THEORY: final features, moras, and roots [was: it's what I do]

On Tue, 3 Oct 2000, H. S. Teoh wrote:

> On Mon, Oct 02, 2000 at 05:17:19PM -0700, Marcus Smith wrote:
> > Wow.  Now I understand final features. :-)
> [snip]
>
> Hmm. I tried reading the chapter on final features, but can only get a
> vague impression. Anyone care to explain what it really means, in layman
> terms? :-P
In layman's terms? Okay, I'll try. (If I don't succeed, ask again.)
The term "final feature" was coined in 1974 by a linguist working on
Northern Paiute, and it's stuck ever since. (A more appropriate name
IMO would be "final segment", but now we're already in the realm of
analysis.) The basic idea is this: part of the lexical information of
a noun is the effect it has on a following stop or nasal. Some nouns
are always followed by a geminate stop or nasal (about 1% of the total
lexicon), others are always followed by a nasal-stop cluster (about
28%), others are followed by a voiceless fricative or nasal (about
22%), and the rest (about 49%) are followed by a voiced fricative or
nasalized glide. the trick is that when these nouns occur finally in a
phrase, the final feature is silent. Here are some examples:

        [tToo] 'bead'
        [tToombai] 'have beads'

        [tToo] 'great grandparent'
        [tTooBai] 'have a great grandparent'

In the first pair of words, a nasal appears between the stem and the
suffix [-bai]; this nasal wasn't pronounced on the stem in isolation.
The second pair of words shows a neat contrast; the stems in isolation
are pronounced identically, but when the suffix is attached, it is
apparent that the first ('bead') also contains a nasalizing final
feature, while the second ('great-grandparent') does not. Final
features only show up when a consonant follows. And then only if the
consonant is a stop or nasal; /s, h, w, y/ are unaffected by final
features.

(Chapter 1 has a useful summary of the final features of Gosiute
without the theoretical baggage.)

 I'm especially confused about the part involving counting
> morias... what *are* root nodes and morias??
A mora is a unit of syllable weight. A syllable with a long vowel has
two moras, while a syllable with a short vowel has one. In many
languages, syllables which are closed or checked by a consonant also
have two moras. Linguists often refer to syllables with two moras as
"heavy" and syllables with a single mora as "light."

Mora count is relevant for reckoning the proper placement of stress.
In Latin, for example, main stress was placed on the next to last
syllable (the penultima) if that was heavy, or on the syllable before
that if the penultimate syllable was light. In Gosiute, syllables
containing an odd mora counting from the left are stressed:

        .PI.ca.HI.pi.TE. 'drinking milk'

Sometimes when the second syllable contains a long vowel, the stress
passes over the first syllable and rests on the second syllable
instead:

        nuTAAnuh 'ran'

Syllables closed by a consonant are not counted as heavy for the
reckoning of stress:

        .TO.im.PI.te. 'finally came out'

The syllable [im] is closed by a consonant, but it doesn't receive
stress, showing that closed syllables are not heavy, and therefore
that consonants which close syllables are not moraic.

A root node is the organizational element of a speech sound; its the
point in the representation of a speech sound at which all of the
features of that speech sound are gathered together; it lends the
speech sound its character as a unit separate from its neighbors. The
consonant /p/ can be represented as follows:

               o           root
              /|\
        -voice | \         laryngeal feature
               |  \
               |  stop     manner feature
               |
            labial         place feature

Another way to think of root nodes is as "place holders" in the stream
of speech sounds; it anchors a segment in the speech stream. The root
node is a hard sell for some linguists because it is rather an
abstract notion.

> Also, what's a geminate consonant?
A geminate consonant is long; compare the pronunciation of English /n/
in 'penny' with the /n/ in 'pen knife'; you should notice that the /n/
in 'penny' is short, while that of 'pen knife' is long. The lengthened
/n/ in 'pen knife' can be considered a geminate.

Dirk

--
Dirk Elzinga
dirk.elzinga@m.cc.utah.edu