Greek and English Bilinguals and Thai Bilabial Stops Bilingualism - Essay Example

Greek English Bilinguals and Thai Bilabial Stops Bilingualism and the Relationship between Perception and Production: A Critical Essay This paper is a critical review of a journal article which documented two research experiments aimed at investigating the association between perception and production among bilingual Greek / English bilinguals as regards to Thai Bilabial stops. A total of 60 respondents equally representing Australian / English, Greek / Australian – English and Thai speakers participated in two experiments. Three Thai speech contrasts, /ba/ versus /p}a/, /pa/ versus /p}a/, and /ba/ versus /pa/, were utilised as stimuli in Experiment 1. A pretest block of 12 trials and 3 blocks of 12 actual test trials were conducted in a sound-attenuated chamber were administered and mean discrimination indices were calculated. Data were further analysed using repeated measures analysis of variance. Meanwhile, two production tasks namely, imitation and reading were used in Experiment 2. Statistical measures using correlation and analysis of variance were carried out. Results revealed no significant difference among the three speaker groups in terms of discriminating the three aforestated Thai distinctions in the first experiment. In the second experiment, the imitation task production of the Greek / Australian – English speakers deviated from the Thai model significantly more than the Australian – English speakers. In the second experiment, there was no difference between the groups which produced voice onset times (VOTs) with a long lead as well as the group which produced the short lag. However, VOTs with long lag were significantly higher than those with a short lag. 1.0. The First Experiment The goal in Experiment 1 was to examine bilingual subjects speaking Greek and Australian-English bilinguals and monolingual subjects speaking Australian-English are able to discriminate non-native constants, particularly Thai bilabial stop contrasts when they are exposed to two sets of phonemes. The researchers assumed that the presence of phonetically similar sounds used as allophones in the Greek and Australian native languages will facilitate discrimination of the Thai sounds. While the hypothesis for the first experiment was not explicitly stated, it may be readily deduced from the presentation of the research problem. The null hypothesis may be stated as follows: there is no significant difference in the discrimination of Thai contrasts among the three speaker groups. To evaluate the null hypothesis, discrimination indices were computed for the three groups and the resulting values were analysed using repeated measures analysis of variance (ANOVA). The study was written for an audience where level where research and statistical competency is high. This observation is given scaffold by the fact that the answer to the research problem in the first experiment, as well as the hypothesis testing, were not cued using words such as “the null hypotheses is accepted or rejected”, but rather, as indicated in the Beach, Burnham and Kitamura (2001) study, “Greek / Australian – English speakers showed a tendency for better discrimination than the Australian-English speakers … due to the fact that Greek bilinguals have /p/ in their Greek inventory” and “all groups performed consistently more poorly with the /b/ - /p/ contrast”, respectively (Beach, et al., 2001, p. 227). The reader had to do a lot of “reading between the lines” to get a better grasp of the substance of the study. Repeated-measures analysis of variance (ANOVA) was used in this study to analyse the same group of subjects in different treatment conditions, specifically the three Thai contrasts. Specifically, the three Thai distinctions were analysed in 3 x 3 language groups (i.e., the contrast) and the repeated measures were applied on the second factor. Repeated-measures ANOVA was also an excellent choice by Beach, et al. (2001), because this technique is used to determine if differences identified between treatment conditions are significantly greater compared to differences due to chance. A distinguishing feature of repeated-measures ANOVA is the removal of individual differences from the denominator of the F-ratio or residual variance. The residual variance or error variance gauges how much variance is expected after the treatment effects and individual differences have been taken out (Gravetter and Wallnau, 2008). It was also apparent from the description in the journal article that the respondents were not selected at random, but through purposive or judgmental sampling. Purposive sampling is a non-probability sampling procedure based on “selecting the individuals as examples according to the purpose of the researcher … due to good evidence that he is a representative of the total population” (Calmorin and Calmorin, 2007, p. 104). As explained by Chambliss and Schutt (2010), each of the sample elements chosen by the researcher in a purposive sampling is selected because of their unique position and is, therefore, targeted because they possess attributes which are of central importance in the issue under investigation. In this study, the respondents were purposively chosen by virtue of the languages they speak. The respondents chosen should be speakers of Australian / English, Greek, Australian – English and Thai. In this study, discrimination index (DI) was designated as the dependent variable and was computed for each participant for all of the three speech distinctions. The number of the incorrect responses in each of the different trials is subtracted from the number of correct responses in the same trials, and the difference is divided by the number of different trials. Use of the discrimination index as a variable in this study seems appropriate except that the respondents were not sampled using probability-type sampling. Bell (2008) indicated that discrimination index may be used in any analysis which involves categorisation and the probability that any two individuals selected at random possess a different pre-specified characteristic. However, owing to the nature of the study, which may not naturally proceed on random selection of respondents or subjects, the randomisation requirement may be put aside. This is this writer’s personal opinion as a critic of the paper. It was noted from the description of the respondent demographics that there were equal number of males and females among the Thai control group. This suggests that basically, there was an attempt to institute control for gender among the control group but not for the other two groups. It was, however observed that the age of the respondents, although, close in value, were not very similar. In this regard, it may not be safely assumed that the researchers attempted to establish control over the age variable. It is this writer’s belief that age is an important intervening factor in this regard. From previous readings regarding this topic, particularly in the work of Burnham and Mattock (2007), Australian English – speaking subjects and Thai-speaking children found the non-phonemic contrast [ba]-[pa] more difficult compared to phonemic contrast [p]-[ ph]. This observation suggests that the [ba]-[pa] contrast takes a longer time to master. In addition, results show that Thai speakers show convergence of proficiency on both contrasts, an indicator of phonemic experience. Inversely, Australian English-speaking children exhibited ease in identifying the [ba]-[pa] contrast compared to discriminating tones. Meanwhile Australian English-speaking adults are more adept at identifying discriminating tones compared to the [ba]-[pa] contrast (Burnham and Mattock 2007). The Burnham and Mattock (2007) study carried out an experiment on the relationship between tone (Thai) and none-tone (Australian-English) language perception in a developmental context was conducted on adults and children aged 4, 6, and 8 years old. Eight groups of 24 subjects each were tested using a same/different AX two-alternative forced choice speeded button-pressed discrimination task divided into three trial blocks consisting of 5 randomly selected exemplars of each available phoneme. It may also be recalled that the Central Thai language is characterized by five tones and three bilabial stops for /b/, /p/, and /ph/ (Burnham and Mattock 2007). The findings of the Beach, et al. (2001) study was supported by the more recent Burnham and Mattock (2007), specially in the aspect that Australian English-speaking adults are more adept at identifying discriminating tones compared to the [ba]-[pa] contrast as evidenced by their ability to treat the experiment as a perceptual discrimination task with minimal use of linguistic processes. This ability enables Australian English-speaking adults to perform better in identifying psycho-acoustically salient non-native pitch differences compared to non-native consonant differences (Burnham and Mattock 2007). Burnham and Mattock (2007) also showed that the difficulty encountered by French speakers regarding stress contrasts may be likened to phoneme contrasts. More importantly, the study provided two crucial findings on language specific speech perception. First, children’s speech perception tends to be more language specific for a lexical tone compared to consonantal voicing. Second, children prefer small phonemically distinctive intonational units compared to random and non-phonemic intonational variations (Burnham & Mattock 2007). These findings points towards the age as an important factor in language studies 2.0. The Second Experiment The second experiment has three-fold goals: (1) to assess the effect of experience with more than one language; (2) verify the relationship between the perception and production of sounds; and (3) evaluate the effect of an individual’s language background on the association between sound perception and sound production. The same respondents in Experiment 1 were used in Experiment 2. The experiment used imitation and reading tasks recorded via Sony Professional Walkman WM – D6C. The sample were processed through digitisation at a sampling rate of 50kHz, while voice onset times (VOTs) were measured through spectrograms generated from the Kay 4300 Computerized Speech Laboratory (CSL) spectogrammetry. The second experiment tested the following null hypothesis: (1) there is a significant relationship between language background and sound production deviation; (2) there is significant association between production deviance score and perceptual ability; and (3) Language background does not mediate the relationship between sound production and sound perception. The first and third hypotheses were verified using ANOVA, while the second hypothesis was tested using correlation analysis. Use of analysis of variance in testing hypotheses 1 and 3 is valid in the sense that the experiment compared the mean VOT times among the three groups of speakers. Mendenhall, Beaver, and Beaver (2009) elucidated that analysis of variance divides the total variation in the measurement into parts which may be attributed to one or more factors of interest to the researcher. In the second experiment, the measurement involved is the VOTs of the three speaker groups and the factor is language background. Hence, analysis of variance is a good statistical technique in this regard. There is just one observation about the use of analysis of variance in this study, including that of repeated measures ANOVA in this experiment – there was no mention in the journal article about the verification procedure applied to be able to use ANOVA. Berk and Carey (2010) maintained that assumptions for the use of ANOVA necessitate that data be normally distributed, and therefore, the variance is equal. It was observed that either no test was conducted or no such procedure was reported to confirm such normality of data, such as Levene’s test of homogeneity of variance. Levene’s test for checking homogeneity of variance is believed to be a convenient procedure which can be implemented using various statistical software packages (Ellison, Barwick, and Farrant, 2009) Values of VOT were found to be similar to the modeled speech, but the imitation for the /b/ and /ph/ were steadily negative and positive, respectively compared to the Thai model based on the production performance in the Thai imitation task. Specifically, the bilingual subjects deviated significantly more than the monolingual subjects. Grounded on the research objectives of the second experiment, the production tasks used are deemed suitable. Jilka (2009) explained that production tasks are performed to determine differing types of intonational configurations, speaking rates, and degrees of fluency. If speakers are allowed to use any accent they are comfortable with, the degree of foreign accent is expected to vary as show in this study. Likewise, reading tasks are utilised to provide coverage of the phoneme inventory and certain phonotactic constellations, as well as elicitation of pitch accent distribution and tunes in a controlled environment. Reading task is recognized as good methodology in this regard because it discourages the use of avoidance strategies in relation to problematic sounds, words, and sentence structures to minimize error during evaluations. Meanwhile, Rathvon (2004) maintained that imitation tasks involve letting the speaker listen to sentences or phrases and have them repeat the same in verbatim (i.e., word for word). Sentence imitation is used to determine syntactic and semantic language skills. Normally, sentences are made up of real words following standard English syntax (Rathvon, 2004). However, in this study nonsense words were included in the task. As Rathvon (2004) rationalised, this strategy helps minimize the influence of semantic knowledge. Other techniques might involve sentence imitation tasks which makes use of real words in nonsensical but syntactically correct sentences. It should be recalled at this point that the production of speech is dependent on the synchronized operation of three subsystems: (1) respiratory; (2) phonatory; and (3) articulatory. The respiratory subsystem is responsible for the outflow of air required to produce sound. Meanwhile, the phonatory system interacts with the outflow of air to produce vibrations. On the other hand, the articulatory system regulates the air flow and vibration to form vowels and consonant sounds (Goldstein 2010). The second experiment also demonstrated substantial variation among the productions of the Greek/Australian – English speakers. A deeper analysis revealed that these group of speakers may be bracketed into three groups: those who generated Greek bilabial stops with short lead for [b] and short lag for [p] (i.e., S - S), while another group produced stops with long lead for [b] and long lag for [p] (i.e., L – L). On the other hand, the third group generated VOTs which are similar to the corresponding Thai stop. From related readings, it may be recalled that voice onset time (VOT) is the “interval between the release of a stop and the start of a following vowel” (Ladefoged, 2005). VOT for English is 50-60 ms for k, and a little less for t and p. The Spanish VOT is considerably shorter than for English (Ladefoged, 2005). As a general rule, English, German and Danish have long VOTs. Some languages such as Thai contrast three kinds of stops. Speech contrasts which do not conform phonologically or phonetically from a listener’s naïve language are limited through language experience. Language experience is much evident in adults, but has its beginnings during infancy until early adulthood. Best and Tyler (2007) evaluated the effect of non-native speech perception findings on phonological and phonetic aspects of second language perceptual learning by identification of commonalities and complementarities between inexperienced listeners and those studying a second language. A common belief exists that late learners find it difficult to learn certain second language segments and contrasts due to non-native speech perception. Evaluations on speech perception and learning should take into consideration that models of non-native speech perception such as the Perceptual Assimilation Model involves inexperienced listeners. In contrast, models of second language speech acquisition such as the Speech Learning Model gravitate towards experienced listeners (Best and Tyler 2007). Another explanation which may be forwarded for the substantial variations in the productions of the Greek / Australian – English speakers may be due to phonemes and phonological contrast. As Probert (2010) argued, phonology varies among different languages. It is, therefore, but natural for students of a foreign language to experience difficulty in pronouncing the unfamiliar sounds. Probert (2010), however explained that other than the aforementioned basic difference, languages also differ in the condition of the specific differences between sounds in their phonetic system. A very familiar example of such differences in English are the unaspirated [t] and the aspirated [th], but most English speakers are not actually and consciously aware of such difference (Probert, 2010). The differences revealed even among the same groups of speakers may also be explained by the age factor among the subjects. In the Kuhl, Conboy, Padden, Nelson, and Pruitt (2005) study which investigated the relationship between early speech perception and later language development in terms of its implications during the critical period of phonetic learning, even infant phonetic discrimination is widely observed. However, the rate of phonetic discrimination for non-native phonemes decreases by the end of the first year. Therefore, the authors postulated that the decrease in phonetic discrimination may be attributed to native language phonetic learning. Results of standard speech discrimination measurement among infants and language abilities during the 14th, 18th, 24th, and 30th month of life reveal a negative correlation between infants’ early native and non-native phonetic discrimination skills. In addition, native language discrimination at 7 months show improved later language abilities. On the other hand, non-native language discrimination revealed diminished later language abilities (Kuhl, Conboy, Padden, Nelson, and Pruitt 2005). The Beach, et al. (2001) study is similar to the more recent Tsukada (2004) study in that it compares native and non-native listeners in the cross language perception of final stops in Thai and English. Both studies involved bilinguals living in Sydney-Australia but participants of the Beach, et al. (2001) study have been residents of Australia for about 15 years while the Tsukada (2004) participants have resided for a mean of 5.5 years. The experiments were, however, quite different, particularly for the second experiment. The Thai bilinguals showed their ability to discriminate stop contrasts with consistent accuracy in both languages, whereas the English-speaking listeners discriminated well only in English. Both studies, however, illustrated that native Thai speakers showed better perception by virtue of their background with the Thai language. Another study, this time by Tsukada and Roengpitya (2008) examined the discrimination of English and Thai words with voices stops /p/, /t/, and /k/ in first and second language involving three groups of native Thai subjects. The first group of subjects consisted of 18 Thai listeners living in Australia. Meanwhile, the second and third group consisted of 12 university and 12 high school students, respectively. A control group of 18 Australian English listeners were included. Results of the study revealed a reasonably accurate discrimination for both English and Thai words on all three Thai groups, which is somewhat different with the present study. However, only the first group showed discrimination accuracy at par with the Australian English group in terms of English words. In addition, the third group of Thai students was more accurate in discriminating unreleased Thai compared to English stop contrasts (Tsukada & Roengpitya 2008). Again this study proved that language background affected speech production in the Thai Language. The latest Tsukada (2009) study is an empirical study on durational characteristics of English vowels /i/ and /I/ produced by Thai and Japanese L2 learners. In the Thai and Japanese language, vowel length is phonemic. Two set of experiments were performed for the study. The first experiment compared the /i/ and /I/ contrast in monosyllabic words ending with /p/, /k/, and /t/ by Australian English and Thai-speaking subjects. On the other hand, the second experiment compared the /i/ and /I/ contrast in monosyllabic words ending in /t/ and /d/ by Australian English and Japanese-speaking subjects. Based on the results, both the Japanese and Thai groups show a greater duration differentiation compared to the Australian English counterparts. It was concluded that language-specific factors such as allophonic variation should be included in language acquisition studies (Tsukada 2009). This is a common finding among Thai language studies, particularly the Beach, et al. (2004) work being critiqued. On the whole, it is interesting to note that the Slayden (2009) argument that most studies of the Thai language distinguish among five contrastive tones. The current study is one such example. Slayden (2009), however, pointed out that there is actually a sixth, glotally constricted tone. Hence, there is some verity in Slayden (2009) observation that although the Thai language is well-studied, and is, therefore, considered as well-described. It is on a good note that this critique ends knowing that the richness of the Thai language has been gaining more interest from contemporary researchers. 3.0. References Beach, EF, Burnham, D & Kitamura, C (2001), Bilingualism and the relationship between perception and production: Greek / English bilinguals and Thai bilabial stops. International Journal of Bilingualism, 5(2), 221-235. Berk, KN and Carey, P (2010) Data Analysis with Microsoft Excel: Updated for Office 2007, 3rd ed. Boston, MA: Brooks Cole / Cengage Learning. Best, CT & Tyler, MD (2007) Non-native and second-language speech perception: Commonalities and complementarities. In OS Bohn & MJ Munro (ed), Language Experience in Second Language Speech Learning: In Honour of James Emil Flege. Philadelphia, PA: John Benjamins North America. Burnham, D & Mattock, K (2007) The perception of tones and phones. In OS Bohn & MJ Munro (ed), Language Experience in Second Language Speech Learning: In Honour of James Emil Flege. Philadelphia, PA: John Benjamins North America. Calmorin, LP & Calmorin MA (2007) Research Methods and Thesis Writing, 2nd ed. Manila, PHL: Rex Book Store. Chambliss, DF and Schutt RK (2007) Making Sense of the Social World: Methods of Investigation, 3rd ed. Thousand Oaks, CA: Pine Forge Press / Sage Publications. Ellison, SLR, Barwick, VJ, and Farrant, TJD (2009) Practical Statistics for the Analytical Scientist: A Bench Guide, 2nd ed. Cambridge, GBR: Royal Society of Chemistry Gravetter, FJ & Wallnau, LB (2008) Essentials of Statistics for the Behavioural Sciences 6th edition. Belmont, CA: Thomson Higher Education. Jilka, M (2009) Assessment of phonetic ability. In G Dogil & SM Reiterer (ed), Language, Talent, and Brain Activity. Boston, MA: Walter De Gruyter. Kuhl, PK, Conboy, BT, Padden, D, Nelson, T &Pruitt, J (2005) Early speech perception and later language development: Implications for the “critical period”. Language Learning and Development, 1(3-4), 237-264. Ladefoged, P (2005) Vowels and Consonants, 2nd ed. Oxford, GBR: Blackwell Publishing. Mendenhall, W, Beaver, RJ & Beaver, BM (2009) Introduction to Probability and Statistics, 13th ed. Belmont, CA: Brookes / Cole / Cengage Learning. Probert, P (2010) Phonology. In EL Bakker (ed), A Companion to the Ancient Greek Language. Oxford, GBR: Blackwell Publishing, 85-103. Rathvon, N (2004) Early Reading Assessment: A Practitioner’s Handbook. New York: The Guilford Press. Slayden, G (2009) Central Thai Phonology, viewed 18 May 2011, http://www.thai-language.com/resources/slayden-thai-phonology.pdf Tsukada, K (2004) Cross-language perception of final stops in Thai and English: A comparison of native and non-native listeners. Proceedings of the 10th Australian International Conference on Speech Science & Technology, Macquarie University, Sydney, December 8-10, 2004, pp. 563-568. Tsukada, K & Roengpitya, R (2008) Discrimination of English and Thai words ending with voiceless stops by native Thai listeners differing in English experience. Journal of the International Phonetic Association, 38(3), 325-347. Tsukada, K (2009) Durational characteristics of English vowels produced by Japanese and Thai second language (L2) learners. Australian Journal of Linguistics, 29(2), 287-299. Read More