PTLC2005 Ole Stig Andersen Correction Typology:1 A Typology of Pronunciation Corrections Ole Stig Andersen, Copenhagen To widen teachers' correction arsenal and general understanding of what correction is all about, a typology was developed for use in training courses for Teachers of Danish as a Second Language for Adult Immigrants, where teacher and students often do not have a common language, dramatically so at beginners' levels. Ideally this typology should be supplemented with a typology of pronunciation mistakes. 0. GENERAL CONSIDERATIONS You can only correct another adult's pronunciation, student or otherwise, a few times before it becomes socially awkward since correction is also a status interaction requiring mutual succes. If the correction doesn't succeed, both parties may lose status in their own and/or in the spectators' eyes, since neither is up to the task, the student because he "can't" pronounce the correct sound, the teacher because she can't supply an efficient intervention. A teacher quickly learns to understand the intention behind the students' mistakes. You will not teach Arabic students, e.g., for long before you know that when they say [u] they might also mean [y]. In fact, understanding the intention is a prerequisite for correcting. You can only correct what you have understood. Correction is thus meaningless from at communication point of view. Indeed the teacher herself or the textbook has often supplied the students' sentences. So in a sense you could say that the students are teaching the teacher to understand their mistakes. This effect can be diminished by increasing the amount of unplanned communication in the class. It also predicts that learners acquire a better pronunciation if they have to speak Danish in real communications than if they have the opportunity to teach their teacher. The general inefficiency of prevalent pronunciation correction methods has led to its relegation to the periphery of teaching Danish. It has also led to a situation where teachers have pet mistakes that they do correct—often in vain—and a whole host of other mistakes which they may not even notice. 1. NATURAL (NAIVE) METHODS These methods are by far the most widespread in the classroom. Repetition is part of normal interaction skills and does not require any specialized knowledge of the description of the language. Teachers will—almost by default—correct the student's mistake by repeating the target (often the whole word or the syllable). In some cases this approach works, but in most cases it does not, and the teacher generally has no idea why which. In real class this impasse often leads to absurd sequences like Teacher: kylling ["chicken"] Student: killing PTLC2005 Ole Stig Andersen Correction Typology:2 Teacher: No: kylling Student: killing Teacher: OK Repetition blends into explicitation of target sounds by speaking slower, louder or more "distinctly"; by prolonging or isolating the problematic sound(s); prominence by applying non-standard stress. Some types of explicitation seem to be part of natural language skills whereas others are extensions from learning to read and writr the mother-tongue, a context in which it is reasonable to stress an unstressed vowel to make it more prominent by using "pedagogical stress". The unduly stressed syllable will "fall back" to its natural unstressedness because the mother-tongue pupil commands the prosodic system. But a foreign student with no grasp of Danish prosody will not be able to distinguish pedagogical stress from correct stress. This is aggravated in a highly stress-timed language like Danish, which tends to have the same time distance between strong stresses in a sentence irrespective of the number of intervening weak stresses, yielding a large number of radical word reductions, in contrast to what you'll find in syllable-timed languages. E.g. in a sentence like "Jeg skal i skole ("I'm going to school", lit. "I shall in school") pronounced ['ja sga i 'sgool] many students will leave out the "i", some for grammatical reasons others for phonetic. Teachers will often try to correct by making the missing element explicit with pedagogical stress, leaving the student with a somewhat correct "i", but a wrong prosody. A far better way to point out the lacking sound is backward correction: Start with the last strongly stressed syllable and build up the whole utterance by adding weak syllable after weak syllable. In this way you preserve correct stress all the way through the correction sequence. 'sgool i'sgool sgai'sgool 'jasgai'sgool This approach is also the best choice to give longer sequences a natural prosodic flow, far better chopping up the utterance in the direction of writing. 2. INTELLECTUAL METHODS These methods intend to give the students access to some of the teacher's world of sound descriptions and assume that learners can guide their sound production by conscious constructs. Learning to read and and write basically implies acquiring a model or theory of the language. Through prior education or during the training teachers acquire a more refined version of the alphabet theory, a more scientific version. A number of highlights, lists, charts, standard descriptions, pet examples, etc from these two complexes, the school model PTLC2005 Ole Stig Andersen Correction Typology:3 and the university model, filter down into the teaching of Danish to adult foreigners. Typical examples are Articulation instruction, mainly used for lip and tongue position, either by picture, by explanation or by explicit demonstration (which may in some cases transgress the student's personal space and feeling of decency. After all the speech organs are also erogenous mucuos membranes.) Articulation instruction includes a host of idiosyncratic "tricks" og micro-gymnastic exercises, use of mirrors to watch the mouth, etc. If for example you want to teach the Danish [y], you might tell an Arabic student to pronounce an [i] AND round (or even protrude!) the lips. This is fully in line with the theoretical decription of "y" a rounded, high, front vowel. But if the student cannot hear (or rather, categorize correctly) the difference, it is in fact possible to compensate with the tongue so that something [i]-like comes out although the lips are rounded. But it is possible to use minimal pairs to test whether a certain pronunciation problem is a hearing/interpretation problem or an articulatory problem. Another theoretical construct often employed is letter-sound correspondences, or "pronunciation rules", which are very numerous and complicated in Danish. And transcription may be used, putting en extra leeaning burden on the student, if done with special symbols. Teachers will also routinely use analytical concepts like vowel and consonant (often meaning the letter, not the sound), front and back tongue, syllable, stress, melody/intonation etc which only have meaning for few of the students. A special variety of the intellectual methods is comparisons with students' languages where you contrast features of the student's own language(s) with Danish. The contrastive approach probably has more clout in teaching grammar, and it is also difficult to use in multinational classes. 3. "MANIPULATORY" METHODS vary the corrective input according to the mistake made. If the student pronounces a vowel too high, you correct with a vowel that is too low. This approach aims at giving the learner subconscious information as to which direction in the auditory space the next attempt should take. A speaker of Spanish may pronounce Danish [v] as something Danish ears hear as [b]. You then correct with [f], which underscores the labio-dental and fricative character of the target sound though not its voicedness. Then return to the target sound. and again you correct the student's mistake by making the opposite mistake. This somewhat unconventional method is called "compensatory correction" or "servo-correction". It does not require the student to "understand" the teacher's world or even language. But it DOES require the student to accept that a mistake might be corrected by another "mistake". The inherent hierarchy and authority system of schools makes many students expect that the teacher is always correct, or should be at least, so that the student always has a perfect model to try to be like. PTLC2005 Ole Stig Andersen Correction Typology:4 A similar compensatory principle is used with difficult consonant clusters, which Danish has a fair amount of. Speakers of Turkish may make a typical difficult cluster like [sg-] in "skole" (“school") pronouncable by inserting a vowel after the [s]. Compensatory correction therefore corrects either by moving the vowel Teacher: sgool Student: segool Teacher: esgool or by establishing prosodic priority since a consequence of such supporting vowels is an extra syllable. Teacher: sgool Student: esgool Teacher: gool Another category of "manipulative methods" is simple elision of problem sounds or their replacement by more easily pronounced though less correct sounds that will be understood by Danes. 4. MACHINE FEEDBACK gives the student real-time feedback on a variety of acoustic parameters such as intensity and pitch contours, spectrograms etc. The student is only required to create the same picture on screen as on a given model. Neither student understanding nor teacher intervention is required, ideally. An intensity display will show a clear difference between "v" and "b", and you can leave the Spanish student with the computer and a wordlist and a picture of the target sound as it shows on the screen. Then she can train at her own pace and endurance and without feeling judged by the other students or even by the teacher. A student that learns the difference this way may not be able explain it. Complicated displays like sonograms could be used in combination with filters to show the student the absence or presence – whatever is wanted – of a vowel formant within a certain frequency range. Standardised pitch contours could be used to train rythm and intonation. In spite of the great proliferation of computers in Denmark, these possibilities have not been explored to any degree yet, as far as I know. But I think that there are great promises in such an approach. Utopically I envisage a program library you can refer any student to, to train any difficulty on their own using computer feedback. But unequivocal one-to-one correlations between machine analysis displays of the acoustic output and the descriptive categories of language are few, so there is some way to go.