Wodzimierz Sobkowiak Wiesawa Ferlacka Calibrating the Phonetic Difficulty

Włodzimierz Sobkowiak & Wiesława Ferlacka Calibrating the Phonetic Difficulty Index Twenty English words of four different phonetic difficulty levels (as measured by the Phonetic Difficulty Index – PDI) were read in carrier sentences by 38 Polish learners of English aged 17 -18. A total of 617 word-readings yielded 1211 errors, for the grand mean of 1. 96 phonetic errors per reader per word. The primary aim of this experiment was to verify empirically the intuitively arrived at lexico-phonetic difficulty judgements encapsulated in the PDI assignment algorithm applied to the English lexicon. The secondary aim was to likewise verify PD intuitive ratings collected in 2000 from 208 university students of EFL. 1

Background • Phonetic difficulty is an important factor to be reckoned with in vocabulary teaching and learning. • There has been so far relatively little research on phonolexical difficulty of English vocabulary to foreign learners. • The PDI was conceived of to be a practical researcher's and teacher's tool to help measure phonolexical difficulty in the context of teaching EFL pronunciation to Polish learners. • Some empirical verification is necessary to calibrate the intuitively constructed tool to account for and predict the actual phonetic problems of learners. 2

2000 study In early 2000 PDI was validated experimentally by running a questionnaire asking the students of the first two years of English philology to rate 20 PDIstratified words on a four-level scale of pronouncing difficulty. Students were asked to judge the words' phonetic difficulty to beginning Polish learners of English. A full account of this study appears in my unpublished paper at this web address: http: //elex. amu. edu. pl/~sobkow/diffind 2. doc. 3

Results of the questionnaire 0 2 4 6 carry belief relax taxi defect almost debate kingdom oblige dissolve appear survive tired server awkward mother author southern youngster coloured mean=1. 45 mean=2. 03 mean=2. 15 mean=2. 46 The respondents' difficulty ratings correlate very highly with PDI (r =. 684) 4

2005 subjects The subjects in the present experiment were learners from a secondary technical school in Świnoujście, Poland. These students represent the beginner to low pre-intermediate level of English, with an average EFL learning experience of 5. 8 years, with 1 -2 English lessons a week. The sample consisted of 14 boys and 24 girls, totalling 38 learners aged 17 -18. 5

2005 questionnaire sample 1. Pallbearer - someone who helps to carry ('k. Vr. I) a coffin at a funeral 2. Hocus-pocus - an activity or a belief ('belfel) that you think has no value and is intended to trick people 3. Melt - if you melt into or against someone you relax ('relaks) as they hold you close in a romantic way 4. Hail - to signal a taxi ('taksi) or bus so that it stops for you 5. Defect ('defekt) - a fault in someone or something 6

Results (1) A total of 617 word-readings yielded 1211 errors (60. 55 per word on average) from 30. 85 readers, for the grand mean of 1. 96 phonetic errors per reader per word. The easiest words appeared to be: server, carry, youngster and taxi. The hardest: oblige, coloured and awkward. word errors readings error/reading student PDI rating 1 carry 22 36 0, 61 1, 46 0 2 belief 51 32 1, 59 1, 50 0 19 youngster 22 30 0, 73 2, 07 6 20 coloured 119 31 3, 84 2, 07 6 sum 1211 617 mean 60, 55 30, 85 1, 96 2, 02 7

Results (2) The product-moment correlation between the intuitive rating of PD made by students and the mean number of errors per word is 0. 59, i. e. highly statistically significant. This means that students had a good feeling for the actual relative PD of the twenty words "to beginning Polish learners of English". The most seriously student-underrated words in terms of PDI were oblige and coloured, two of the three most phonetically difficult words. The most seriously student-overrated words were server, southern and youngster. 8

Results (3) The product-moment correlation between the reading errors and the PDI value for the twenty words is 0. 21, which is statistically not significant. This means that in its present form the PDI metric does not capture correctly the actual phonetic errors made by beginning-to-intermediate Polglish learners. The PDI metric needs calibrating to better account for and predict the actual phonetic errors made by Polglish learners at this level. 9

Calibration: example The PDI algorithm predicts no phonetic difficulties for relax, i. e. PDI(relax)=0. The actual number of errors counted for this word in the experiment is about two. Sixteen of the twenty-five erroneous readings of this word show the same pattern: /'relaks/, which scored three errors: (1) wrong stress, (2) wrong vowel in the first syllable, (3) wrong vowel in the second syllable. These difficulties will need to be accounted for in the PDI algorithm, at least in its beginner-Polglish implementation. 10

Conclusion Phonetic difficulty (both subjective and objective) of English to Polglish learners should be comprehensively studied in various contexts and settings: different proficiencies, age-groups, teachers, geographical locations, didactic approaches and methods, textbooks and dictionaries, multimedia, etc. There has been surprisingly little research along these lines. A suitably calibrated, flexible PDI can be instrumental in studying English texts and wordlists of all sorts for their phonetic difficulty in the Polish EFL setting. 11

Thank you! Full text of this presentation to be found at: http: //elex. amu. edu. pl/~sobkow/ 12