Phonological Factors Affecting Taiwanese EFL Students’ Predictions of Voiced or Voiceless Interdental Fricatives in Unfamiliar English Vocabulary

全文

(1)Phonological Factors Affecting Taiwanese EFL Students’ Predictions of Voiced or Voiceless Interdental Fricatives in Unfamiliar English Vocabulary Wayne T. Schams National Pingtung University of Education Abstract This study analyzes Taiwanese university EFL students’ predictions of voiced and voiceless interdental fricatives in unfamiliar vocabulary words. The subjects comprised 47 non-English major undergraduate students in a national university in southern Taiwan. From a list of thirty vocabulary words containing interdental fricatives (ten each in onset, mid, and coda positions), students were asked to guess whether the interdental fricatives in each word were voiced or voiceless and also provide a reason for their guess (from a multiple choice list of possible reasons). The results showed that the highest percentage of correct guesses (94.5%) occurred with words containing voiceless interdental fricatives in coda position, supported by McGuire’s (2003) contention that voiced interdental fricatives are far less frequent than voiceless interdental fricatives but contradicting his claim that voiceless interdental fricatives are more stable than their voiced counterparts when they occur in onset position. The subjects had the greatest difficulty predicting voiced interdental fricatives in coda position and voiceless interdental fricatives in onset position (each with an accuracy rate of 34.9%). These and other findings are discussed, as are pedagogical implications and suggestions for additional research. Keywords: interdental fricatives, Markedness Differential Hypothesis, Rule-based Theory, Optimality Theory. 240.

(2) Phonological Factors Affecting Taiwanese EFL Students’ Predictions of Voiced or Voiceless Interdental Fricatives in Unfamiliar English Vocabulary Introduction The voiced and voiceless English interdental fricatives, [ð] and [θ], respectively, are among the rarest phonological sounds that occur in the world’s languages and are also arguably among the last and most difficult sounds to acquire be it in a language that is learned as a L1 or a L2. The physical articulation of these sounds is visually unsightly and undoubtedly considered impolite if not downright vulgar by many cultures throughout the world. Perhaps this social stigma associated with the physical production of these two sounds is partly to blame for why they are so difficult to acquire by speakers who are learning them in a language that is not their mother tongue. However, the absence of these sounds in most languages undoubtedly is a crucial factor in the challenges associated with acquiring them. Indeed, these sounds are quite regularly substituted with sounds in one’s native language but with similar points or manners of articulation. There are relatively few studies focusing exclusively on the interdentals. However, it is no surprise that those which do deal with interdental fricatives focus primarily on what sounds are substituted for them among L2 learners and the phonological rationale for making such substitutions (e.g. Ritchie, 1968; Ahn, 2003; Gough, 1999; Lombardi, 2003; Lee, 2006). Yet, as noted by Lee (2006), English spelling does not make a distinction between the voiced and voiceless interdental fricatives; that is, both of them are orthographically realized as “th.” That being the case, it is natural for ESL/EFL learners to encounter pronunciation difficulties when learning new vocabulary containing the interdental fricatives. Curiously enough, though, I have not found any research studies that deal with this issue. This study, then, is an attempt to help fill this gap. More specifically, the aims of this study are to: 1) see how successful EFL students in Taiwan are at correctly determining whether interdental fricatives in new and unfamiliar vocabulary words are voiced or voiceless; and 2) analyze what factors affect EFL students’ choices as to how they think such interdental fricatives should sound like and be pronounced. Literature review Even among native speakers of English, it has previously been pointed out that the voiced and voiceless interdental fricatives are the last sounds to be mastered and the ones most often substituted (Menyuk, 1968). More specifically, Menyuk notes that it is the [+strident] and [-strident] features which distinguish [θ] and [ð] from [s] and [z], respectively, and which native English speakers master last. Other researchers (e.g. Schmidt, 241.

(3) 1987) concur with this analysis and the principle is also reflected in Eckman’s (1977) Markedness Differential Hypothesis. It must further be noted that most of the more recent studies that analyze substitution of interdental fricatives among L2 learners approach it from the framework of McCarthy and Prince’s (1993) Optimality Theory, or OT (e.g. Lombardi, 2003; Lee, 2006). In order to see why the problem of interdental fricatives among L2 learners whose native languages do not contain such sounds has typically been handled by OT (particularly in substitution-related studies), it is useful to go back a bit further to see how it was dealt with prior to OT. Chomsky and Halle’s (1968) adherence to the idea of universal marking conditions prompted them to claim that since [s] is the most unmarked continuant, it will universally replace the voiceless English interdental fricative [θ]. However, this notion is not supported by the data. Although the theory holds true for many languages (e.g. Japanese, French, German, etc.), it does not hold true in many other languages (e.g. Russian, Dutch, Mandarin, etc.). Similarly, Rule-Based Theory makes it no easier to explain interdental fricative substitutions with across-the-board explanations since, if children have no interdental fricative sounds in their L1 inventories, then they have no rules at their disposal for substituting them with sounds that they do have in their L1 (e.g. [t] or [s]). That is to say, for languages that have both the [t] and [s] sounds, we are hard pressed to explain from a rules-based approach why it is that some languages prefer the former while other languages gravitate toward the latter, as just noted above in reference to Chomsky and Halle’s universal grammar theory. It is problems such as these that have caused many researchers to look to OT as a satisfactory vehicle for explaining such discrepancies. For example, Lombardi (2003), Ahn (2003) and Lee (2006), among others, all agree that L1 constraint ranking is the common denominator in accounting for L2 interdental fricative substitution. However, there is some disagreement over the precise ranking of such constraints among various languages. Nonetheless, OT has to date been the most viable means of explaining L2 interdental fricative substitution. Methodology The present study examines interdental fricatives from a different angle. My aim here is not so much to see what sounds L2 English learners substitute interdental fricatives with but rather analyze how well they do at predicting whether “th” sounds in unfamiliar words are voiced or voiceless and what the phonological reasons for their decisions are. The subjects in this study were comprised of 47 university undergraduate students in Taiwan, including 33 females and 14 males, in an elective English course taught by the researcher. They were all non-English majors and most had relatively low-level English ability. The 242.

(4) subjects ranged in age from 18-23 with a mean age of 20.1. The procedures were as follows. In order to determine the subjects’ choices of voiced or voiceless interdental fricatives in unfamiliar words, it was first deemed necessary to compile a list of words containing such sounds. It was further determined that the sounds in question should occur in as wide a range of environments as possible. Accordingly, factors in addition to the different sounds themselves (i.e. voiced or voiceless) included their position within the word (i.e. word-initial, medial or final), stress placement, adjacent vowels (and consonants), syllabicity (i.e. number of syllables), and similarity of parts of the words to other words that the subjects already know or would be more likely to know. These factors were not merely selected randomly. They largely correspond to the potential factors McGuire (2003) adopted in his study of interdental fricatives, including such phonological factors as preceding and following segments, prosodic context (i.e. onset, mid, or coda position), voicing, etc. as well as prosodic factors like syllable stress. McGuire’s (2003) study also incorporated social factors about the subjects (e.g. age), although such factors were not analyzed in the present study. With the factors in McGuire’s study as a base, then, it was believed that these factors might have some bearing on the subjects’ choices. However, it was believed that the subjects would have little background in phonology and phonological processes. Hence, rather than asking them to provide open-ended responses as to their reasons, it was deemed appropriate to put these factors in a multiple choice-type format (see the appendix) in order to gain more concise and succinct findings than what otherwise might be obtained. As pertaining to the wordlist, the researcher was unable to find any similar studies and hence was unable to obtain any preexisting wordlist used to test predictions of interdental sounds. With the above-mentioned factors in mind, the researcher therefore compiled a list of thirty words. Ten words contained interdental fricatives in the onset position, ten in a medial position, and ten in the coda position. Within the two categories in which the interdentals appeared in mid and coda positions, care was taken to assure an equal balance between voiced and voiceless ones (e.g. five voiced and five voiceless ones in mid position; five voiced and five voiceless ones in coda position). However, all ten of the interdentals in onset position were voiceless. The reason for this is that the only voiced interdentals the researcher could find in onset position were in common function words (e.g. there, that, this, etc.) that most of the students would most likely already know. Since the aim of the study was to test unfamiliar words, it was decided to use only voiced interdentals for the ten in the onset position category. Fifteen of the thirty words were selected because segments of them containing the interdentals bear resemblance, to some degree or another, to words the students would likely already know (e.g. the word brothel differs from brother and smother differs from mother by only one letter). Some of the other words among these fifteen have segments that are 243.

(5) similar to other words that are less obvious but still present (e.g. the word scathing contains the word thing and the word thanatology has than in its onset). Of the fifteen words chosen with such similarities (see Table 2), three contain interdentals that share the same sound as the word that would most likely be associated with them (e.g. in smother, the interdental fricative is voiced as is the “th” in the similar term mother). However, in the other twelve, the interdental fricative has the opposite sound of what would be predicted from the associated word (e.g. it might easily be predicted that the “th” in brothel would be voiced because the “th” in the similar-looking word brother is voiced). Finally, attention was also given to the vowels (and in a few cases, consonants) adjacent to the interdentals. An attempt was made to select words with a relatively equal balance between tense and lax vowels flanking the interdental fricatives. The words also represent various stress placements and syllabicity (the number of syllables among the thirty words ranged from one to five). The complete survey, then, in addition to asking about the respondents’ gender and age, included the 30 words listed randomly, a space for the subjects to indicate whether they think the “th” in question is voiced or voiceless, and a space for them to indicate a reason for their choice. Six reasons are listed at the top of the survey with an option for respondents to specify any additional reason not listed (see appendix). One week prior to administering the surveys, the researcher, verified with the subjects that they did not already know the thirty words being tested. The surveys were then administered during class period one week in the spring semester of 2009. The stress placement and vowel length in each word were provided and clearly explained immediately before administering the survey and the subjects were told that using dictionaries to help them with the survey was not permitted. After the subjects completed the surveys, they were collected and analyzed. Results & Discussion The results of the survey revealed widely divergent findings. In regard to the three general categories (i.e. in which the interdental fricatives were in word-initial, mid and final positions), which corresponds to the first reason in the survey, the subjects performed best at correctly predicting voiceless interdentals in word-final position (with a correct prediction rate of 94.5%). They showed the most difficulty predicting voiced interdentals in word-final position and voiceless interdentals in word-initial position—both with a correct prediction rate of 34.9% (see Table 1). This finding is perhaps explained due to low. Table 1 Subjects’ correct predictions (in percent) of interdental fricative (IF) pronunciation in 244.

(6) unfamiliar vocabulary Onset position Voiced IF. N/Aa. Medial position 60.3. Coda position 34.9. Voiceless IF 34.9 39.4 94.5 a No voiced interdental fricatives in onset position were included in survey items due to the difficulty of obtaining new and unfamiliar words containing this feature.. frequency of interdentals, as pointed out earlier. According to the Hoosier Mental Lexicon corpus, the interdentals have two of the lowest frequencies among all English sounds, as has already been noted earlier. More relevant to the above finding is that the voiced interdental fricative is even far less frequent than its voiceless counterpart (McGuire, 2003). This being the case, the subjects are far more likely to encounter words with voiceless interdentals in coda position than voiced ones. If they were unable to make correct predictions based on similarities between the words in this study with words already in their mental lexicons, then the odds worked in their favor of making a correct prediction anyway. A further explanation might be that most words with a voiceless interdental fricative in coda position tend to be monosyllabic and more closely resemble words that students likely already know. Also, the final “e” that is typically added to words that have the voiced interdental fricative in coda position (which, as just noted above, is far less common) would be rarely encountered by low level EFL students and thus be more confusing and/or less understood from a phonological point of view. Data from this study shows evidence that many respondents didn’t realize the final “e” in words ending in a voiced interdental sound is silent. In nearly one-third (29.9%) of such cases, the subjects chose reason number five (i.e. that the vowel following the “th” influenced whether that interdental fricative is voiced or voiceless) as a reason for their prediction. This is obviously incorrect because the final “e” in these words is not phonetically realized. In regard to the second reason listed in the survey (i.e. that the word in question contains a segment resembling a likely more familiar word), it was found that the subjects generally predicted the interdental fricative sound (i.e. voiced or voiceless) that is also present in the segment of the word resembling a more familiar word. For example, the second item in the survey is hearth. The more familiar word earth obviously contains a voiceless inderdental fricative in the coda. Hence, the supposition was that the subjects would be more likely to Table 2 Items with segments similar to more commonly recognized words 245.

(7) Item scathing [+voice]. Similar segment thing [-voice]. Correct predictions (%) 27.7. Reason 2 for prediction (%) 21.7. thanatology [-voice]. than [+voice]. 36.2. 3.0. atheroma [-voice]. the [+voice]. 17.0. 15.2. thistle [-voice]. this [+voice]. 36.2. 21.7. thesaurus [-voice]. the [+voice]. 23.4. 8.7. theurgy [-voice]. the [+voice]. 31.9. 6.5. wherewithal [+voice]. with [-voice]. 53.2. 28.3. thanatopsis [-voice]. than [+voice]. 31.9. 10.9. brothel [-voice]. brothe(r) [+voice]. 31.9. 13.0. thistly [-voice]. this [+voice]. 42.6. 23.9. ethereal [-voice]. the [+voice]. 25.5. 15.2. thane [-voice]. than [+voice]. 23.4. 23.9. hearth [-voice]a. earth [-voice]a. 91.5b. 21.7. girth [-voice]a. (b)irth [-voice]a. 97.9b. 15.2. smother [+voice]a mother [+voice]a 74.5b 39.1 Note. Bracketed features are for the interdental fricative sounds within the items and “similar segment” words. a In these three, the phonological feature of the interdental fricative in the item and similar segment are unchanged. bThe overall average percentage of correct predictions for these three items is 87.9. For the first twelve items, the overall average percentage is 31.7.. 246.

(8) predict that the “th” in hearth would also be voiceless based on the similarities between them (see Table 2 for a list of the fifteen items and the results). Of the fifteen items (out of the thirty total) containing such segments, three of them contained “th” sounds that were the same in the segments of familiar words (as just noted above with the earth/hearth example). However, twelve of those fifteen words contained interdental fricatives that were in fact opposite of the expected sounds in the segments with more familiar words. For example, the first item in the survey is scathing. The obvious segment is thing, where the “th” is voiceless. However, the “th” in scathing is voiced. As might be expected and can be seen in Table 2, for those examples in which the interdental fricative sounds were unchanged, correct guesses were significantly higher (87.9%) compared to when the sounds were different (31.7%). Such a wide ranging difference suggests that the subjects were likely basing their decision in large part on comparing those similar-looking segments with words they already know (i.e. reason number two). Yet, for those fifteen items, reason number two was listed only 18.7% of the time. Obviously, some of those 15 words had segments that more naturally catch the eye (e.g. scathing, wherewithal, etc.) and indeed it was such words which had the highest number of responses with the second reason listed. For many words, the segments were less obvious (e.g. atheroma, theurgy, etc.) and this may at least partially explain the above discrepancy. Another more probable explanation is that two of the last three items in Table 2 (i.e. those in which the interdentals were predicted most accurately) contained voiceless interdentals in coda position. In all of the other items, the voiced or voiceless interdentals were in onset or mid position. This finding further reinforces the data in Table 1 showing the highest accuracy in predicting voiceless interdental fricatives when they appear in coda position. The third reason in the survey dealt with stress placement. The general results for the whole thirty items revealed that only in 12.2% of the possible cases did subjects indicate that the word’s primary stress was at least one of the reasons, if not the only reason, influencing whether they thought the “th” in question was voiced or voiceless. This translated into a 38.5% accuracy rate at correctly predicting the “th” sounds among those who responded that stress placement plays a role in whether the interdentals are voiced or voiceless. An interesting related finding was that among these same subjects, nearly half of them (42.6%) indicated that in the words that had three or more syllables, stress placement was a determining factor. A breakdown of the results showing items with differing numbers of syllables is shown in Table 3. As seen, the general rule was that the correct prediction rate gradually decreased as the number of syllables increased. Also, the greater the number of syllables, the more responses there were indicating that reason three was a factor in their predictions about the interdental sounds. However, the only general constraint regarding voiced or voiceless interdental fricatives as they pertain to stress is that an intervocalic 247.

(9) Table 3 Effect of perceived importance of stress placement and syllabicity on interdental predictions. Correct predictions (%)b. One (11)a. Number of syllables in test items Two(10)a Three(5)a Four(3)a. Five(1)a. 56.7. 54.3. 36.2. 44.3. 30.5. Reason 3 indicated as 9.7 10.7 16.1 19.6 19.6 reason for prediction (%)_______________________________________________________ a Numbers in parentheses indicate the number of test items with that number of syllables. b Correct predictions are listed for all respondents, not only those who listed Reason 3.. interdental fricative is voiced if the syllable that follows is unstressed (i.e. V [ð] V [-stress]), although this constraint is occasionally violable (e.g. brothel). The length of vowels appearing either immediately before or after the interdental fricative (reasons four and five in the survey) was also a factor. However, subjects indicated that this was a factor twice as often in the case of vowels following the “th” (26.4%) compared to when vowels precede the “th” (13.2%) when the “th” appears in either onset, mid, or coda position. Yet, the only apparent constraint here involves interdental fricatives in coda position. That is, vowels that precede a voiced interdental fricative in coda position are held slightly longer than vowels that precede the voiceless interdental fricative. An example from the survey can be seen with item numbers 5 and 26 (i.e. the vowel “a” is held longer in swathe than it is in swath). However, although vowel length was considered a factor in 39.6% of the overall cases, only in 8.7% of the cases was it indicated as being a factor in words where the “th” was in coda position. This also reflects the relatively low percentage of correct predictions (34.9%) of voiced interdental fricatives in coda position. The sixth reason in the survey was concerning the occurrence of a consonant immediately following the interdental fricative. This phenomenon occurred only in items 10 (rhythmic), 13 (loathful), and 29 (wreaths). While 36.2% of the possible responses for this particular reason indicated that it was a factor influencing whether the “th” in these items is voiced or voiceless, 44.7% of the predictions regarding whether the “th” in these three items were voiced or voiceless were correct. This latter statistic would undoubtedly be higher if not for the fact that one of these three items (wreaths) contains an inflectional affix which alters the “th” from the voiceless coda in the root to the voiced form. This example 248.

(10) in particular is an anomaly since most other words with similar structure (e.g. breath/breaths, death/deaths, depth/depths, etc.) retain the voiceless interdental fricative. Finally, an open-ended seventh reason was listed if subjects guessed the “th” in question to be voiced or voiceless for some other reason and they were asked to specify the reason. Only a few responded by selecting “other” and wrote something like “it sounds natural” or “it intuitively sounds correct” without giving more specific phonological reasons. Since only a few such responses were made, they were discounted as being insignificant and not shown in the results given in the tables. Conclusion The results of this study tend to show that for low level EFL students, predicting whether the highly marked interdental fricatives in unfamiliar vocabulary words are voiced or voiceless is quite voluble and in many instances a hit-and-miss affair. In terms of the position of interdentals within words, this study showed that voiceless interdentals in coda position were highly predictable and thus assumedly more stable. Yet, this finding contradicts that of McGuire (2003) who claimed that voiceless interdentals are more stable in onset position due to a smaller chance of perceptual confusion. However, his study focused on native speakers of African American English while the present study examined EFL students in Taiwan. More empirical studies in cross-linguistic settings should be conducted so that interdental prediction rates can be compared among English learners in different cultural contexts and language backgrounds. This study also confirms that low-level EFL learners’ knowledge of phonological processes and the interrelationship of phonological sounds within words is generally quite weak, despite having a basic grounding in English phonetics (e.g. most have learned the K.K. phonetic system which is commonly used in Taiwan). Perhaps better training not only in phonetics but also phonological processes and relationships would help students to better predict not only interdentals but also other sounds in new or unfamiliar vocabulary in a more systematic way. This is a subject that should be further explored in future studies. Finally, it is hoped that with greater understanding of how students learn interdental sounds and the specific challenges they face in mastering them due to L1 interference, it can help educators know better how to teach them more effectively.. 249.

(11) References Ahn, S.C. (2003). English interdental substitution. English Language and Literature, 49(5), 981-1004. Chomsky, N. & Halle, M. (1968). The sound pattern of English. New York: Harper & Row. Eckman, F. (1977). Markedness and the contrastive analysis hypothesis. Language Learning, 27, 315-330. Gough, D. (1999). L2 interdental substitution. Linguist List, 10, 662. Lee, J. (2006). Typology of interdental fricatives with reference to loanword adaptation. Studies in Phonetics, Phonology and Morphology, 121, 127-148. Lombardi, L. (2003). Second language data and constraints on manner: Explaining substitutions for the English interdentals. Second Language Research, 19(3), 225-250. McCarthy, J. & Prince, A. (1993). Prosodic morphology I: Constraint and interaction and satisfaction. Ms. University of Massachusetts. McGuire, G. (2003, May 16). Realizations of interdental fricatives in Columbus AAVE. Paper presented at The Ohio State University Colloquium Fest. Retrieved April 12, 2009, from http://www.ling.ohio-state.edu/~mcguire/Interdentals/Handout.doc Menyuk, P. (1968). The role of distinctive features in children’s acquisition of phonology. Journal of Speech and Hearing Research, 11, 138-146. Ritchie, W.C. (1968). On the explanation of phonic interference. Language Learning, 18, 183-197. Schmidt, R. (1987). Sociolinguistic variation and language transfer in phonology. In G. Ioup and S.H. Weinberger (Eds.), Interlanguage Phonology (pp. 365-377). Rowley, MA: Newbury House Publishers.. 250.

(12) Appendix Phonology Survey Personal information: Gender (male or female): _______________ Age (in years): _______________ Directions: The questions below contain vocabulary words with “th” in them. Read the words carefully and decide if the “th” is voiced (i.e. [ð]) or voiceless (i.e. [θ]). Put a check under the appropriate column. Then in the column on the far right side, write the number that corresponds to the key below in regard to the reason why you checked the answer that you did. 1 = Based on the position of the “th” in the word (e.g. front, middle, final) 2 = Based on the fact that a part of the word containing “th” and nearby letters are similar to another word that I already know 3 = Based on the position of the word’s primary stress 4 = Based on the length of the vowel immediately before the “th” 5 = Based on the length of the vowel immediately after the “th” 6 = Based on the consonant immediately after the “th” 7 = Other (please specify) Word. Voiced [ð]. Voiceless [θ]. number) 1. scathing 2. hearth 3. thanatology 4. atheroma 5. swathe 6. thein 7. tether. 251. Reason (write.

(13) 8. monolith 9. thistle 10. rhythmic 11. seethe 12. thesaurus 13. loathful 14. girth 15. theurgy 16. smother 17. scythe 18. thyroid 19. wherewithal 20. froth 21. thanatopsis 22. brothel 23. writhe 24. thistly 25. ethereal 26. swath. 252.

(14) 27. thane 28. wither 29. wreaths 30. thixotropy. 253.

(15)