Sounds & Words. Week 3 Michaelmas Mark J. Jones

Transcription

1 Sounds & Words Week 3 Michaelmas 2010 Mark J. Jones markjjones@cantab.net

2 Last week Airstreams pulmonic, glottalic, velaric Vowel quadrilateral height + fronting + rounding Phonemes & allophones physics the actual sounds heard function lexical contrasts & sound patterns

3 This week Talk in more detail about phonemes & allophones in English some English allophones are phonemes in other languages some English phonemes are allophones in other languages Different languages divide up phonetic space in different ways

4 Phonemes & allophones Phonemes are sounds which are lexically contrastive pin vs. tin vs. kin vs. sin vs. fin vs. thin these words are minimal pairs they differ by one sound only Minimal pairs main way of identifying contrasts commutation test Need not be real words: tap vs. dap dap not a real word, but it could be

5 Phonemes & allophones The sounds we hear are called phones aspirated alveolar plosive [tʰ] in tin aspirated velar plosive [kʰ] in kin Allophones are contextual variants of the same phoneme in different contexts tell with dark final [lˠ] let with light initial [l] complementary distribution within the same lexical context, e.g. cat with final [t] vs. [ʔ]

6 Allophones REMEMBER: phonemes are not real, they are abstract what we hear are (allo)phones allophones are grouped into phonemic categories We ignore phonetic differences and focus on contrast stan vs. tan red vs. thread pill vs. lip cull vs. cud

7 Allophones - Issues Do listeners really link phones in different contexts together? some lexical patterns: tell (dark [lˠ]) linked to telling (light [l]) fourteen (aspirated [tʰ]) and sixteen (unaspirated [t]) set off ( [tʰ] ) and set down ( [ʔ] ) ticker ( [kʰ] ) and ticking ( [k ʰ] ) Why wouldn t they? Humans are good at pattern recognition group stars into constellations, see images in clouds, flames, shingle on a beach

8 Allophones issues Experimental evidence listeners compensate for context [s] with lip-rounding sounds more like [ʃ] create a synthetic fricative half-way between [s] and [ʃ] in a rounded vowel context, listeners claim to hear more [s] it isn t really [s]; it s neither [s] nor [ʃ] they compensate for contextual rounding hear the hybrid fricative as [s] That said, directly proving the existence of the phoneme is tricky (cf. the atom) it s an abstract entity

9 Phoneme what symbol? As the phoneme is abstract the choice of symbol is technically arbitrary English /r/ is not a trill in most accents (i.e. [r]) square brackets = physical reality slash brackets = abstraction in recent phonological theory this is not the case blurring of boundary between phonetics & phonology division of labour content of abstract symbolisation

10 Some RP allophones Two major causes for allophones: segmental context syllabic context Segmental context cannot move seamlessly from one segment to next coarticulation articulations interact effects predictable to some extent Syllabic context more mysterious effects not predictable

11 Syllable structure Nucleus is obligatory usually a vowel can have syllabic consonants too in some languages Onsets are very common onset-nucleus most common syllable shape (transition?) onsets obligatory in some languages Codas much rarer many languages do not have codas Clusters are also rarer

12 Syllables structure

13 Some RP allophones Allophones of /l/ coda /l/ is dark [lˠ] (older [ɫ]) velarised: back of tongue raised onset /l/ is light non-velarised pal vs. lap pill vs. lip tell vs. let

14 Some RP allophones

15 Some RP allophones Allophones of voiceless plosives initial plosives are aspirated produced with abducted vocal folds final plosives are pre-glottalised produced with adducted vocal folds pre-pause and pre-consonantal pill [pʰɪlˠ] lip [lɪʔp ] Use of tie-bar to show that symbols are not a cluster we can use the same thing for affricates, i.e. [tʃ ]

16 Some RP allophones Vowel duration + final obstruent obstruent = plosives, fricatives, affricate Vowel is longer before voiced obstruent not the vowel length contrast, just phonetic peace /piːs/ vs. peas /piːz/ hiss /hɪs/ vs. his /hɪz/ phonetic transcription uses half-length diacritic [ˑ] use is logical (once you think about it!)

17 Some RP allophones Short vowel + voiced > half-long hiss /hɪs/ vs. his /hɪz/ hiss [hɪs] vs. his [hɪˑz] Long vowel already long becomes half-long + voiceless peace /piːs/ vs. peas /piːz/ peace [pʰiˑs] vs. peas [pʰiːz] This works even if the final sound is devoiced, i.e. no vocal fold vibration, so [hɪˑz ] (cf. whispered speech)

18 Some RP allophones Nasalisation Nasal consonants produced with lowered velum allows air (and sound) into nasal cavity Lowering the velum (velic lowering) takes time vowel + nasal consonant > nasalised vowel sad [sæˑd ] sang [sæ ŋ]

19 Some RP allophones Fronting of velars velar sounds produced with tongue back (dorsum) place of articulation border of hard & soft palate adjacent vowels modify tongue shape front vowel velar becomes fronted (pre-palatal) back vowel velar becomes backed (post-velar) keep [kʰ iˑʔp ] curb [kʰɜːb ] carp [k ʰɑˑʔp ]

20 Some RP allophones Vowel retraction before dark /l/ bid [bɪd] bill [bɪ lˠ] Produces a range of allophones tell vs. ted [tʰɛ lˠ] pal vs. pad [pʰæ lˠ] The retracted vowels shift in the vowel space

21 Retracted vowels

22 Crosslinguistic differences Phonemes in English /r/ vs. /l/ rap vs. lap rack vs. lack not phonemes in Korean or Japanese park /ɑ/ vs. pack /æ/ quality difference not phonemic in Arabic duration is!

23 Crosslinguistic differences Allophones in English [ɕ] in sheep and [ʃ] in shark both contextual variants of /ʃ/ phonemes in Mandarin & Polish, /ɕ/ vs. /ʃ/ [ t ] in eighth and [t] in eight phonemes in Malayalam, Gunywingu

24 Reading Ashby, Michael, & John Maidment (2005) Introducing Phonetic Science, CUP (chapter 9) Giegerich, Heinz J. (2003). English phonology. CUP, Cambridge (chapter 2) Roach, Peter (2000). English phonetics & phonology. CUP, Cambridge (chapters 5 & 8) Cruttenden, Alan. (2001). Gimson s Pronunciation of English. Arnold (chapter 9) Questions? Mark: markjjones@cantab.net