Introduction to Contemporary Linguistics November 18, 1998 Phonology 1: sounds in the mind [WARNING! ALL OF THE PHONETIC FONTS HAVE BEEN RUINED! FOR CORRECT DETAILS, PLEASE SEE A REAL HANDOUT!!] OVERVIEW: 1. The existence of phonology 2. Phonemes and phonemic contrasts 3. Phonological rules 4. Distinctive features 5. Summary so far ============================================================= 1. The existence of phonology >If phonetics is the physical part of language, connecting two brains so that mental stuff like morphology, syntax and semantics can be communicated, then what is phonology??? >Phonetics (»y­µ¾Ç) studies the physical sounds of language. >Phonology (­µÃý¾Ç) studies the sound systems of language. >That is, phonology is the mental part of the physical part of language. [OVERHEAD] >The difference between phonology and phonetics is mysterious but important: it is the difference between mind and body. >A Japanese professor theorized that the reason why Tone 0 (»´Án) in Taiwan is different from that in Mainland China is because the air in Taiwan is more humid!!! The real explanation is mental, not physical (see next week). >Communicating animals like birds, whales, and bees have their own "semantics" (information) and "phonetics" (physical signals), but no morphology and syntax (systems of meaningful units like words) and no phonology (system of meaningless units like sounds). >Thus phonology is a special property of the human brain. Whatever it is, it is a part of what makes us human, just like syntax, morphology and other parts of grammar. >Since phonology is mental, phonological "sounds" really live in your brain, not in your mouth or in the air. >These sounds are phonetically different, but your brain thinks they are the same: Real phonetics What your brain thinks ½Í [tha~n35] /than15/ ¨¥ [iE~n35] /ian15/ >These sounds are phonetically the same, but your brain thinks they are different: Real phonetics What your brain thinks ¶R°¨ [mai35ma214] /mai11ma11/ ®I°¨ [mai35ma214] /mai15ma11/ >The consonants below sound the same to English speakers (and so they are written the same way in English spelling and in KK), but English speakers do not pronounce them the same way: top [th] stop [t] trip [tS] butter [D] mountain [?] winter 0 I met your sister. [tS] I met another one. [D] I saw them last week. 0 2. Phonemes and phonemic contrasts >The strange concept of a "sound" that is mental, not physical, gets its own name: phoneme. >A phoneme is a sound segment that serves the mental function of distinguishing words (e.g. /t/ in English) >For example, /t/ is a phoneme in English because it can distinguish words (minimal pairs that differ only in one sound): tape vs. cape still vs. skill tape cape still skill Phonemic representation: /tep/ /kep/ /stIl/ /skIl/ Phonetic represenation: [thejp] [khejp] [stIl] [skIl] >The different physical pronunciations of a single mental phoneme in a language are called allophones (allo = "other"). >In English, /t/ has the allophones [th], [t], [tS], [D], [?], 0 >A sound difference is phonemic if it distinguishes words, even if the difference is not a whole segment (e.g. tones, or phonetic properties). >In English, the phonetic property of voicing is phonemic: bet bed sue zoo lap lab fine vine pick pig chunk junk >However, in English, aspiration is not phonemic: [th]op vs. s[t]op are separate words, showing that /s/ is a phoneme in English. [th]op vs. *[t]op are not separate words, showing that aspiration is not *s[th]op vs. s[t]op phonemic in English. >Remember we are talking about MENTAL sounds that are part of a specific GRAMMAR. That means that the same physical sounds can play different roles in different languages. >In Mandarin (unlike English), aspiration is phonemic: [tha51] vs. [ta51] are separate words in Mandarin. >In Mandarin (unlike English), lexical tone is phonemic: [kou51] vs. [kou55] are separate words in Mandarin ...but in English, these same pitches do not create separate words: Go! [falling intonation] Must you go? [can be high level intonation] 3. Phonological rules >Even though allophones are physical things, the pattern of their appearance is not. In fact, the mental grammar contains rules that tell how the allophones should be used. >Phonemes live in the lexicon. They represent the arbitrary, unpredictable sounds of different words. >Why does "top" start with /t/? No reason! You just have to memorize that information! >Allophones are created by rules. They represent the systematic, predictable sound patterns. >Why does "top" have [th] but "stop" has [t]? This seems to be predictable: you only get aspiration when you have no /s/. >Allophonic rules are important because: >They are the major reason for "accents": non-native speakers have a "foreign accent" is because they don't use allophonic rules correctly. >Like syntactic rules, they are often quite complex, and yet children seem to learn them easily: a deep mystery! >The correct analysis of allophonic rules takes careful work. For example, what is the rule for aspiration in English? >Here's our first try: /t/ becomes [th] when it is not after /s/ >That's too negative, so let's put it in a positive way: /t/ becomes [th] at the beginning of a word >That's good, but it's too specific. Let's make it more general: All voiceless stops become aspirated at the beginning of a word. (So /p, t, k/ -> [ph, th, kh].) >But wait! You can get aspirated consonants even when they are not at the beginning of a word! "Tom" "atomic" [tham] [@thamIk] >So, how about this rule: All voiceless stops become aspirated at the beginning of a syllable. >No, no, that doesn't work: "atom" "atomic" [aeD@m] [@thamIk] >So here is the correct rule for Aspiration: All voiceless stops become aspirated at the beginning of a stressed syllable. >Using the same method, many other allophonic rules in English can be described, such as: >Flapping: /t/ and /d/ become the voiced flap [ä] between a vowel and an unstressed syllable "atom" "quality" "latter" "ladder" [aeD@m] [kwalIDi] [laeD@r] [laeD@r] >Nasalization: any vowel becomes nasalized before a nasal consonant that is in the same syllable "fat" "fan" "fantastic" "fanatic" [faet] [fae~n] [fae~ntaestIk] [f@naeDIk] >What are some allophonic rules in Mandarin? >At least one of them is the same as in English: Nasalization. ¤j ·í ³o ¥¿ [ta51] [ta~N1] [tS@51] [tS@~N1] >Another Mandarin allophonic rule: >Low vowel fronting: The low back vowel /a/ becomes the low front vowel /A/ before /n/ and /i/: ¤j ·í ³J ±a [ta] [ta~N] [tA~n] [tAi] >Other Mandarin allophonic rules are harder to figure out, since it isn't clear which sounds are the phonemes, and which are the allophones. This can be seen by comparing ª`­µ²Å¸¹ and Pinyin, both of which are phonological, not phonetic systems. >ª`­µ²Å¸¹ treats [u] and [O] as allophones but Pinyin doesn't: word IPA ª`­µ²Å¸¹ Pinyin ... [tu] £x£¹ du ªF [tON] £x£¹£¶ dong >Pinyin treats [E] and [@] as allophones but ª`­µ²Å¸¹ doesn't: word IPA ª`­µ²Å¸¹ Pinyin ½º [tiE] £x£¸£® die ±o [t@] £x£­ de >Both systems treat [kh] and [tSh] as phonemes, but are they? word IPA ª`­µ²Å¸¹ Pinyin ©@ [kha] £}£« ka Õt [tShia £¢£¸£« qia >Is there a rule of Palatalization? Velars become palatals before front vowels. >Or are we just wrong about the phonemic representations of these words, so there really is no rule at all? ©@ Õt /kha/ /tSha/ ??? (no /i/) (also no /i/) >Here's another Mandarin example showing the same problem. >Why do Mandarin speakers think that words like these contain the same vowel, when they really don't? ª`­µ²Å¸¹ Pinyin IPA ½Í £y£³ tan [tha~n] ¨¥ £¸£³ yan [iE~n] >Notice that /a/ and /E] are distinct phonemes, since you can find minimal pairs: words ª`­µ²Å¸¹ Pinyin IPA ®a £¡£¸£« jia [tSia] ±µ £¡£¸£® jie [tSiE] µM £§£³ ran [Za~n] ¤H £§£´ ren [ZE~n] >So do we have phonological rule like this? a-Fronting/Raising: /a/ becomes /E/ between a front vowel and /n/ >Or are speakers just being fooled by ª`­µ²Å¸¹ and Pinyin, and actually the phonemic representations are like this? ½Í ¨¥ /tan/ /iEn/ >One last point about allophonic rules: they can change over time. >ª`­µ²Å¸¹ and Pinyin represent ½Í and ¨¥ with the same ending because that was how these words used to be pronounced: a-Fronting/Raising is a "new" rule. >Palatalization is also "new", so when the French first went to Beijing they spelled the name as "Peking", because at that time the /k/ had not yet become [tS]. >In Modern English, /f/ and /v/ are phonemes: there are minimal pairs (e.g. "face" vs. "vase"). But in Old English (around 1000 AD), [f] and [v] were predictable allophones. >What allophonic rule describes the distribution of [f] and [v] in Old English? [SORRY -- THE EXAMPLES ARE UNREADABLE!] >First notice that [f] and [v] are in complementary distribution: [f] never appears where [v] does, and vice versa: >[f] at beginning or end of word >[v] only between vowels >Old English rule: /f/ becomes [v] between vowels. 4. Distinctive features >We've already seen a lot of examples showing that phonology seems to use phonetic properties. How is this possible if phonology is mental and phonetics is physical? >Phonology actually uses mental units called distinctive features: Distinctive features Phonetic properties [+voiced] ==> short VOT (voice onset time) [-voiced] ==> longer VOT [+aspirated] ==> even longer VOT >Distinctive features distinguish phonemes, e.g. in English: /t/ /d/ /s/ /z/ [voiced] - + - + [continuant] - - + + >Phonemes always fall into nice neat patterns like this. For example, English once had a gap in the feature pattern shown below. Now this gap has been filled by /½/ (e.g. in vision). Older form of English: [+labial] [+alveolar] [+palatal] [+velar] [+cont, -voiced] /f/ /s/ /S/ [+cont, +voiced] /v/ /z/ ____ >Such patterns can be explained if we recognize that segmental phonemes are really just combinations of distinctive features. >Most combinations of these features are allowed in English: Place cont voiced Labial + + Þ /v/ Labial + - Þ /f/ Labial - + Þ /b/ Labial - - Þ /p/ Alveolar + + Þ /z/ Alveolar + - Þ /s/ Alveolar - + Þ /d/ Alveolar - - Þ /t/ Palatal + + Þ /½/ Palatal + - Þ /§/ Palatal - + Þ *[+palatal, -continuant] Palatal - - Þ *[+palatal, -continuant] Velar + + Þ *[+velar, +continuant] Velar + - Þ *[+velar, +continuant] Velar - + Þ /ý/ Velar - - Þ /k/ >Thinking about features this way helps solve a mystery: Why do many people pronounce ±V as [ai35], instead of the standard [iai35]? >Maybe it's because the syllable [iai] goes against Mandarin grammar in some way. Think about all logically possible three-vowel syllables in Mandarin. Only four are common: Last vowel i u y i *iei iou *ioy *iai (±V) iau *iay First vowel u uei *uou *uey uai *uau *uay y *yei *you *yoy *yai *yau *yay >So Mandarin speakers don't like to have syllables with two [+high] vowels that are both [-back] or both [+round]! >Because segments act in natural classes in phonological rules, distinctive features allow us to simplify their description: >Mandarin Vowel Nasalization: Using segments instead of features (12 symbols!) Œ, i, u, y, { ® Œ~, i~, u~, y~, {~ before n, ÷ Using features instead of segments (only 3 symbols!) [-consonantal] ® [+nasal] / _ [+nasal] >What about features for tones? >The phonemic tones in Mandarin may be simpler than the phonetic forms that are actually pronounced: Phonetic form Possible phonemic form 1: [55] /55/ (high) 2: [35] /15/ (rising) 3: [214] /11/ (low) 4: [51] /51/ (falling) >Tone 3 has different pronunciations in different contexts: >Final low rule: /11/ -> [214] at end of a phrase °¨ /ma11/ -> [ma214] >Tone sandhi rule: /11/ -> [35] before /11/ ¶R °¨ /ma111/ /ma11/ -> [mai35] [ma214] >Elsewhere: nothing changes. ¶R ¨® /mai11/ /tS@55/ -> [mai11][tS@55] >Thus the tones in Mandarin are just all four combinations of high and low tone features, plus some allophonic rules. 5. Summary so far [SORRY -- FONTS ARE ALL MESSED UP AS USUAL.] >Here is a suggested feature chart for a suggested set of Mandarin phonemes. Remember that these are mental, not phonetic. >Note: [cons]=[consonantal], [son]=[sonorant], [hi]=[high], [lo]=[low], [rnd]=[round], [nas]=[nasal], [vcd]=[voiced], [asp]=[aspirated], [cont]=[continuant], [lab]=[labial], [dent]=[dental], [vel]=[velar], [pal]=[palatal], [ret]=[retroflex]. cons son hi lo back rnd i - + + - - - u - + + - + + y - + + - - + a - + - + + - @ - + - - - - cons son nas vcd asp cont lab dent vel pal ret p + - - - - - + - - - - ph + - - - + - + - - - - m + + + + - - + - - - - f + - - - - + + - - - - t + - - - - - - + - - - th + - - - + - - + - - - n + + + + - - - + - - - l + + - + - +? - + - - - k + - - - - - - - + - - kh + - - - + - - - + - - x + - - - - + - - + - - t‚ + - - - - -+ - - - + - t‚ú + - - - + -+ - - - + - ‚ + - - - - + - - - + - tê + - - - - -+ - - - - + têú + - - - + -+ - - - - + ê + - - - - + - - - - + ü/• + + - + - + - - - - + ts + - - - - -+ - + - - - tsú + - - - + -+ - + - - - s + - - - - + - + - - - ÷ + + + + - - - - + - -