and production; next, the 12 setting governs perception while Lt continues to goven production; finally 2 sting are achieved for both perception and production Leather and Juss (1996) review other studies which provide no clear evidence of any constant interelation between perception and production Another study which suggests that production isin advance of perception is Flege (1997) who studied the perception and production of a new vowel eategory by German speakers of English The result is tat the enpevienced German speakers of English didnot differ significantly from the native English speakers in the production experiment However, in the perception experiment, they didnot filly match the native English speakers “Perception and production may be les interdependent, in theta learner's ability to perceive distinctive phonetic detail inthe speech of others may not correlate well with his or her own phonetic production (Leather, 1999:7) Ia short, the relationship between perception and production is rather complex. In this study, we will examine te perception of 8} and 8} by Cantonese ESL leamers and attempt to identify the perception pattern regarding thse ‘ovo segments to examine whether they confuse [9] and [0] with [f] and [4] in perception 4. Methodology 41. Research Questions ‘The research questions that are being addressed in this study are four-fold 1, Does the proficiency level ofthe lenmers affect their perception of [8] and [8]? 2. How frequently do Cantonese ESL learners confuse [8] with [) and [8] with [4]? 3. What is the pattem of misperception, ifany? 4 Does the position where 8] oF [0] oveurs affect how Cantonese ESL. leamers perceive these to segments? 42 Subjects The data presented in this study were obtained from 65 Cantonese.speaking leamers of English in Hong Kong All of them are students fom secondary schools in Hong Kong. 40 of them were Secondary | students, aged between. ‘leven and fifteen while 25 of them were Secondary 6 students, aged between sixteen and nineteen. The $6 students Jad five more years of exposure to English in the classroom context than the SI students Their proficiency in English ‘ay be characterized as beginners and intermediate 43. Test Materials In order to investigate whether leamers are able to perceive the diferences among the four segmens: [8], (6), [1] ‘an (sets of minimal pairs were used ina perception test A total of 60 minimal pairs were randomized to form alist oftestitems The wor consisted often deen types of word par, si of which were par contesting two ofthe for sounds in question: [8], {3} {f] and [d] and four types were identical por actingas control Each type contained 3 pir for wordt poston and 3 pairs fr wor-fnal poston For instance the minimal pairs contrasting [0] and {fin the word-nal position ere: wreath [vi] verous effi, loath Lou] versus loaf. Some pairs Were repected because ditional pais that belong toa puticulr type could at be found. ‘The fil et of tet items canbe found in sppendi 44 Data Coleedon Procedure The word pairsin the word ist were randomized native speaker of Bish English wes asked o read aloud the randomized list and the speech was recorded by a minidisk (MD) recorder in the Speech Laboratory atthe Hong Kong Polytechnie University. In order a alow ample time for the subjects to choose the answer, there was a fve-second pause between pairs ‘The perception test wes cared out in loop room atthe secondary school so that ech subject cold listen tothe 1aterials hough hiser ovm earphone ‘This ensured that they could listen tthe word clearly The two groups of subject did the test separately. Subjects were aed to isten caeflly to exch pair of words, and then decide whether the two words were the same or diferent by circling either (for same) or D (fr ferent) onthe answer set provided Two native specker of Bolish, one male and one female, were also asked to do the same percepion test as contol 4S Data Analysis ‘The daa collected were compared wit the intended targets thats, either SAME or DIFFERENT Eech conect response sored one pint For example, fa subject cheseS (sume) forthe frst word par thay [ot] and thaw [99H], one point would be awarded for this ICD (different) was selected for this word pair, no marks would be given 5. Findings 51 Overall Performance On average the subjects achieved am average accuracy of 72% The average scores oblained from the two roups of subjects are shown in tems of percentages in Figure I below The 6 students obtained a higher average than theSi students. The avemge percentage is 76 78% and 68296 respectively Although the sores are rather high, detiled examination ofthe results indicetes tha the lamers did conse the four sounds (8), (0), (7, [] The performance on the minimal pairs of (0 and (] was the best whereas the minimal pairs of [0] an (f] was the most problematic 72.00% 4 : ae 68.00% be 66.00% | | ee ae i 64.00% 62.00% : Subject groups Fig 1 Average scores inthe perception test 52. Proficieney Level and Perception Ability ‘The performance ofthe Secondary 1 and Secondary 6 sdens ere compared inorder to examine whether developmental patter s evident. Table 2 below illustrates the percentages of scores obtained by the two groups of subjects Though the average pacenage obtained by the 86 students (16 78%) higher than tha ofthe 1 students (68.2%), the test sul indicates thatthe difference is not significant (two-tailed, 3 45, p<0 05) Table2 Comparing the scores of 1 and $6 students “Ditta ewe L SI and $6 (S681) “ype B (6) vs [ni Ta 765% ype [Ola [tint 607% DG BOM [ype [Ove [d)inital 99198 wae ‘SypeF 8) initia 575% wa Bam FType @ (6) vs (jinitiad 825% oe “aS% Faypel [f] vs [d) initiat 99.17% 100% 7% Type (O)vs(6\ fina EI TH3% T3% ype (v(m 10% HypeD _[O]vs (a) final 95% 100% ‘ype (6) (6) final 85% WO Hype @ (8) dl nal S17 100% Aypet(f] ve [d] inl 925% WO From the above table, it can be seen thatthe types in which the S6 students outperformed the $1 students all involve segments in the word-initial position The three types that show the greatest difference ere Type B-{O] versus (6} (26.33%), Type C-{0] versus [f] (22 66%), and Type F-{6] versus ff] (23 83%), indicating that (6] and (6) occurring in ‘the word-initial postion are especially difficult to be perceived accurately by beginning learners 5.3. Performance on the Six Types of Minimal Pure ‘The subjects performed differently on the six types of minimal pair Its found thatthe minimal pairs of [8] and [i] are the most difficult ones for the subjects to perceive while the minimal pairs of [6] and (d] are the easiest Tuble3 Scores according to test type TRENIMAL PATE "AVERAGE PERCENTAGE Tyee Bai Ta Type B 16) vs (0), me Ter flvsEf] Ba Teo wld) wim Tet held) wi Typed (8}s5 (4 TAIN Table 3 illustrates the averoge sare ofthe six types of minimal pai from the owestto the highest. kis evident ‘hat he minimal pais consining [9] and [£] eonfute the subjects most requntly. The sore is oly 12.31% The rents suggest thatthe subjects have lificltyin dferentoting these two sounds The minimal pis containing [9] an (8 also caus problems fr the subjects The sore is elatvely low (60%) compared to the other four types but it isrmuch higher thn that of[6} and [| Based on hesetwo observations, we may conclude thatthe ocses inerdentl Sicative [8 is very dificult fr thesetwo groups of lamers. Howeser, the minimal pairs containing [9 nd (dl donot ‘pose a problem for the subjects. The average score for this type is as high a8 98.21% ‘The fect that [8] seems extremely dificil for these lamer to perceives only partly acurte because when itis contasive wit it isnot dificult at ell (98.21%). In other words, for these two groups of Teamers, the distinction of stop ([a})versusa clive ((8) is esier to perceive than the distinction between a voiceless ({9}) and a voiced fictive ((3) ora nterdenta icatve (19) and bilabial cave (fF). In effect, it suggests thatthe diference in manner of artieuatio s mere realy distinguishable than a dillerence in place of aticulation Whether thisisaplcable to othe eontasve pis or nts ubjeto further sty 54 Word-initial positon versus wordtina position Tn this setion, the retlts of the ten types of minimal pir ae compared in terms ofthe potion where the segments oer worn vers wedi. Table below shows the average percentage foreach ype of minimal its in both word-inis and word-inal positions Onaverge, the leamers were able to perceive the dierence between the segments beter in wor-finl poston tha in word-nitil poston However, the est result indicates ‘thatthe difference is not significant (two-tailed t=4 75, p>0.05) Tabled Average score in terms of syllable position CONTRASTIVE PAIRS Tad Tyee sth 15% Typed (hws (a) 39% TypeB he} 30% TpeF helt om Tipe Bela Ti Tpel veld) 100% TDENTICAL PAIRS Tnitial TypeA Tae 3% TpeE (oh (5) DH Type oe ro Typet Tala) 1% ‘Average (alten pes) TH Although the difference isnot statistically significant, ftom the scores associated with the four identical pairs, it is apparent that all the four sounds under scrutiny are more readily perceived in word-final postions since the subjects 10 ‘schieved higher stores in most eases excep fr], which shows difference of only 196) Onthe other hand, when these four sounds are being cotrstd, 8] isthe easiest to perceive in word-final position athe sores asoatd with ‘tare ll inthe word-fnl position In conclusion, although the elect ofthe positon where the four sounds occur does nt show any ttiscal significant correlat ‘with perception, «pattern thatthe word-final postion seem to facilitate perception can be discemed, especially in the case ofthe voiced interdentalfricative 8} 6. Discussion The aim of this study isto investigate the pate of perception and misperception of Cantonese speaking ESL Tears with eer tothe four segments (0, (3, and [2] and io examine whether the position where these four segments occur and whether the evel of prfiseny ofthe leamers fet their perception 6.1, Differential perceptual difficulties: [@] is more difficult than [8] Its found that these eamers id have diioulty in perceiving he diference between [0] and [9] thus providing a possible explanation ofthe production dla eer in Ballon & Kwok (1990) and Hang 2000) Iti quite possible that Comes ESI. leamers persive [8] 48 fs hey naturally produce (instead of] because theater is absent inthe Lt Although the two sounds (0 and [6] are replace by and) nprodvetioneccording to previous findings, the ‘esults in perception of these two sounds shown in the present study suggest the two sounds merit different status in the phonological system ofthese leamers. ‘The subjects could not differentiate the minimal pis contesting (6) af ‘The average score chsined bythe subjects is 12 31% A observed in Hung (2000) some of his subjects reduce instead of [6] in al environments and he contuded that 0] is apparently not pat ofthe consonant system of thee subjects, The results rom the presen study leds upport to Hung’s egument Nevertheless, the same subjects were able to perceive the dference inthe minimal pars of [8 anf} unlike the 8) / {f] counterpart. The average score is 85 13% This is very different from the case of [6] and [f]. In other words, u

You might also like