Introduction



Tomasz Ciszewski

Instytut Anglistyki

Uniwersytet Gdański

The English interdentals and the Polish learner:

perceptual difficulties and substitution patterns.

Introduction

Little seems more frustrating to the prescriptive ear of a Polish teacher of English phonetics than a mispronounced interdental. Aesthetic judgements aside, the problem is not uniquely Polish but it also torments other foreign learners of English all around the world. Many native speakers, however, feel equally uneasy about their own dental fricatives.

Inadequate perceptual clues on the one hand and relatively small contrastive potentials on the other, make English interdentals likely candidates for phonological obliteration, or, at least, some phonological restructuring. Given the fundamental auditory problem concerning the interdental fricatives, it is hardly surprising that native speakers of various dialects of English have developed certain substitution strategies, traditionally referred to as ‘ stopping’ (alveolar plosive) and ‘ fronting’ (labiodental fricative) (Wells 1982). Although both processes are argued to be spreading geographically across Britain (Trudgill 2001), it is unclear why they simultaneously appear in such geographically and linguistically remote dialects as Shetland[i] (van Leyden 2004) and Dublin (Hickey 1998), which adopt stopping, and Glaswegian (Wells 2004), in which fronting is increasingly popular, or Cockney (Wells 1982) and African American (Bailey and Thomas 1998), in which both substitutions are used.

The present study addresses the following questions concerning the perception and production of English interdentals by Polish learners:

• Which consonantal contrasts are specially difficult to perceive in isolated words?

• Does the non-phonetic context facilitate the perception of interdentals?

• What is the relation between the knowledge of the word and its correct perception/production?

• Is there any correlation between the perceptual difficulty of interdentals in certain consonantal contrasts and their substitution patterns? and finally

• What is (are) the most common substitution pattern(s) for interdentals in the word-initial position?

Three experiments were designed and carried out in November 2004 among a group of the first year students of a BA teacher training college. All participants were acquainted with places and manners of articulation of English consonants. Prior to the experiment, however, the students had not practiced the contrasts analysed in the present research. Their participation in the experiment was voluntary and anonymous. The number of participants varied in each experiment: (a) the perception experiment 1: 66, (b) the perception experiment 2: 116, and (c) the substitution experiment: 40.

Perception Experiment 1

The objective of the first experiment was to examine the perception of the voiceless dental fricative confronted with the labiodental fricative /φ/, the alveolar plosive /τ/ and the alveolar fricative /σ/ in a series of minimal pair tests. Eighty words were arranged in 20 quadruplets. Each of them contained the same vowel. The material contained both existent and non-existent words. For the latter, the most typical spelling was chosen and each minimal pair was invariably presented with identical vowels. Due to its defective distribution word-initially and the insufficient number of possible consonantal contrasts, [Δ] was not included in this experiment.

• thick fick tick sick

• thin fin tin sin

• theme feme team seem

• Thea fear tear sear

• thieve feeve teethe sieve

• thake fake sake take

• thed fed Ted said

• Thannet fannet tannet sannet

• thammy fammy Tammy Sammy

• thug fug tug sug

• thummy fummy tummy summy

• thar far tar Saar

• thart fart tart sart

• Thor four tore sore

• thaw faw taw saw

• thought fought taught sought

• thong fong tong song

• thog fog tog sog

• therm firm term serm

• thirst first tursed sirced

The informants were presented with 3 lists of 20 minimal pairs which obligatorily contained a form confronted with another word of the same quadruplet in the following configuration: List 1 /Τ/-/φ/ contrasts, List 2 /Τ/-/τ/ contrasts and List 3 /Τ/-/σ/ contrasts. Then, 20 words were read[ii]. The informants were asked to tick the word which they heard in each minimal pair. The actual choice of the form presented in each minimal pair was fully random in order to exclude any systematic patterning. The procedure was repeated with identical lists of minimal pairs but in the second attempt the selection of items was complementary to the first attempt.

The informants were not able to see the reader so as to exclude any visual information from the input. This was particularly important in the case of /(/–/φ/ contrasts which, according to some researches (Walden et al.1977), may be perceived correctly even by people with severe hearing loss. These are claimed to belong to different visemes, whose auditory recognition is strongly supported by non-phonetic, visual information.

Perception Experiment 2

This experiment utilized two perception tests by Baker (2001: 137, 142). The informants were given two lists of existent English words, each of which contained 6 minimal pairs. For each list 6 pre-recorded sentences were presented. The participants’ task was to correctly identify the word which was used in a particular sentence. After the auditory test the participants were asked to indicate the words with which they were not acquainted.

List 1 Recorded sentences

(target words in bold type)

1. sink / think I always sink in the bath.

2. mouse/ mouth What a small mouth!

3. tin / thin Don’t burn it. The saucepan is only tin.

4. taught/thought The teacher thought quickly.

5. moss/ moth Look at that moss on that stone.

6. fought/thought The two men fought very hard.

List 2 Recorded sentences

1. Ida/either I don’t like her sister, Ida.

2. day/they Day came suddenly.

3. dare/there's Jim dares his friend.

4. size/scythe That’s a very large size.

5. bays/bathe Did you sea bathe?

6. boos/booth The booth echoed loudly.

All the sentences were basically context-free which, at least in theory, eliminated any semantic information from the input.

Production Experiment: Patterns of substitution

This experiment was designed to compare the most typical patterns of misperception with the most typical patterns of mispronunciation and to check the frequency of various word-initial substitutions of the English dental fricative made by Polish speakers. The assumption was that the mispronunciation and misperception patterns should be rather congruent, i.e. the perceptually difficult contrasts will match the most frequent substitution patterns.

The informants were asked to read aloud a list of 20 words (including some nonsense words). No special preparation time was allowed.

thummy thought thirst thin thar

thieve thammy thick thong therm

theme thog thed Theo thaw

thart Thannet thake thug Thor

After the experiment, the participants were asked to indicate which words they knew by writing their Polish equivalents. This was meant to provide information concerning the students’ acquaintance with the form, which confirmed their previous exposure to an item, rather than the precise and correct meaning of the word. In this way, five words were singled out (>60% of correct answers): thick, thin, thieve, thought and thirst. These ‘core’ items were used as reference words in the analysis of the correlation between the acquaintance with the phonetic form and the substitution pattern.

Results

Perception Experiment 1

The overall misperception of the initial dental fricative in the minimal pair test was 27.75%. For each contrast, however, the rates of misperception differed considerably. When paired with /τ/ or /σ/, the dental fricative was correctly identified by 96.3% and 96% of participants, respectively. For the /(/-/f/ contrast, however, the correct perception rate for the dental fricative plummeted to a mere 56.7% and /f/ in this particular contrast was misperceived by as many as 22% of participants. The results are presented in Table 1 below.

Table 1. /(/ misperception rates in selected consonantal contrasts

| |/Τ/ |Other |TOTAL |

| | |consonant * | |

|/Τ/−/φ/ |43.30% |22% |65.30% |

|/Τ/−/τ/ |4% |9% |13% |

|/Τ/−/σ/ |3.70% |1.20% |4.90% |

* /f/, /t/, /s/ respectively

These results are perfectly congruent with Balise and Diehl’s (1994) findings, whose experimental study showed a particular difficulty for native speakers to recognize the /(/-/f/ contrast correctly, and Jongman et al. (2003), who claim that speakers of General American have greater difficulty in identifying correctly words starting with these fricatives even when they are presented in a semantically congruous context.

Another question which was independently addressed in perception experiment 1 was to what extent the vowel context in monosyllables can affect the perception of the initial consonant. Contrary to some suggestions made in literature, e.g. LaRiviere et al. (1975), who argued that the cues to the perception of labiodental and dental fricatives are located not in their noise portion but rather in the transition and the vocalic portion of fricative-vowel combinations, the results of this study are not conclusive in this respect. Although some vowel contexts have considerably higher recognition rates, e.g. /(/, no general patterns seem to have emerged.

Perception Experiment 2

The overall misperception rate for both the voiced and the voiceless dental fricative was 17.2% (compare with 27.75% in experiment 1).

Table 2. Misperception rates in selected consonantal contrasts

|/Τ/−/φ/ |24.10% |

|/Τ,Δ/−/τ,δ/ |28% |

|/Τ,Δ/−/σ,ζ/ |4.90% |

However, while the misperception rate for the /(/−/σ/ contrast remains constant in both experiments (4.9%), the results for the other two contrasts depart radically from those obtained in perception experiment 1. Instead of the total error rate of 65.3% for the /(/−/φ/ contrast, the present result decreases to 24.1%. Conversely, for the /(/−/τ/ contrast the misperception rate rises to approximately 28% (compare with 13% in the previous experiment). The reasons for this divergence, therefore, must not be entirely auditory.

Although the test sentences were all context-free, i.e. their semantic contents was in no way indicative of the actual choice of the word, they contained certain non-lexical clues which aided the correct recognition of an item. These recognition enhancement conditions included:

• assimilations and co-articulation effects

• stress and intonation

• the informants’ knowledge of the word

The voicing assimilation effect in [Ο:λωεΙσ(σιΝκ] (sentence 1, list 1), however, has not improved the incorrect recognition rate which was 5.2%. Similar results (4.9%) were obtained in perception experiment 1, where the target words were presented in isolation. The unanimously correct recognition of the target word mouth in sentence 2 (list 1) may have been potentially reinforced by the fact that the coronal /(/ facilitates a more extensive lip rounding of the glide element of the diphthong /αΥ/ (as argued in Gonet and Pietroń 2004). However, in contrast with /σ/, as in mouse, the same degree of lip rounding may be anticipated. Here the incorrect recognition figures were roughly the same as in sentence 1 (list 1), i.e. 6%, and, again, fully congruent with the overall misperception rate for the /(/−/σ/ contrast in perception experiment 1 (4.9%).

The stress and intonation could have hypothetically improved the correct recognition of the target word bathe in sentence 5 (list 2). This is due to the fact that in the sentence Did you sea bathe? the tonic stress falls on the syllable sea rather than bathe, which means that the intonation falls on the tonic and rises on the post-tonic bathe. In the sentence Did you see bays?, on the other hand, the fall-rise contour would have been realized over one syllable, i.e. the tonic bays. Regardless of the intonation clues, the incorrect recognition rate for this item is 4.3%, which again is only insignificantly different from the perception results in experiment 1. The relatively robust consonantal contrast between /(/−/σ/, therefore, gains little from various other auditory aids and its misperception rate invariably oscillates around 5%. The detailed results are presented in Table 3.

Table 3. Misperception rates (%) in experiment 2

List 1 List 2

|think |5.2 | |either |42.2 |

|mouse |6 | |they |40.5 |

|thin |17.3 | |there's |27.6 |

|taught |14.7 | |scythe |0 |

|moth |5.2 | |bays |4.3 |

|thought |24.1 | |boos |8.6 |

Under the assumption that the success in the perception of English interdentals is based entirely on their acoustic qualities, the results of both experiments should be fairly similar, if not identical, for each of the contrasts tested. Such convergence, however, does not obtain, as shown by comparing the perception of particular contrasts in isolated words with those in context-free sentences.

Table 4. Perceptual difficulty scale: Experiment 1 and 2 compared.

|Experiment 2 |Experiment 1 |

| Contrast* |% |% |Contrast** |

|scythe |size |0 |4.9 |ALVEOLAR FRICATIVE |

|bays |bathe |4.3 |4.9 |ALVEOLAR FRICATIVE |

|think |sink |5.2 |4.9 |ALVEOLAR FRICATIVE |

|moth |moss |5.2 |4.9 |ALVEOLAR FRICATIVE |

|mouse |mouth |6.0 |4.9 |ALVEOLAR FRICATIVE |

|boos |booth |8.6 |4.9 |ALVEOLAR FRICATIVE |

|taught |thought |14.7 |13.0 |ALVEOLAR PLOSIVE |

|thin |tin |17.3 |13.0 |ALVEOLAR PLOSIVE |

|thought |fought |24.1 |65.3 |LABIODENTAL FRICATIVE |

|there's |dares |27.6 |13.0 |ALVEOLAR PLOSIVE |

|they |day |40.5 |13.0 |ALVEOLAR PLOSIVE |

|either |Ida |42.2 |13.0 |ALVEOLAR PLOSIVE |

*The target words in bold type.

** The voicing distinctions have been ignored in this comparison.

The most interesting correlations, however, are revealed between the participants’ knowledge of an item and its perceptual ease. This is presented in Table 5 below. The figures in the left column represent the participants’ ignorance of the words in each pair. The percentage result was arrived at by adding the ‘unknown word’ indications for both items and dividing the result by the total number of participants. This procedure was meant to provide information about the participants’ overall degree of lexical uncertainty rather than the actual ignorance of a particular word in each pair. The figures in this column are arranged decreasingly. The figures in the right column restate the misperception rates from Table 4.

Table 5.

|Ignorance of items (%)| | |Misperception |

| | | |rates (%) |

|72.5 |booth / boos |8.6 |

|53.75 |moss / moth |5.2 |

|45 |size / scythe |0 |

|25 |bathe / bays |4.3 |

|12.5 |tin / thin |17.3 |

|8.75 |dares / there's |27.6 |

|6.25 |fought / thought |24.1 |

|2.5 |taught / thought |14.7 |

|1.25 |sink / think |5.2 |

|1.25 |Ida / either |42.2 |

|0 |Mouth / mouse |6 |

|0 |day / they |40.5 |

The more robust /(,Δ/−/σ,ζ/ contrasts display no particular regularities as they are distributed more or less equally in Table 5, i.e. there is no correspondence between the level of ‘unknown word’ indications and the perception results for the words tested in each pair. Surprisingly, the correct perception of the other contrasts /(/−/φ/ ανδ /(,Δ/−/τ,δ/ is inversely proportionate to the acquaintance with the item tested. This suggests that lexical confidence ‘switches off’ the phonetic monitoring and, hypothetically, a ‘probabilistic’ discrimination method is employed which in each pair selects either the lexically more frequent item or the one which seems more probable in a particular position in the sentence. This is especially evident in pairs: dares-there’s, Ida-either, day-they. Phonetic information, therefore, becomes to a large extent irrelevant. This observation explains the divergence of the results in both perception experiments.

Production Experiment: Patterns of substitution

Out of all 800 pronunciation attempts (40 participants x 20 items) 330 were incorrect. This amounts to the total mispronunciation rate of 41.25%. The uneven distribution of error types, however, seems more interesting than the overall mispronunciation rate. It appears that the substitution that outnumbers all other errors in the word-initial position is the /t/ replacement: 58.2% of all incorrect attempts. The runner-up is the labiodental fricative substitution: 26.1% of all incorrect attempts. Other consonant substitutions include

/(/: (5.2%), [((]: (5.2%) and /(/: (3.9%). Incidentally (3 attempts), the dental fricative was replaced by the palato-alveolar affricate /((/. There were also two isolated cases of the /ϖ/ and /τφ/ substitution. Surprisingly, there were no instances of the /σ,ζ/ replacement although many teachers will report this to be a noticeable problem[iii]. The distribution of error types is shown in Table 6 below.

Table 6. Word-initial substitutions patterns: general results.

|/τ/ |58.20% |

|/φ/ |26.10% |

|[τ?] |5.15% |

|/δ/ |5.15% |

|/Δ/ |3.90% |

|/τΣ/ |0.90% |

|/ϖ/ |0.30% |

|/τφ/ |0.30% |

In the group of reference (‘core’) words (thick, thin, thieve, thought and thirst), however, the types of error differ considerably, both from the general results and the non-core words substitution results, especially in the proportion of stopping and fronting preferences, as shown in the comparison below.

Table 7. Word-initial substitution strategies:

general results vs. ‘core’ / ‘non-core’ words

| |General |Core |Non core |

| | | | |

|Mispronunciation |41.25% |31.50% |44.50% |

|rates | | | |

| | | | |

|/τ/ |58.20% |47.60% |60.67% |

|/φ/ |26.10% |46% |21.35% |

|other |15.70% |6.40% |17.98% |

The overall mispronunciation rates bear out a rather trivial conclusion that the well-known words are slightly easier to pronounce than the unknown ones. Nonetheless, if both the semantic contents and the phonetic form of the ‘core’ words were known, the figure would not be significantly higher than 0%. In this light, the difference between the incorrect articulation rates in the ‘known’ and ‘unknown’ words (13%) remains insignificant. Remarkable differences, however, exist in the actual proportions of error types, i.e. the /τ/ and /φ/ substitutions.

Stopping or fronting?

The present study suggests that in the word-initial position the speakers of Polish predominantly select the stopping strategy (60.67%) and disfavour the fronting option (21.35%) in unknown English words. Thus, the ‘stopping’ procedure seems a phonologically natural solution. In familiar words, on the other hand, the two substitution strategies are equally popular and other mispronunciations are considerably rarer (6.4%). This means that, aware of the interdental problem, in a particular word Polish learners would make the ‘stopping’ and ‘fronting’ substitutions much more randomly. A further study, however, is required to show whether the same results obtain word-medially and finally.

These results differ from the perception results (experiment 1 above) in which the /Τ/−/φ/ contrast was the most difficult to perceive. It is surprising, then, that a perceptually easy consonantal contrast /Τ/−/τ/ does not prevent, or, at least, limit the scope of stopping.

Conclusions

The participants’ reactions to the experiments show their general awareness of the interdental articulation problem in English phonetics. After the experiments many of them commented on how difficult [(,(] sounds are in their subjective sense. On the other hand, most of these comments indicated a fairly good passive knowledge of dental articulation. The results of the three experiments discussed in this paper may be summarized as follows:

• In isolated words, Polish learners find the /Τ/−/φ/ contrast particularly difficult to perceive, whereas in context-free sentences the level of perceptual difficulty is more or less equal for both /Τ/−/φ/ and /Τ/−/τ/ contrasts. In the latter case (perception experiment 2), however, the correct perception is inversely proportionate to the knowledge of the word. As far as the /Τ/−/σ/ contrast is concerned, the misperception figures are not affected by the knowledge of the word or other phonetic clues such as assimilations, intonation or stress.

• The perceptual and articulatory difficulty do not coincide, at least for the word-initial contrasts. While the /Τ/-/φ/ contrast is most frequently misperceived, it is the alveolar plosive which serves as the natural substitution for the dental fricative word-initially.

• Finally, although in the group of ‘core’ words two substitution patterns (/φ/ and /τ/) seem to be equally popular, the phonologically natural substitution for the dental fricative in word-initial position in the case of ‘unknown’ words is definitely the alveolar plosive.

Notes:

-----------------------

[i] Unlike the relatively close Orkney.

[ii] Ideally, the words should be pre-recorded or presented as electronic dictionary entries (Jones 2003, for example). This was impossible due to the fact that a large portion of possible contrasts contained non-existent lexical items.

[iii] The alveolar fricative substitution may be either idiosyncratically bound to certain words, e.g. /(((((((/ and /((((/ and never (?) /(((sð/ or, more probably, it may coσ/ or, more probably, it may constitute a product of a two-stage assimilation of place and subsequently voicing: /Τ/→/σ/→/ζ/ ⁄ ___ /d/. A surprisingly similar process has been observed by Bailey and Thomas (1998) in African American (Memphis data) where bathroom and birthday are pronounced as [β(σρ(μ] and [β3?ρσδεΙ].

References:

Bailey, G. and E. Thomas, (1998), ‘Some aspects of African-American Vernacular English

phonology’. In S. Mufwene, J. Rickford, G. Bailey and J. Baugh (eds.), African

American English Structure, History and Use, London: 85-109

Baker, A. (2001), Ship or Sheep, Cambridge.

Balise R.R., and R.L. Diehl (1994), ‘Some distributional facts about fricatives and a

perceptual explanation’, Phonetica, 51: 99-110.

Gonet, W. and G. Pietroń (2004), ‘The Polish tongue in the English ear’. In Sobkowiak, W.

and E. Waniek-Klimczak (eds.) Dydaktyka fonetyki języka obcego w Polsce,

Tom III: Zeszyty Naukowe Instytutu Neofilologii PWSZ w Koninie: 56-65.

Hickey, R. (1998) The Dublin Vowel Shift and the historical perspective. In Fisiak, J, and M.

Krygier, (eds.), Trends in Linguistics, Studies and Monographs 112 Mouton de

Gruyter New York

Jones, D. (2003), English Pronouncing Dictionary, CD-ROM Version 1.0., Cambridge.

Jongman, A., Wang, Y. and B.H. Kim (2003), ‘Contributions of semantic and facial

information to perception of nonsibilant fricatives’, Journal of Speech, Language and

Hearing Research 46: 1352-1367.

LaRiviere, C., Winitz, H. and Herriman, E. (1975), ‘The distribution of perceptual cues in

English prevocalic fricatives’, Journal of Speech and Hearing Research 18: 613-622.

Trudgill, P. (2001), Sociolinguistic Variation and Change, Edinburgh.

van Leyden, K. (2004), Prosodic Characteristics of Orkney and Shetland Dialects. An

experimental approach, Utrecht.

Walden, B., Prosek, R., Montgomery, A., Scherr, C., and C. Jones (1977), ‘Vocalic transitions

in the perception of voiceless initial stops’, The Journal of Acoustical Society of

America. 57.2: 470-475.

Wells, J.C. (1982) Accents of English, Cambridge.

Wells, J.C. (2004) http:/www/phon.ucl.ac.uk/home/estuary/Glasgow.htm [2004-12-11]

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download