1. Ambisyllabicity2. Rules for ambisyllabicity3. More blurred syllables in English 4. Ambisyllabification and Tapping 5. Some additional consequences of ambisyllabification5.1 /l/ allophones5.2 Stop Aspiration5.3 Affrication of /s/ after /n/5.4 Vowel nasalization5.5 Distributional patterns6. Secondary stress and syllabification7. Returning to the /ol/ Data, with syllables 8. Summary 9. Ambisyllabicity in child speechChapter 13A: Syllabification in English Bruce Hayes Department of Linguistics UCLA This is a chapter that I decided not to include in my textbook Introductory Phonology (2009, Malden, MA: Blackwell). I’m fond of the data, but ambisyllabicity seems to be a sufficiently controversial hypothesis in phonology that the chapter may not be suitable for use in a broadly-distributed text. Comments and corrections welcome: [email protected]. Copyright 2009 by Bruce Hayes. This material may be freely used for teaching, study, or any other nonprofit purpose. 1. Ambisyllabicity English has some puzzling cases with regard to syllabification. Consider words like butter, camel, upper, Lenny, etc. Native intuition seems to waver on whether these should divide as /b.t‘/ or as /bt.‘/. This is different from what we see in many other languages, where intuitions on how to divide syllables are clearer. The phonological treatment of the English “blurred syllabification” is a long-standing issue in phonology, and various analyses have been proposed. The one I will describe here seems to work fairly well. The basic idea is that some consonants can belong to more than one syllable at a time; in standard terminology they are called ambisyllabic. Ambisyllabic consonants can most easily be depicted using the tree notation for syllable structure: they are consonants that are dominated by more than one σ: butter camel upper Lenny σ σ σ σ σ σ σ σ b  t ‘ k æ m l  p ‘ l  n iIntroductory Phonology Chapter 13A: Syllabification in English p. 2 Such a representation would account for the ambiguous intuitions speakers have concerning the syllabification of such words. English syllable division is not blurry in every instance. When a stressed vowel follows the consonant sequence at issue, syllabification seems pretty clear: balloon /b.lun/, proclaim /proÉ.feÉn/, approach /.poÉt/, insane /n.seÉn/, attract /.trækt/, etc. It is only where the second vowel is stressless, as in the cases shown above, that we get ambisyllabicity. Our analysis of syllabification should take account of this difference. 2. Rules for ambisyllabicity We can derive English ambisyllabification if we augment the rules for syllabification seen in the chapter 13. The first three are repeated from before. σ Assignment Assign syllable nodes (σ) to be in one-to-one correspondence with [+syllabic] sounds. Onset Formation Join consonants to the following syllable, provided the resulting cluster can occur at the beginning of a word (Maximal Onset Principle).1 Coda Formation Join any consonants not yet syllabified to the preceding syllable. Ambisyllabification I σ σ → σ σ where the second syllable is stressless C C The idea is that the first three rules assign the sort of syllabification that most languages have, then the last rule, Ambisyllabification I, blurs the syllabification of the initial consonant of a stressless syllable. For example: the word approach would get two syllables, one each for its two syllabic sounds: σ σ  p oʊ t 1 As we saw in the last chapter, English needs limitations: for instance, /tw/ and /dw/ are treated as onsets only when they are word-initial.Introductory Phonology Chapter 13A: Syllabification in English p. 3 The consonants work like this: first we take as many consonants as can to become onsets: σ σ  p oʊ t And then we make the remaining consonants into codas: σ σ  p oʊ t The rule of Ambisyllabification I can’t apply here, since the vowel of the second syllable isn’t stressed. A word like Lenny would first be syllabified like this: σ σ l ɛ n i Here, since the second syllable is stressless, Ambisyllabification I can apply, and we get: σ σ l ɛ n i Why does ambisyllabification exist? One conjecture is that it represents a compromise between two contradictory “goals,” as follows: • Stressed syllables want to have more segments, while stressless syllables want fewer. • All syllables want to have onsets. The ambisyllabification seen in Lenny lets the first syllables be a relatively beefy /ln/, while still letting the second syllable have an onset. 3. More blurred syllables in English We’ve seen earlier that languages differ in how they treat /...C]word [wordV... / (word ending in consonant, followed by word beginning in vowel). German syllabification respects word boundaries, so the first syllable of the second word cannot acquire an onset by “stealing” a consonant from the preceding word. Spanish works the opposite way, with word-final consonants transferring fairly freely to the following syllable.Introductory Phonology Chapter 13A: Syllabification in English p. 4 English is a peculiar intermediate case: the word-final consonants seem to be blurred in their syllabification, for example the /t/ in bite a (melon), or the /p/ in up a tree. Plausibly, these represent another case of ambisyllabification, and we can posit the following rule: Ambisyllabification II σ σ σ σ C V → C V (phrase bounded) This rule assumes that words are first syllabified by themselves, then together. I haven’t put any ]word expressions into the rule, because they are not needed—in the usual case, it is only when words have first been syllabified separately that we ever get a consonant that doesn’t belong to the same syllable as the following vowel. Let us look at an example. A brief preliminary: since the expression ]word[word, which designates the break between words, is a little bit verbose, I will use a conventional (fairly commonly used) symbol “#” instead. The basic syllabification rules for up a tree would give the following: σ σ σ  p #  # t  i