| |
Abstract:
It is still a matter of debate to what extent revision
difficulty is affected by the length of the ambiguous region.
According to Fodor & Inoue (1998), distance does not affect
reanalysis difficulty as long as the error cue and the ambiguous
element are in a grammatical dependency relation. According to
Frazier & Clifton (1998) on the other hand, length of the
ambiguous region has an effect if it involves thematic role
assignment: reanalysis becomes hard when the erroneous analysis has
been semantically confirmed by a thematic role (cf. also Ferreira
& Henderson, 1991). Data from Sakha (a Turkic language) are
problematic for both accounts. In Sakha possessive constructions,
the possessee is marked for number of the possessor. However, the
possessee itself is ambiguous in number if the possessor is third
person plural:
(1)
a. Kiniler yt-tar-a ürer. t
hey dog-pl-3pers barks 'Their dog barks.'
b. Kiniler yt-tar-a ürel-ler.
they dog-pl-3pers bark-pl 'Their dogs bark.'
c. Kiniler yt-tar-a kiehee xaranaqa ürel-ler.
they dog-pl-3pers evening in.the.dark bark-pl
'Their dogs bark in the dark in the evening.'
The singular interpretation (1a) is preferred. We assume this is
because the structural representation of the singular is simpler as
no NUM node has to be postulated. Intuitively, revision is easy
when the disambiguating information is close to the noun, as the
plural verb in (1b), but can lead to a severe garden path effect
when the ambiguous region is lengthened, e.g. by inserting
adverbial material (1c). In both long and short cases, however, the
verb and the noun form a syntactic dependency and no thematic role
is assigned before the verb.
We therefore propose that not (only) thematic assignment, but
the completion of the DP affects the resolution of DP-internal
ambiguities. Evidence comes from the fact that it is easy to obtain
the plural reading when the disambiguating information is inside
DP, even if the ambiguous region is longer than in (1c), and even
if the disambiguating cue is pragmatic rather than syntactic (cf.
Fodor & Inoue, 1994), as in (2) where plural 'noses'
pragmatically enforces a plural reading of 'dog':
(2) kiniler yt-tar-y-n nahaa ulaxan xara munnu-lar-a
they dog-pl-3pers-gen very big black nose-pl-3pers
'their dogs' very big black noses'
At the position DP+1, the interpretation of the DP can be
completed. After this point, the parser shifts its attention to
following information, and the DP-internal subject becomes
inaccessible. Hence, revision is easy in (1b) and (2), but hard in
(1c). A study to experimentally confirm the above intuitions is in
progress.
Ferreira, F. & Henderson, J. (1991). Recovery from
misanalysis of garden-path sentences. Journal of Memory and
Language, 30, 725-745.
Fodor, J.D. & Inoue, A. (1994). The diagnosis and cure of
garden path sentences. Journal of Psycholinguistic Research, 23,
407-434.
Fodor, J.D. & Inoue, A. (1998). Attach anyway. In: Fodor, J.D.
& Ferreira, F. (1998). Reanalysis in sentence processing.
Kluwer, Dordrecht.
Frazier, L. & Clifton, C. (1998). Sentence reanalysis, and
visibility. In: Fodor, J.D. & Ferreira, F. (1998). Reanalysis
in sentence processing. Kluwer, Dordrecht.
|
