Words and Syllables

[HallowXIII]

The SCA does not currently handle suprasegmentals internally. While it is possible to hack together rules approaching this, it would be better to represent words as sequences of syllables, each of which can be specified for suprasegmental features.

Syllables themselves should be sequences of segments, with the syllable being aware of the type and position of elements within it -- along the lines of public Segment getNucleus().

The initial iterations of this feature will require syllable breaks to be explicitly specified in the lexicon and should be made strictly optional. A future integration as the default system will require an intelligent segmenter.

[samanthamccabe]

Think this is quite feasible. All data are stored in Sequence objects, which could be enhanced to have a "syllable" tier representing which segments belong to which syllables.

Requiring the user's lexicon to reflect syllable boundaries is excellent at this point, since a lot of people may not need them, and it reduces the burden on us. We'd still need a way to indicate the nucleus.

When a user uses whatever command to enable syllables, they might also provide a series of syllable patterns (using regular expressions). The SCA could use these to determine what segment or segments represent the nucleus. If they were using features, this might be a little easier.

[HallowXIII]

PyLaut use(s?/d) features everywhere, so what we did there was just check for [+syllabic] and arrange the consonants around that -- would probably mesh nicely with syllable patterns.

[samanthamccabe]

Well, for better or worse, Haedus is feature-system-agnostic. I'm sure there's some reasonably simple way of doing it.