Pietilä, P. & O-P. Salo (eds.) 1999. Multiple Languages – Multiple Perspectives. AFinLA Yearbook 1999.
Publications de l’Association Finlandaise de Linguistique Appliquée 57. pp. 165–170.
AN ERROR ANALYSIS OF PHONETIC
TRANSCRIPTION: RESULTS FROM A PILOT STUDY
Pekka Lintunen, University of Turku
This article deals with phonetic transcription, more precisely phonemic or broad trans- cription, in which only the phonemes are transcribed. I have made an error analysis of the transcriptions produced by first-year university students of English. I also created a typology of errors to present the results. The results show that the students improved quite a lot in their transcription skills during a 14-week course. The most common error types are the same as in pronunciation. In further research I will compare individual students’ pronunciation and the transcription that they produce. The aim is to find out how trans-cription can be efficiently used in teaching second language pronunciation.
Keywords: phonetic transcription, error analysis, second language acquisition
1 THEORETICAL ASPECTS
Phonetic/phonemic transcription is interesting as a concept, since it is a writing system that tries to imitate speech as closely as possible. There are fundamental differences between speech and writing, but transcription tries to lie on the borderline. There are many kinds of transcriptions that differ in their level of abstraction and/or in their treatment of individual sounds. This article deals with phonemic or broad transcription. That is, transcription where only the phonemes are written and all other phonetic details of the pronunciation are excluded. Phonemic transcription exercises have traditionally been used as a means to teach English pronunciation – sometimes even exclusively. However, it is surprising that there is not much research done on transcription. Though phonetic symbols often exist in school books from the lowest levels, it seems that this resource may be underused at schools today. Children are not always taught how to read the instructions for pronunciation written in transcription. On higher levels,
transcription is a theoretical method that can be helpful in modern language laboratories when combined with other teaching techniques. In addition, transcription is essential when teaching phonetics. The advantage of transcription is that it gives the students the ability to write and speak about pronunciation on an abstract level, which can be required with reason as a part of a university degree in English.
As a part of the syllabus at the university level, transcription - based on student reactions – is often found very difficult, and students seem to have problems with associating correct sounds with the written symbols that represent the English sound system. What makes it even more difficult is the fact that many students have not yet acquired all the phonemes of English and cannot use them in their production.
The system of symbols used in teaching needs a lot of attention. The symbols have to be phonetically accurate, but at the same time also pedagogically helpful and easily legible, since transcription is used for example in schoolbooks and dictionaries. The guidelines of representing sounds with symbols have been established by the International Phonetic Association, but there are different ways of categorising sounds into phonemes and various principles that can be followed when symbols are chosen for these phonemes. In this article the approach is pedagogic rather than phonetic. What is under consideration is the way transcription can be used in the teaching of second language pronunciation. The first step in trying to answer this question is this pilot study on transcription.
2 THE PILOT STUDY
A study was carried out at the University of Turku. Material was collected from a spoken English course that was aimed at first year students. The course (28 contact hours) dealt with pronunciation, and transcription was used as one teaching method. Each student produced three phonemic transcriptions of short (ca. 10 sentences) passages of text during one term with a few weeks’ intervals. The transcriptions were made from a written text. The last transcription was the final test of the course. The subjects were 35 Finnish-speaking students, and they produced more than 3 000 errors in transcription. A typology of errors was created to present the results. The aim of the study was to find out what kinds of errors students make and, in addition, by analysing the errors to try to shed some light on the fact why students find transcription difficult.
The typology has been introduced and explained more thoroughly (for example category boundaries and the motivation behind them) in an earlier article (Lintunen 1999), but I will summarise the categories here briefly. The phonological system that the typology is based on is the latest version of Gimson’s system of English sounds (Cruttenden 1994), which is used in many pronoun-cing dictionaries. The first distinction was drawn between errors in the sound symbols, errors in stress, and other errors, viz.
punctuation or the use of slashes. The sound symbol errors consist of errors in vowel and consonant sounds. These errors were first divided into system and sound errors. The motivation behind this was that system errors were caused by the transcription system used, but in sound errors there was a possibility of a pronunciation error. System errors can be of two types: incorrect phonemic symbols (e.g. using existing symbols in incorrect combinations of diphthong or affricate symbols as in / / and /d/) or non- phonemic symbols (introducing a new symbol – such as /M/ or /o/ – to the transcription system). Sound errors had several subtypes, some of which were different in vowel and consonant errors. Errors in the number of symbols are quite straightforward: there can be either extra symbols or symbols missing. Quantity errors occur when there is either the same consonant transcribed twice as in /ll/ or the members of the tense-lax vowel pairs are mixed up with each other (e.g. // instead of /iw/). Quality errors consist of schwa (/°/) errors and fortis/lenis errors (e.g. /p/ instead of /b/ and /s/ instead of /z/) and other incorrect sounds (e.g. /æ/ instead of // or /s/
instead of / /) that could not be categorised under the types listed above.
The results show that the students learnt transcription quite well during the course. In the first text an error occurred in approximately every second word, but in the third text only in every fourth word. The major error categories could be predicted on the basis of pronunciation difficulties that are typical of Finns. The total number of segmental errors was 3126. There were more errors in the vowel sounds (63.0 %) than in the consonant sounds (37.0 %). Unsurprisingly, sound errors (84.4 %) were more frequent than system errors (15.6%). In system errors the largest type was incorrect phonemic symbols in vowel sounds (11.2 % of all errors). This suggests that the vowel symbols are more difficult to learn. Especially diphthong sounds were problematic through the whole course. This is due to the fact that different symbols are used in diph- and monoph-thongs. A common error was, for example, to mix the symbols in /a/, /†w/ and // and produce such symbols as /†/ and //.
The clearly largest error type was a full vowel replacing a schwa sound /°/ (25.1 % of all segmental errors). This kind of error is also very common in pronunciation (e.g. Lehtonen, Sajavaara & May 1977; Morris-Wilson 1992). These errors occur in weak forms or unstressed syllables, which are the stum-bling blocks of the English pronunciation for Finns. This error type was very common in such words as and, of and that (relative pronoun).
However, the subjects improved their transcriptions during the course. In the beginning 28.5 % of all errors were of this type, but in the final text 18.4 %. As it is well estab-lished that Finns often make this kind of error in pronunciation, the schwa sounds are emphasised in the teaching of pronunciation and transcription, too. The subjects learnt to use this sound more and more, but they still did not use it often enough. The following most difficult error types of the vowel sounds were other incorrect sound (7.5 %), a tense sound instead of a lax one (6.0 %), schwa instead of a full vowel (5.1 %) and a lax sound instead of a tense one (4.0 %).
The largest type of errors in the consonant sounds was a fortis sound instead of a lenis one (16.8 %). Again this is not surprising as Finnish does not have a distinction between fortis and lenis sounds, which makes errors in these sounds very common in pronunciation. The subjects did learn how to use these sounds more during the course, and there were also some traces of hypercor-rection in the final test, because many students began to use too many lenis consonants (especially the sibilant /z/), which was the opposite of the situation in the beginning. However, throughout the course this was the most common error type with the consonant sounds. The other frequent error types with consonants were other incorrect sound (5.2 %), sound missing (4.8 %), lenis instead of a fortis sound (4.5 %) and extra sound (3.5 %). The errors with the number of symbols reflect the frequent omissions of linking sounds and, on the other hand, the tendency to include silent consonant graphemes of the source text in the transcriptions (for example in know transcribed as /kn°/).
The subjects made a total of 301 stress errors. These were especially frequent in the beginning of the course. The reason behind many of these errors was that the subjects simply forgot to mark the main stress. Also in the beginning some subjects had the tendency of placing stress marks in various incorrect places, for example before syllabic sounds and in correct syllables but between wrong symbols (e.g. /s°pra’z/). The subjects had clearly problems with identifying syllable boundaries. The marking of the stress remained to be an important source of errors, but the subjects became more used to it. However, it is difficult to say to what extent the
errors made are in fact due to the trans-cription system used or to difficulties in pronunciation.
The other errors did not prove to be the most problematic ones. The use of punctuation was common at first, but after the subjects had been reminded not to use any punctuation marks in their transcriptions, they all learnt to follow this principle. Also the technical skill of the correct use of the slashes in broad transcription was acquired reasonably well. In the beginning it was difficult, because the subjects were not used to the technique of writing broad transcrip-tion. Mostly the subjects did not use slashes at all in the beginning. In the final test, however, all subjects used the slashes, but more than half of them used the slashes incorrectly (e.g.
separated every word with a slash, which is not the standard in continuous text). These other errors are naturally less important when transcription is used only as a pedagogic tool to improve students’ pro-nunciation, because they deal with technical details.
The reasons behind errors can be divided into (transcription) system internal and system external reasons. System external reasons are connected to pronunciation; some errors are clearly due to difficulties in producing words or sounds. System internal reasons cause errors in, for example, difficult symbols for sounds (e.g. sounds that are represented with different symbols in transcription and orthography) or some technicalities (e.g. the use of slashes). It is very difficult to say which motivation has caused more errors by looking at transcrip-tions alone. Moreover, in some errors both motivations may be involved.
The teacher has to choose whether to concentrate on the technique or the development of pronunciation when teaching transcription due to the fact that the teaching time is always limited. It is self-evident that transcription should be used as a means to aid pronunciation and not as a goal per se. Therefore, the optimal transcription system for Finns should be used. This would facilitate concentrating on the sounds involved instead of the technique. However, authorities (dictionaries, descriptions of the phonology of English etc.) are often aimed at an international audience and not made for Finns alone, which makes it more difficult in practice to teach a system created for Finnish learners.
3 THE FUTURE OF THE PROJECT
This project goes on, and more material is being collected. The material covers the first year of English studies ranging from the first pronunciation
test at the beginning of the spoken English course to the final exam in the spring term. Focus is widened from transcription alone to the area of pronunciation. This study has shown that transcription can be analysed and that the problem areas seem to be the same as in pronunciation. The development of pronunciation and the fact how different tests used in teaching (discrimination tests, phonemic transcription exercises) correspond to this development will be studied more closely. The study deals basically with the difference between perception and production; as it is generally assumed that perception precedes production, the difficult sounds (the ones that do not exist in the first language) should first be identified correctly in discrimination and thereafter in transcription and pronun-ciation. The transcriptions made from spoken and written texts will also be compared in order to explore whether the frequency of the error types depends on the source text. Also individual differences will be examined more closely.
The goal is to find out how phonetic transcription could more efficiently be used in the teaching of English pronunciation for Finns or should it be taught at all. At the moment transcription is not systematically taught before the university level, and it is probable that pupils (or even all teachers) cannot read – let alone produce – transcription in their books. Based on my own experience, I believe that learning to transcribe can improve (at least adult) pronunciation, because transcription makes the transcriber think about the target sound system more consciously. Future research will hopefully either confirm or reject this hypothesis.
References
Cruttenden, A. 1994. Gimson’s pronunciation of English. London: Edward Arnold.
Lehtonen J., K. Sajavaara & A. May.1977. Spoken English. Gummerus.
Lintunen, P. 1999. Analysing phonetic transcription: a typology of errors. In S-K.
Tanskanen & B. Wårvik (eds.) Proceedings from the 7th Nordic conference on English studies. Anglicana Turkuensia 20. University of Turku, Department of English, 125–132.
Morris-Wilson, I. 1992. English segmental phonetics for Finns. Helsinki: Finn Lectura.
