12 TOPIC 2: PHONEMES AND ALLOPHONES Question: Why would most (all?) English speakers say that lips, slip, spill, Pils, and lisp comprise the “same sounds” in different orders? Answer: Although the physical SOUNDS differ from word to word, the words do comprise the same set of PHONEMES. phoneme: (first try at a definition) “a basic sound unit of a language” allophone(s): the phonetic variant(s) of a phoneme phonological rules: generalized statements defining the conditions for the appearance of the particular phonetic realizations (allophones) of an “underlying” phonemic form Some formal conventions / /: Slant brackets enclose phonemic (“underlying”) forms, e.g. /pɪls/ ‘Pils’ [ ]: Square brackets enclose phonetic forms, e.g. [pʰɪɫs] ‘Pils’ ([ɫ] =“dark” l) X  Y / __Z: “X is realized as Y in the environment before Z” X  Y / Z___: “X is realized as Y in the environment after Z” Recipe for phonemic analysis ASSEMBLE A FULLY REPRESENTATIVE DATA SET, ACCURATELY TRANSCRIBED. a. Do phonetic charts for consonants and vowels. b. Collect local environments for the sounds being investigated. c. Look for patterns in environments—do particular sounds appear exclusively in certain environments? d. If the answer to (c) is “yes”, state the environment in the most general way possible. Often one allophone will most easily be stated as “elsewhere”. e. Set up an underlying form for the phoneme (usually the “elsewhere” allophone), and write phonological rules (A  B / P ___ Q) deriving the other allophones. Write the rules in as general a form as possible, leaving out unnecessary material. f. See whether the pattern worked out in (e) can be generalized to other sets of sounds. For example, a rule that aspirates all the voiceless stops is more general than a rule that just aspirates /t/.Linguistics 120A 2. Phonemes and Allophones 13 Example 1: English velars The English phoneme /k/ can be described as having (at least) three allophones: a “neutral” velar allophone, which we can symbolize as [k], a backed allophone, which be can symbolize as [k!] and a raised or fronted allophone, which we can symbolize as [k!]. Submit the following words to the “recipe” above 1. keen [ ] 6. cool [ ] 11. creep [ ] 2. kin [ ] 7. could [ ] 12. clean [ ] 3. Kate [ ] 8. coal [ ] 13. crone [ ] 4. kettle [ ] 9. cut [ ] 14. clone [ ] 5. catch [ ] 10. cod [ ] 15. crop [ ] 16. clod [ ] Thinking like a phonologist: What additional questions does our analysis raise? - Can the analysis be extended to a larger class of segments than just /k/? - We have only looked at /k/ in / [word ___ (C) V. What happens in / C___V / V___ ]word / V___V (in particular, if the vowels “conflict”, which one wins?) Some other issues of phonology that this data set illustrates: - HOW DO WE KNOW THAT THERE IS ANYTHING TO LOOK FOR IN THE FIRST PLACE?! - The messiness of dealing with real life data. - The idealized nature of phonological rules vs. precise phonetic descriptions. - The usually “local” nature of phonological rules. - The phonetic naturalness of phonological processes. - A tendency for “anticipation” of following environments to be stronger than “perseverance” of preceding environments. Example 2: Papago (Focus: [t, t!ʃ, d, d"ʒ) 1. [ˈbid!ʒim] ‘turn around’ 12. [ˈhɨwgid] ‘smell’ 2. [ˈtaːpan] ‘split’ 13. [ˈt!ʃihaŋ] ‘hire’ 3. [ˈhidoɖ] ‘cook’ 14. [ˈtoɲi] ‘become hot’ 4. [ˈt!ʃɨkid] ‘vaccinate’ 15. [ˈwiɖut] ‘swing’ 5. [ˈgatwid] ‘shoot’ 16. [ˈtaːtaɖ] ‘feet’ 6. [ˈt!ʃuku] ‘become black’ 17. [ˈkiːt"ʃud] ‘build a house for’ 7. [ˈdagʂp] ‘press with hand’ 18. [ˈdoːdom] ‘copulate’ 8. [ˈtoha] ‘become white’ 19. [ˈtaːtam] ‘touch’ 9. [ˈd!ʒuːki] ‘rain (noun)’ 20. [ˈd!ʒɨwɨd] ‘soil, earth’ 10. [ˈwɨːmt] ‘help, marry’ 21. [ˈt!ʃɨgig] ‘name, reputation’ 11. [ˈd!ʒɨːk] ‘taste’ 22. [ˈt!ʃiːwia] ‘settle, establish residence’Linguistics 120A 2. Phonemes and Allophones 14 The Vowel System of Papago front central unrounded back rounded high i, iː ɨ, ɨː u, uː mid o, oː low a, aː The Consonants1 labial alveolar palato-alveolar retroflex palatal velar glottal voiceless stops p t k voiced stops b d ɖ g voiceless affricates t!ʃ voiced affricates d!ʒ voiceless fricatives s ʂ h nasals m n ɲ ŋ liquids ɾ glides w j The Data Sorted by Immediate Context (numbers index the word used, from data set above) t t!ʃ d d!ʒ 2,16,19 [word ___aː 13 [word ___i 3 i___o 1 i___i 8,14 [word ___o 4 [word ___ɨ 4,5,12 i___ ]word 9 [word ___uː 15 u___ ]word 6 [word ___u 17 u___ ]word 11 [word ___ɨː 5 a___w 17 iː___u 7 [word ___a 20 [word ___ɨ 16,19 aː___a 21 [word ___ɨː 18 [word ___oː 10 m ___ ]word 22 [word ___i 18 oː___o 16,19 [word ___aː 20 ɨ___ ]word 1 Data and analysis assembled by Bruce Hayes, based on Saxton, Dean, Lucille Saxton, and Susie Enos (1983) Dictionary: Papago/Pima-English, English-Papago/Pima, University of Arizona Press, Tucson.Linguistics 120A 2. Phonemes and Allophones 15 Left Contexts Only: No Pattern t t!ʃ d d!ʒ 2,8,14,16,20 [word ___ 4,6,13,21,22 [word ___ 3,4,5,12 i___ 1 i___ 15 u___ 17 iː___ 15 u___ 9 [word ___ 5 a___ 7,11,18 [word ___ 20 [word ___ 16,19 aː___ 18 oː___ 10 m___ 20 ɨ___ Right Contexts Only t t!ʃ d d!ʒ 2,16,19 ___aː 13 ___i 3,18 ___ o 1 ___i 8,14 ___o 4 ___ɨ 4,5,12,15, 20 ___ ]word 9 ___uː 10,15 ___ ]word 6,17 ___u 7 ___ a 20 ___ɨ 5 ___w u ___ɨː 18 ___ oː 11 ___ɨː 16,19 ___a v ___iː There is a simple and coherent generalization (the kind that phonologies tend to favor). “The palato-alveolar affricates occur before high vowels, and the alveolar stops occur elsewhere.” Few Data Especially for [d!ʒ]. But among the values of a precisely formulated phonological analysis are (1) that it suggests areas where you should look for further data and (2) that it makes predictions that can be tested by such data. Formalizing to Achieve Generality • Assume underlying /t,d/: these are what you get if no rule perturbs the basic pattern. • State rule as simply as