Introduction to the idea of Phonological features Keith Johnson 1 Let s start by considering how big the phonetic alphabet needs to become in order to transcribe to an arbitrary degree of phonetic detail a I need a volunteer to say food b all write a super narrow phonetic transcription of this word The definition of super narrow transcription is that with this transcription and a talent for reading phonetic transcription exactly it possible to repeat a word exactly as it was said before What is the problem with this exercise What happens when you ask the speaker to say it again 2 Our experience in attempting to provide super narrow transcription suggests that there is a limit to the amount of detail we can get in phonetic transcription There is a theoretical point here as well as a practical one It would be a pain to provide a super narrow transcription but also we suspect that in using speech as a communicative system people ignore super narrow differences and attend to some more gross level of phonetic description a broader transcription if you will 3 Cross linguistic phonetic variation We use the same IPA symbol but the sound is a little different from one language to the next i and in English are phonetically different from i and in Kimatuumbi Kimatuumbi i is higher than English i and Kimatuumbi is a bit lower than English i German is different from i in English not sure how to describe this difference exactly any German speakers in the class Hindi retroflex consonants have much milder retroflexion than does Telugu We will x rays of this in the section on place of articulation Disner s cross linguistic study of vowel formants found that the height of mid vowels in five 1 vowel systems in Italian versus Yoruba differed Nartey s cross linguistic study of fricative spectra found different spectral shapes in different languages perhaps correlating with degree of laminality Many other examples see Ladefoged Out of chaos comes order 4 Such differences are audible and we could expand the phonetic alphabet to cover them e g Kimatuumbi i versus English i But as we saw with our food experience there has to be a limit 5 The IPA answer only one of two possible answers The gross level of phonetic description that the IPA is designed to capture is phonetic properties that may be used in some language of the world to contrast meaning Contrastive phonetic property may distinguish words in some language Any phonetic property that can participate in a minimal pair in some language of the world gets a symbol in the IPA Interestingly people who study language acquisition starting perhaps with Werker and Tees 1984 feel that infants are sensitive to just these dimensions of phonetic variation and are not inclined to use other phonetic dimensions e g loudness 6 The IPA answer is unsatisfying because it forces us to one of two levels of phonetic transcription broad or narrow Using narrow transcription we are committed to transcribing any property that is contrastive in some language and have tools for transcribing many phonetic details Using broad transcription we focus on recording only those properties that are contrastive in this language or dialect 7 Consider for example how these two transcription systems work in the case of stop consonant voicing in English 2 As we have seen for many speakers of English a voiceless unaspirated t stop with the s cut off sounds just like an instance of the voiced stop d Both t and d become flap intervocalically before weak vowels e g later and ladder However in some speech communities I ve seen this in the speech of African Americans the voiced stops of English are really voiced IPA has symbols for all of the variants that we are talking about here t t and d but we might not use them all in narrow transcription of my variety of English because my initial stops are not usually voiced narrow broad American English t t t d African American English t t and d t d What the IPA solution is missing is a way of talking about how sounds pattern together within a language in any more sophisticated way than this We need a way to talk about the transcriptions at different levels of analysis Score keeping a phonetic transcription abstracts away from phonetic detail the same word can be spoken different ways b phonetic detail is part of what speaker hearers know about language sounding native is not easy to do c contrast universal or language specific is one guide to abstraction but narrow is too narrow and broad is too broad 8 Now here is an observation parallel to the food observation Just as food may be pronounced differently in different contexts stems can change their phonetic shape divine divinity diva n divin electric electricity l ktr k l ktr s In expressing morphemes people permit phonetic variability even of contrastive elements e g fine vs fin and kin vs sin 3 Paradoxically speakers know both how to ignore phonetic variation for purposes of identifying words or morphemes and how to produce language specific articulatory patterns This is true at two levels of analysis a at the narrow transcription level all instances of food are of the same word b at the broad transcription level all instances of divin are of the same stem Phonology is the study of sound patterning how word and stem forms are realized As we see with instances like divine divin this goes beyond the conventions of the IPA An aside on reality Is the stem of divine diva n really the same as the stem of divinity divin How strong is this relationship Relationships between stems like this seem to be stronger for some stems than others and may differ from one person to the next One semantically related form to another Despite their suppletive morphological relation people don t seem to think of is as a phonetic variant of are they are just two words But refinity if n ti is probably a better nonce word people know what it means refinedness and smile at the clever or silly person who says it than is wa plural of the verb to whizz Also if n ti is better than ifa n ti The gradient strength of these intuitions corresponds to gradient degrees of productivity in studies of the introduction of nonce forms e g Lieber 1983 9 An example of sound patterning We have said several times that phonetic transcription depends partly on an analysis linguist s and speakers of the language s sound pattern In Khmer the syllable rhyme that is pronounced like the vowel in English my is not a diphthong but a sequences of vowel