LINGUIST List 4.1000

Sun 28 Nov 1993

Qs: Optimality Theory

Editor for this issue: <>

Directory

1. Takaaki Hashimoto, Q: Lexical representations in Optimality Theory

A question on lexical representations in Optimality Theory

 I am posting a query on behalf of my colleagues.
We would very much appreciate it if anyone could provide us
with a suggestion for the solution of the problem presented
below. Please respond directly to me, and I will summarize
and post the responses.

1. Problem

In Spanish, a word stress falls onto one of the last three
syllables in a word, and never onto any pre-antepenultimate
syllable. Onto which of the last three syllables the stress
falls is not predictable, while the penultimate stress is the
most common option. For the purpose of discussion, we will
focus on the difference between words with the penultimate
stress and those with the antepenultimate stress in the
following discussion.

2. Question

Under Optimality Theory, how can we formally represent the
lexically conditioned stress position, while deriving the
prohibition against the pre-antepenultimate stress in some
way?

3. Proposals and problems

There are several ways to represent the lexically conditioned
stress, while none of them appears to be problem-free. Some
of them are discussed below.

Account 1:

Solution ...
The rank ordering of the relevant constraints in Spanish is:
 RHT/T >> Align (Ft, R, PrWd, R) >> NonFinality

For words with an antepenultimate stress, the head foot of
the prosodic word and the syllables dominated by the head
foot are given in the lexical entry, with the head syllable
of the head foot in the antepenultimate position.
Containment and Consistency of Exponence ensure that the
lexical head foot and its syllables will be represented in
all the candidates generated from the input.

Or:

For words with an antepenultimate stress, the last syllable
is directly attached to the prosodic word in the lexical
entry. Containment and Consistency of Exponence ensure that
the final syllable will not be a part of the head foot in all
the candidates generated from the input. Align-R ensures
that the optimal candidate has the antepenultimate stress.

Problem ...
There is no explanation for the absence of the pre-
antepenultimate head syllable of the head foot nor for the
absence of the penultimate syllable directly associated with
the prosodic word in the lexical representation.

If the head foot of the prosodic word and its syllables can
be given in the lexical entry, nothing bars the head syllable
of the head foot from being assigned to a pre-antepenultimate
position in the lexical representation.

If a given syllable can be directly attached to a prosodic
word in the lexical entry, nothing bars the penultimate
syllable from being attached to the prosodic word. If the
penultimate syllable is attached to the prosodic word in
the lexical entry, then the optimal candidate will have a pre-
antepenultimate stress, due to RHT/T and Align-R.

Account 2:

Solution ...
The rank ordering of constraints can be different for
different classes of words.

The rank ordering of the relevant constraints for the
unmarked penultimate words in Spanish is:
 RHT/T >> Align (Ft, R, PrWd, R) >> NonFinality

The rank ordering of the relevant constraints for the marked
antepenultimate words in Spanish is:
 RHT/T >> NonFinality >> Align (Ft, R, PrWd, R)

The words with the antepenultimate stress have a lexical
marking on the rank ordering of relevant constraints in the
lexical entry.

Problem ...
If the rank ordering of constraints can be different for
different classes of words within a language, there can be
indefinitely many classes of words, each of which differs
from another in the rank ordering of constraints. The order
can possibly be different for each word.

If it is necessary to evoke lexical marking of some kind to
explain the distribution of word stress, then how can we
constrain the arbitrariness of lexical marking and derive the
prohibition against the pre-antepenultimate stress?

Should we propose constraints on the lexical entries, i.e.
constraints by which possible input forms (rather than the
candidates generated from the input) are evaluated?