Download Arthur Holmer

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Arabic grammar wikipedia , lookup

Integrational theory of language wikipedia , lookup

Esperanto grammar wikipedia , lookup

Scottish Gaelic grammar wikipedia , lookup

Inflection wikipedia , lookup

Lojban grammar wikipedia , lookup

Serbo-Croatian grammar wikipedia , lookup

Old English grammar wikipedia , lookup

Germanic strong verb wikipedia , lookup

Musical syntax wikipedia , lookup

Pipil grammar wikipedia , lookup

Modern Hebrew grammar wikipedia , lookup

Grammatical aspect wikipedia , lookup

Russian grammar wikipedia , lookup

Antisymmetry wikipedia , lookup

Spanish verbs wikipedia , lookup

Swedish grammar wikipedia , lookup

Future tense wikipedia , lookup

Yiddish grammar wikipedia , lookup

Transformational grammar wikipedia , lookup

Lexical semantics wikipedia , lookup

Junction Grammar wikipedia , lookup

Macedonian grammar wikipedia , lookup

Chichewa tenses wikipedia , lookup

Tense–aspect–mood wikipedia , lookup

Polish grammar wikipedia , lookup

Distributed morphology wikipedia , lookup

Grammatical tense wikipedia , lookup

Transcript
Agrammatism dissociations and generative theory:
The evidence reviewed
Arthur Holmer
Agrammatism (Broca's aphasia) selectively affects the patient's production of grammatical and fluent language, while not
affecting lexical competence. This is (not very controversially) taken as evidence that the syntactic component of language is
mediated by other mechanisms than the lexical component. In short, language consists of its grammar and its words, and each can
be affected separately (in Broca’s and Wernicke’s aphasia respectively).
But grammar is not always totally affected: since Grodzinsky (1990) it has been known that agrammatism affects different
subsystems of the grammar to varying extents. Particularly striking is the contrast between Agreement and Tense, cf. Friedmann
2001:74.
Berega ze ha-yeled
now
DET-boy
TENSE:
holex.
walk.PRES.MASC.SG
SUCCESS RATE
Gam
etmol
ha-yeled _____.
(-> halax)
too
yesterday
DET-boy
walk.PST.MASC.SG
AGREEMENT: Berega ze gam
now
too
31%
91%
ha-yelad-im ______.
(-> holxim)
DET-boy-PL
walk.PRES.MASC.PL
Similarly, Hebrew-speaking agrammatics perform poorly in wh-question production and subordination, whereas they perform almost normally in
yes/no questions and non-finite embedding (such as adjectives or participles), cf. Friedmann 2002: 79–84.
Repetition test:
Whquestions
success rate
57%
Relative Complement
clauses
clauses
33%
33%
Yes/no*
questions
Non-finite
embedding
87%
92%
* tested for elicitation, not repetition
First solution: Tree Pruning Hypothesis (TPH)
- agrammatism restricts access to functional nodes (AgrP, TP, CP)
- varying degrees of agrammatism = varying span of structure affected
- agrammatism affects all structure above cut-off point
Spec
CP
3
C°
C'
3< 1
Spec
Cut-off at: 1 => CP impaired (no embedding)
2
=> TP + CP impaired (no embedding + no tense)
3 => Agrp + TP + CP impaired (no embedding + no tense + no agreement)
TP
3
T'
3< 2
T°
AgrP
3
Spec
Agr'
3< 3
Agr°
VP
!
structure based on Pollock (1989)
Baker (1985): The Mirror Principle
One important consequence of the TPH as it stands is that it crucially assumes
that Tense is structurally above Agreement. Since Baker (1985) it has been a
common assumption among syntactic typologists that structural ordering can
be determined independently. Baker’s Mirror principle states that:
Morpheme order reflects movement of V through syntactic structure,
which implies that a morpheme closer to the verbal root corresponds to
a category lower in the structure). This can be demonstrated clearly for Finnish.
AgrP
minä e-n
puhu-isi
2
1s
NEG-1s speak-COND
Spec
Agr'
‘I would not speak’
minä
2
Agr° NegP
-n
2
Neg°
ModP
e2
Mod°
VP
-isi
2
V°
OBJ
puhu-
AgrP
minä puhu-isi-n
2
1s
speak-COND-1s
Spec
Agr'
‘I would speak’
minä
2
Agr°
ModP
-n
2
Mod°
VP
-isi
2
V°
OBJ
puhu-
German morphology
According to the Mirror Principle, facts from
German suggest a structure where AGR is
above Tense. Taken together, the Mirror Principle
and the Tree Pruning Hypothesis predict that
the German agrammatism pattern should be the
exact reverse of Hebrew. German agrammatics
should have more serious impairments in
Agreement than in Tense.
TENSE
arbeite- -tearbeite- -tearbeite- -tearbeite- -tearbeite- -tearbeite- -te-
AGR
-Ø
-st
-Ø
-n
-t
-n
3
‘I worked’
C°
AgrP
‘thou workedest’
3
‘s/he worked’ =>
Spec
Agr'
‘we worked’
3
‘you worked’
Agr°
TP
‘they worked’
3
T°
VP
!
The German data (adapted from Wenzlaff & Clahsen 2004):
AGREEMENT
Sentence completion
Grammaticality judgment
92%
81%
TENSE
68%
57%
This shows that German and Hebrew agrammatics pattern similarly,
despite the assumed structural differences between the languages.
Either the assumed structural differences are a misinterpretation,
or agrammatism dissociations do not depend on syntactic structure.
Wenzlaff & Clahsen (2004) argue that Tense and Agreement (following Chomsky 2000) are located in one syntactic node (T), and that the
dissociations depend on the fact that Tense values are inherently underspecified (Tense Underspecification Hypothesis: TUH). A reason for this
underspecification would be that Tense establishes a relation between event time and the speech act itself, while agreement only establishes a
relation within the syntax (cf. Wenzlaff & Clahsen 2005).
Morphological reasons?
Hebrew tense is derived by vowel alterations, along a number of different patterns.
Hebrew agreement, on the other hand, involves affixation. The same contrast holds
for the most frequent verbs in German (strong verbs with Ablaut inflection). Thus,
Agreement is generally more salient and more linear than Tense. If non-linear
grammatical processes require greater effort, they might be selectively impaired.
Hebrew:
roked-et
rakad-eti
rakad-nu
‘dance.PRES-FEM.SG’
‘dance.PST-1SG’
‘dance.PST-1PL’
German:
geh-t
ging-st
‘go.PRES-3SG’
‘go.PST-2SG’
However, this hypothesis would predict a greater tense impairment for strong verbs than for weak verbs in German.
Wenzleff & Clahsen (2004) note that there are no significant performance differences in this respect.
Syntactic bonsai?
An analysis compatible with German data is presented by Lee (2003), who claims that impairment sets in below, rather than above, cut-off point. A
patient with impaired Tense but functioning Agreement has has an impairment cut-off below AgrP (see below). In analogy with Tree Pruning, this
could be referred to as “Root Pruning” or, indeed, bonsai.
CP
3
This proposal would tally well with the assumed structure for German, but would instead face
problems with Hebrew structure (T above Agr). It would also have problems accounting for
Spec
C'
impairment of embedding.
3
Ouhalla (1993) does, indeed, argue for a T-above-Agr analysis for both Arabic and Berber
C°
AgrP
(which are relatives of Hebrew). His arguments involve, in Arabic, the realization of Tense
3
and Agreement on different elements in the clause (i.e. tensed negations preceding
Spec
Agr'
agreement-inflected verbs), and, in Berber, morpheme ordering within the verb (TNS-AGR-√).
3< HERE Following Baker (1985), both facts point in the same direction (T-above-AGR). Data below
Agr°
TP.....
from Ouhalla (1993).
t-tullabu
lam
yadhab-uu
DET-student-NOM NEG.PST go-3PL
‘The students did not go.’
ad-y-segh
Mohand ijn
FUT-3M.S-buy Mohand 1
‘Mohand will buy a house’
teddart
house
t-tullab-u
lan
yadhab-uu
DET-student-NOM NEG.FUT go-3PL
‘The students will not go.’
ARABIC
BERBER
However, it is not clear that this carries over to Hebrew, however, which lacks tensed negations, and cf. the parallelism between Hebrew and
German morphology. By the Mirror Principle, Hebrew should presumably have Agr-above-T structure like German, and the pattern for both would
fit with Lee (2003) - “root pruning”. However, this would not affect CP, so it still does not permit a structural account for the lack of embedding.
How robust is the selective impairment of Tense?
success rates
Speaker A:
Speaker B:
Speaker C:
...
Agreement
75%
75%
50%
Tense
92%
50%
67%
Burchert et al (2005) present data which show that the impaired category in
dissociation can vary according to individuals. For the selection of speakers shown
here, two have a more severe impairment with Agreement, while one has a more
severe impairment with Tense. This cannot be reconciled with a single cut-off
point. Burchert et al. (2005) suggest that Tense and Agreement can be
separately underspecified (individual varition). This is the Tense or Agreement
Underspecification Hypothesis (TAUH).
However, it is not clear that their data is representative (only 2 speakers in the whole study show Agr>Tns impairment), and the data is unique.
separate underspecification
TUH:
TAUH:
TNS
+
+
AGR
+
Both models are incompatible with Baker’s (1985) Mirror Principle. If either
of these analyses is right, then the quite robust connection between morpheme
order and independently verifiable syntactic structure can not be accounted
for structurally.
Further, these analyses face severe empirical problems in both Arabic and Finnish, where Tense and AGR are demonstrably realized in different
positions in the clause.
Is TAUH / TUH only valid for a subset of languages?
- if TAUH / TUH holds in languages where there is a single node (T) for both Tense and Agreement (e.g. German), and if Hebrew has the same
structure, how does agrammatism dissociation work in languages where we have evidence for two separate nodes (e.g. Arabic, Berber or
Finnish)? It would be interesting to see whether a closer investigation of agrammatism among speakers of these languages can replicate
Burchert et al.’s results.
Does TAUH / TUH require a single head T?
- if the features affected by underspecification need not necessarily share a single node, there is no correspondence between syntactic structure and
agrammatism, and aphasia data has no import for syntactic theory.
References
Baker, Mark. 1985. The mirror principle and morphosyntactic explanation. Linguistic Inquiry 16: 373–416
Burchert, Frank, Maria Swoboda-Moll & Ria De Bleser. 2005. Tense and agreement dissociations in German agrammatic speakers: Underspecification vs. hierarchy.
Brain and Language 94: 188–199
Chomsky, Noam. 2000. Minimalist inquiries. the framework. R. Martin, D. Michaels & J. Uriagereka (eds). Step by step. Cambridge, MA: MIT Press. 89–155.
Friedmann, Na'ama. 2001. Agrammatism and the psychological reality of the syntactic tree. Journal of Psycholinguistic Research 30: 71–90
Friedmann, Na'ama. 2002. Question production in agrammatism: The Tree Pruning Hypothesis. Brain and Language 80: 160–187.
Gleason, J. B., H. Goodglass, E. Green, N. Ackerman & M. R. Hyde. 1975. The retrieval of syntax in Broca’s aphasia. Brain and lanaguge 2: 451–471.
Goodglass, H., J. B. Gleason, N. A. Bernholz & M.R. Hyde. 1972. Some linguistic structures in the speech of Broca’s aphasics. Cortex 8: 191–212
Grodzinsky, Y. 1990. Theoretical perspectives on language deficits. Cambridge, MA: MIT Press.
Lee, M. 2003. Dissociations among functional categories in Korean agrammatism. Brain and Language 84: 170–188
Ouhalla, Jamal. 1993. Functional categories and parametric variation. London: Routledge
Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar and the structure of IP. Linguistic Inquiry 20: 365–424
Wenzlaff, Michaela & Harald Clahsen. 2004. Tense and agreement in German agrammatism. Brain and Language 89: 57–68
Wenzlaff, Michaela & Harald Clahsen. 2005. Finiteness and verb-second in German agrammatism. Brain and Language 92: 33–44.