| |
Abstract:
Agreement errors in sentence production are asymmetrical:
local (nonhead) plural nouns cause far more errors than local
singular nouns (Bock & Miller, 1991):
1. The bridge over the rivers *are made of steel. (frequent
error)
2. The bridges over the river *is made of steel. (infrequent
error)
Bock & Miller attributed this to the fact that plurals are
overt and singulars are null, but Eberhard (1997) adverts to
morphosyntactic feature markedness: singular is unmarked (i.e. it
is the default number), while plural is marked, and only marked
features cause agreement errors.Since overtness and markedness are
confounded in English number, Eberhard examined the effects of
specifying number with quantifiers. Compared to (1) and (2), local
plural errors decreased when the head noun had a singular
quantifier, as in (3), and local singular errors increased when the
local noun had a singular quantifier, as in (4):
3. Every slogan on the posters...
4. The slogan on every poster...
In addition, a plural quantifier on a plural local noun did not
increase error rates compared with the non-quantifier
condition:
5. The slogan on several posters...
Singular is also unmarked in Romanian, but it can be either null
or overt. If these two types differ, then overtness does play a
role in agreement errors. Forty adult speakers of Romanian
participated in a production experiment conducted in Romania,
Canada, Germany, and the U.S. The pattern of results is as follows:
(1) a far smaller effect of local plurals than in Bock &
Miller's English study, perhaps because of the rich morphology of
Romanian; (2) far more errors on local singulars when they are
overt; (3) fewer local plural errors when the head has overt
singular inflection; (4) when both S and P are overt, far more
errors on PS than on SP, suggesting that overt singulars are, in
some sense, less marked than null singulars and more marked than
overt plurals; (5) for uninflected singulars, an additional
singular definiteness marker causes very few errors, but for
inflected singulars, the definiteness marker leads to very high
error rates.
We will also report the results of an English study, currently
underway, in which inflected local plurals are compared with
conjoined local plurals (the bridge over [the river and the
highway]...),which do not carry overt plural inflection. If only
markedness counts, there should be no difference in error rates
between the inflected and conjoined cases. We also vary the number
of the nouns within the conjunct and compare them to nouns within
prepositional phrases (the bridge over the river near the
highway...). Previous research has shown that it is the noun
adjacent to the head whose number affects error rates, rather than
the noun adjacent to the verb. If conjuncts have flat structure,
then the middle noun is syntactically farther from the head noun
than in PPs, and error rates should be lower. But if conjuncts are
asymmetrical, as currently assumed, then there should be no
difference between conjuncts and PPs.
We will discuss these findings with respect to the notion of
morphosyntactic markedness; in particular, we question whether
overtness and markedness can be separated, since overtness is one
of the criteria for markedness.
|
