Introduction to Contemporary Linguistics November 25, 1998 Phonology 2: the sounds of morphology and syntax [WARNING! ALL OF THE PHONETIC FONTS HAVE BEEN RUINED! FOR CORRECT DETAILS, PLEASE SEE A REAL HANDOUT!!] OVERVIEW: 1. Phonology and morphology 2. The ordering of rules 3. Prosody: syllables and stress 4. Phonology and syntax 5. Summary ============================================================= 1. Phonology and morphology >Last time I emphasized how phonology is part of grammar and so in the mind, not in the mouth, air, or ear. More evidence for this claim comes from how phonology interacts with other parts of grammar, such as morphology. >Just as phonological phonemes are linked to phonetic allophones, so are morphological morphemes linked to phonological allomorphs: Morphemes: PLURAL Allomorphs: /z/ /s/ /@z/ /@n/ 0 ... Examples: dogs cats horses oxen deer_ ... >Notice that the allomorphs are phonemic representations, not phonetic. For example, /z/ and /s/ can form lexical contrasts: zoo sue buzz bus >Notice also that some of the allomorphs appear for no reason. They are lexically conditioned allomorphs. Since there is no pattern, there is no rule: the information is just in the lexicon. two oxen *two oxes *two ox two deer *two deers *two deeren >Often, however, allomorphs are phonologically conditioned. That is, they appear according to a phonological rule. Abstract morphemes: CAT CAT + PLURAL | | V V Phonemic representations of the morphemes: /kaet/ /kaet/+/z/ | | V V Phonemic representations of the allomorphs: /kaet/ /kaets/ | | V V Allophones: [khaet] [khaets] >What are the rules for the plural allomorphs [s], [z] and [@z]? Let's start by looking at the first two allomorphs: The allomorph /s/ The allomorph /z/ "tapes" [thejps] "jobs"[dZabz] "cats" [khaets] "beds"[bEdz] "clocks"[khlaks] "dogs"[dOgz] >The above pattern can be described with one phonological rule: >We can first describe the pattern using phonemes: [s] appears after [p], [t], [k], ... [z] appears after [b], [d], [g], ... >Then we can simplify this with phonological features: [s] appears after [-voiced] consonants. [z] appears after [+voiced] consonants. >We can put this information together into a rule: Devoicing: /z/ -> [s] after [-voiced] consonants >Notice that this rule can also explain other examples that I haven't listed yet. This shows that the rule is pretty good: "laughs"[laefs] *[laefz] "baths" [baeTs] *[baeTz] >But sometimes the plural suffix is pronounced [@z]: "horses" [hors@z] "bushes" [bUS@z] "branches" [braentS@z] "garages" [g@raZ@z] >This pattern can also be described with features: Schwa Insertion: 0 -> [@] between two [+sibilant] consonants Note: Sibilants are fricatives or affricates that are "noisy". This rule doesn't apply with other kinds of fricatives: "laughs" [laefs] *[laef@z] "gloves" [gl^vz] *[gl^v@z] "baths" [baeTs] *[baeT@z] >Are our rules correct? It's important to be sure, since we can only say we understand the grammar if the rules really do what we want them to do. So let's check. >Is the Devoicing rule too specific? Notice that we get the same kind of voicing pattern with the past tense suffix: "slipped" [slIpt] "robbed"[rabd] "laughed" [laeft] "loved" [l^vd] "missed" [mIst] "buzzed"[b^zd] "packed" [phaekt] "hugged"[h^gd] Devoicing (revised): [+alveolar] -> [-voiced] after [-voiced] >The past tense suffix also has a kind of schwa insertion, but it happens in a different context: Schwa Insertion II: 0 -> [@] between two [+alveolar, -continuant] C's >Another question: What is the true underlying representation (UR) of the plural morpheme and past tense morpheme? PLURAL: /z/ or /s/ ??? PAST: /d/ or /t/ ??? >Fromkin and Rodman (pp. 293-297) think that the UR's are [+voiced], and there is a Devoicing rule, but why can't the UR's be [-voiced], and then have a Voicing rule? >F&R's hypothesis (°²³] ): Plural UR is /z/ "rats" "dogs" UR /raet + z/ /dOg + z/ Devoicing raets --- Phonetics [raets] [dOgz] >Alternative hypothesis: Plural UR is /s/ (like spelling) "rats" "dogs" UR /raet + s/ /dOg + s/ Voicing --- dOgz Phonetics [raets] [dOgz] >Both hypotheses seem to work equally well so far, but what happens when you have a vowel? "boys" [bojz] *[bojs] "days" [dejz] *[dejs] >Fromkin and Rodman's hypothesis works: "days" UR /dej + z/ Devoicing --- Phonetics [dejz] >Alternative hypothesis doesn't work: "days" UR /dej + s/ Voicing --- Phonetics *[dejs] >These examples illustrate a very important part of doing linguistics: making hypotheses, and then testing them. Linguistics is a science because you don't just follow what experts say, but instead you test claims for yourself! 2. The ordering of rules >There's one more problem with our rules so far. What happens if they don't apply in the right order? WRONG ORDERING: "horses" "dishes" UR /hors + z/ /dIS + z/ Devoicing hors + s dIS + s @-Insertion hors + @z dIS + @z Phonetics *[hors@s] *[dIS@s] CORRECT ORDERING: "horses" "dishes" UR /hors + z/ /dIS + z/ @-Insertion hors + @z dIS + @z Devoicing --- --- Phonetics *[hors@z] *[dIS@z] >Therefore, in standard phonological theory, rules are ordered in a specific way, just like commands in a computer program. >When you "know" a language's phonology, you know at least three things: >A set of UR's for the morphemes and words. >A set of rules. >The ordering of the rules. >Then, to get the phonetic pronunciations of the words, you apply the rules, in the correct order, in a derivation: UR: /X/ /Y/ Rule 1: X1 Y1 Rule 2: X2 -- ... ... ... Rule 99: -- Y52 Rule 100: X37 -- Phonetics: [X37] [Y52] >Here's another example in my dialect of English: >First, remember the Flapping rule that we discussed: Flapping: /t/ and /d/ become the voiced flap [D] between a vowel and an unstressed syllable "atom" "quality" "latter" "ladder" [ae.D@m] [khwa.lI.Di] [lae.Dr] [lae.Dr] >Now here is another rule. When many speakers of North American English pronounce /aj/, we actually say [^j] in certain predictable contexts, thus raising the low, back /a/ up to a mid, back [^]: "tight" "tide" "tie" [th^jt] [thajd] [thaj] Raising: /aj/ becomes /^j/ before [-voiced] >But Flapping changes a [-voiced] consonant /t/ into a [+voiced] consonant [D], so what happens in words where both rules can apply? "ride" "write" "rider" "writer" [rajd] [r^jt] [rajDr] [r^jDr] >That last word [r^jDr] seems very strange, since we get a raised [^j] before a [+voiced] consonant! But it's not strange if you imagine that Flapping is ordered after Raising, so at the time that Raising works, the /t/ is still [-voiced]! "ride" "write" "rider" "writer" UR /rajd/ /rajt/ /rajdr/ /rajtr/ Raising --- r^jt --- r^jtr Flapping --- --- rajDr r^jDr Phonetics [rajd] [r^jt] [rajDr] [r^jDr] >If the rule ordering hypothesis is right, then this is very strong evidence that phonology is in the mind, not in physics. The Raising rule is not operating in the physical world, but in the mental world. 3. Prosody: syllables and stress >Another important piece of evidence that phonology is in the mind is prosody: the phonology of rhythm, timing, and duration. >Prosody groups features and segments into prosodic constituents, such as syllables. Prosodic constituents are very obvious to speakers of a language, but they are hard to define in physical terms: they are more mental than physical. >For example, English speakers may feel that "flower" has one syllable, while "flour" has two syllables, but is there any actual physical difference??? >Like syntactic constituents, prosodic constituents have structure. Chinese linguists have studied syllable structure for centuries, and their theories have influenced ª`­µ²Å¸¹ : Syllable ­µ¸` Initial Medial Final Án¥À ¤¶­µ Ãý¥À £t £¹ £¯ £§ £º £­ ... ... ... >The theory of syllable structure used by Fromkin and Rodman is quite similar: $ = syllable $ O = onset R = rhyme O R N = nucleus C = coda N C s t r E N T s >Basically, the Chinese "initial" is the onset and the Chinese "final" is the rhyme. But what is the "medial"...? >Here are some Mandarin syllables in the modern theory: $ O R N C a ªü th a ¥¦ a n ¦w th a n ½Í >What about medials? Many modern Chinese phonologists think that they are really part of the onset, perhaps as a feature: $ O R N C th i E n ¤Ñ thj E n ¤Ñ (other analysis) >Syllables play an important role in phonology and morphology. >Last week we saw that some phonological rules pay attention to syllables, such as English Nasalization: Any vowel becomes nasalized before a nasal consonant that is in the same syllable "fat" "fan" "fantastic" "fanatic" [faet] [fae~n] [fae~n.taes.tIk] [f@.nae.DIk] >Morphology can also pay attention to syllables. In Ilokano (spoken in the Philippines), syllables can be reduplicated: dait 'sew' ag-da-dait 'sewing' takder 'stand' ag-tak-takder 'standing' basa 'read' ag-bas-basa 'reading' >Differences in syllable structure across languages can help explain how foreign words are pronounced. >Unlike English, Mandarin does not allow high vowels or glides after a vowel when there is a coda: ®ü [xai] º~ [xan] *[xain] [xei] «Ü [x@n] *[xein] >This explains why Chinese speakers sometimes mispronounce my name: "James" [dZejmz] *[dZemz] >Mandarin syllables cannot contain more than one consonant in the onset. Thus some foreign names will have inserted vowels to break them up into more than one syllable: "Clinton" [khlIn.th@n] -> ¬ìªL¹y [kh@.lIn.twUn] >Another prosodic constituent is the stress foot: a constituent of one or two syllables, where one syllable is stressed. Stress feet create a rhythm of stress across a word. >In English, stress feet are stressed on the first syllable: (sAd) (hAppy) (Ele)phant A(mEri)ca (Ori)gin o(rIgi)nal o(rIgi)(nAli)ty (hEma)(mAli)(dAnthe)mum [a kind of flower] (sUper)(cAli)(frAgi)(lIstic)(Expi)(Ali)(dOcious) [a fake word from the film Mary Poppins (Åw¼Öº¡¤H¶¡)] >Phonological rules pay attention to feet: a(tOmic) [@.tha.mIk] (Atom) [ae.D@m] a(ttAck) [@.thaek] (Attic) [ae.DIk] English Aspiration (revised for the last time): All voiceless stops become aspirated at the beginning of a foot >Feet also play a role in morphology. For example, in English slang, "fucking" can infix between two feet: Root Emphatic form (fAn)(tAstic) (fAn)-fucking-(tAstic) (PhIla)(dElphi)a (PhIla)-fucking-(dElphi)a >Does Mandarin have feet? In Mainland China, the answer is YES! Unstressed syllables are pronounced with Tone 0 (»´Án). ¾Ç¥Í (X x) (stress on first syllable) ª¾¹D (X x) (one foot) ¤£ª¾¹D (X x)(X) (two feet) ªF¦è (X x) "thing" (one foot) (X)(X) "east & west" (two feet) >In Taiwan Mandarin, stress is not as clear, but you can still sometimes notice that stress goes on the first syllable: ÁÁ (X x) *(X)(X) (one foot, not two) >I think the reason why stress in Taiwan Mandarin is unclear is that Taiwanese does not use stress feet at all, but I'm not sure. (Anyway, that Japanese professor's "humidity" theory is wrong!) 4. Phonology and syntax >One final indication that phonology is part of grammar is that it can interact with syntax. >For example, English questions often have the intonation [L....H]: [L.............H] (low pitch to high pitch) Do you want tea? >But the intonation phrase depends on syntactic phrase structure: [L............................H] [S' Do you want tea or coffee]? (yes/no) [L.................H]..................L] [S' Do you want tea] or [S' ... coffee]? (choice) >Similar phonological effects happen in Mandarin. Recall the Tone 3 rule, often called tone sandhi (tone change due to another tone): Tone sandhi rule: /11/ -> [35] before /11/ ¶R °¨ /mai11/ /ma11/ -> [mai35] [ma214] >Tone sandhi happens in different ways depending on the syntactic structure of the sentence. [OVERHEADS] ¦Ñ§õ¶R¦n°s¡C [35 11 11 35 214] "Lao Li bought good wine" [35 11 35 35 214] "Lao Li finished buying wine" >Taiwanese, Cantonese and Hakka also have tone sandhi rules that pay attention to syntax, but they are much more complicated! (That would be an interesting paper topic....) 5. Summary [this table is messed up as usual...] English Mandarin Allomorphs and rule ordering A lot Very little (because Mandarin morphology is very simple) Syllable structure complex: CCCVXCC (C=consonant, V=vowel, X=either) simple: CVX (C=consonant including e.g. /túÆ/, V=vowel, X=either) Stress feet Yes Mainland: Yes Taiwan: No? Syntactic phrases Marked by intonation Marked by tone sandhi