How to syllabify 'behave' and 'behaviour'?

Question

I'm well aware of the Maximal Onset Principle which says that 'intervocalic consonants should be syllabified as the onset of the following syllable as long as the Phonotactic constraints allow it'. I was explaining it to someone and got confused by some words such as behave /bɪˈheɪv/ and behaviour /bɪˈheɪvjə/. They're problematic:

• if I syllabify 'behave' as */bɪ.heɪv/; English syllables don't end in lax vowels such as /æ ʌ ʊ ɪ/ etc
• If I syllabify 'behave' as */bɪh.eɪv/, it violates the phonotactics of English in that English cannot have a syllable ending in /h/

Now I have absolutely no idea how to syllabify 'behave' and 'behaviour'. Is there anything I'm missing here? How to get around this problem?

Answer 1

The usual syllable division given for sequences like /ɪh/ in English is /ɪ.h/.

Lax vowels can end syllables in some contexts

"English syllables don't end in lax vowels such as /æ ʌ ʊ ɪ/ etc" is not actually easy to support as an exceptionless rule, so most theories of syllabification recognize exceptions.

The most obvious example of a word with a syllable ending in a lax vowel is tattoo, usually pronounced something like [tʰætʰú] (marking stress with the acute rather than the IPA stress mark to avoid begging the question of the syllabification), where the pronunciation of the middle consonant as an aspirated voiceless plosive [tʰ] is not consistent with most theories about how syllable-final or ambisyllabic /t/ is supposed to be pronounced. So the syllable division is usually hypothesized to be /tæˈtu/, which requires abandoning the principle that English syllables cannot end in a lax vowel. The modified rule replacing it is usually formulated in terms of stress: "English stressed syllables don't end in lax vowels".

This stress-conditioned rule is not restrictive enough to produce the distribution of word-final vowels that we see in practice (at minimum, /æ/ /ɛ/ /ɒ/ are not found word-finally in regular English words); so we either need to say that there are different phonotactic rules for word-final and syllable-final position in this case, or that the word-final distribution of lax vowels in English is not entirely based on phonotactic rules (for example, some vowels might be "accidentally" absent in word-final position).

The question of syllables ending in /ɪ/ is complicated by further variation

In the case of /ɪ/ specifically, there are further complications because in certain positions, the "lax" or "short" vowel /ɪ/ is not well distinguished from the "tense" or "long" vowel /i/. The applicable environments vary by accent/dialect; one of the most well known contexts where /ɪ/ or /i/ are not contrastive and either may occur is in unstressed syllables in word-final position or in hiatus with a following vowel. In John Wells' system of lexical sets, an /ɪ/ or /i/ in this position is assigned to a special diaphonemic set, labeled "happY", representing that in some accents (such as the old-fashioned British accent referred to as "RP"), the vowel in words like *happy or recreation is identified with the phoneme /ɪ/, while in other accents, the vowel is identified with the phoneme /i/. In an accent where happy is pronounced /hæpɪ/, it's clear that /ɪ/ can be both syllable- and word-final, so a syllabification like /bɪ.heɪv/ poses no special problems in an accent of this type.

John Lawler noted in the comments that behave shows another phenomenon: the vowel in the first syllable is sometimes pronounced as tense /i/, and sometimes as lax /ɪ/ (or a further reduced "ɨ" or /ə/). Wells also discusses this class of words in the following blog post: believing descriptions

Answer 2

There's a dialect question as to what the pronunciation of "behave" is – I say [bəˈhɛiv], Jlawler says (I guess) [biˈhɛiv]. I say [biˈhɛd] but I've heard [bəˈhɛd]. We grew up in different towns, I think. You can't have the lax vowels in unstressed open syllables, which is why I reduce it to schwa, and it's just random chance as far as selecting a tense vs. lax vowel, as to by there is a [bi], [bə] distinction. Such doublets exist in economic, otolaryngologist.

[bə.ˈhɛiv] is not a problem, because (unstressed) schwa is allowed word-finally (vanilla, Pamela) which is the basis for interpolating claims about syllable position. In case the unstressed vowel is actually phonetic [ɪ], you also have to determine that the vowel is not "basically" a tense vowel. Final [i] is surface lax in some dialects (southern in my experience, but I might not notice it in e.g. California English), so the generalization about no final ɪ is not an output condition, it's slightly deeper into the grammar.