the skeleton: its contents, its shape, its function

notes to session 3

an easy case of compensatory lengthening

OE [nixt] > ME [ni:t] (> ModE [najt] night); this is “easy” because the skeletal slot is vacated right next to the lengthened segment.

difficult cases of compensatory lengthening

H [hida] > [hiːd] ‘bridge’, Old Church Slavonic [bógŭ] > pre-Serbo-Croation [bôːg] (=[bóog]) ‘god’, OCS [bobŭ́] (you should see an acute on a breve on the u) > pSC [bóːb] (=[boób]) ‘bean’, proto-Greek [odwos] > Ionic Greek [oːdos]; these are difficult because we have to get the vacated skeletal slot across a consonant.

solution: the mora (μ)

Assign a mora (the unit of syllable weight) to all vowels and in some languages to some (notably preconsonantal) consonants. But just so, this solution turns out to be too loose: the mora can reassociate to practically any segment.

To control the possibilities Hayes proposes to abandon the skeleton and replace it with moras. A drawback: if moraless (typically onset) consonants are directly linked to the skeleton, their order will remain undecided in complex onsets.

SW Finnish

Samothrake Greek

null hypotheses of the representation of syllables

• the syllable type every human language exhibits is CV
• anything more complex must be represented as an extension of this type
• if a language has V syllables there are two options
  1. we posit a new type of syllable: V
  2. we assume that the C part of our CV syllable remains unpronounced
• if a language has CVC syllables there are two options
  1. we posit a new type of syllable: CVC
  2. we asume that the V part of our CV syllable remains unpronounced, hence CVC = CVC〈V〉
• we must always follow either the (A) options or the (B) options, any departure (like standard government phonology) must be argued for