the skeleton: its contents, its shape, its function

notes to session 1

Properties of sounds are extractable, “autosegmentalizable”, if they are often found to spread, i.e., assimilate or harmonize. Besides tones, other vowel properties (height, frontness-backness, rounding, tenseness) are like this. Vowel length is not like this. Of consonantal properties place of articulation and voicing/aspiration looks like autosegments, manner properties (vocalicness, stopness, glideness) do not look like autosegments.

Compensatory lengthening provides a strong argument for separating sound properties from timing (i.e., the skeleton).

Grassmann’s law

• there can be only one instance of aspiration within a stem (in Ancient Greek and Sanskrit): see Wikipedia entry
• examples: tʰrik-s ‘hair-nom’ vs trikʰ-a ‘hair-acc’, trepʰ-oo ‘I feed’ vs tʰrep-s-oo ‘I will feed’
• e-tʰrepʰ-tʰɛɛn ‘feed-1sg-aor-med/pass’ shows that the aspiration of the [pʰ] in this stem does not “count”, it is simply the interpretation of the aspiration of [tʰɛɛn]

yes-no question intonation in Hungarian

• LHL, interpreted as
  • low-high-low if three syllables are available
  • low-falling if two syllables are available
  • rising-falling (or only rising) if only one syllable is available

compensatory lengthening

Reconstructed Proto-Greek *esmi ‘I am’ is eemi in Attic Greek, emmi in Aeolian Greek, illustrating that either of the neighbouring segments may lengthen to compensate for the loss of the coda [s].