Click on talk title for handout/slides
8:30-9:30am: Breakfast
9:30-9:45am: Bill Badecker: Welcome to the Workshop and Introduction to Today's Symposium
9:45-10:35am: Stephen R. Anderson: Some Points of Agreement
10:35-11:00: Coffee Break
11:00-11:50am: Kay Bock: Reaching Agreement
11:50am-12:15pm: Coffee Break
12:15-1:05pm: Stephen Wechsler: Agreement and Grammatical Theory
1:05-2:30pm: Lunch!
2:30-3:20pm: Lee Osterhout: How Brains Respond to a Failure to Agree
3:20-3:45pm: Coffee Break
3:45-4:35pm: Naama Friedmann: Agreement in Agrammatism
4:35-5:00pm: Coffee Break
5:00-6:00pm: Discussion
6:00pm: Reception
7:30pm: Speaker Dinner
In normal language production, systematic variations in number agreement point to several basic mechanisms of agreement implementation. Two of these mechanisms involve (a) notional valuation, the categorization of referents with respect to numerosity, and (b) lexical specification, the grammatical subcategorization of morphemes with respect to number. The roles played by notional and lexical factors have been explored in studies that manipulate the properties of agreement controllers for verb and pronoun targets. The results of an array of experiments imply that verb and pronoun number are differently sensitive to the notional number variations underlying canonical agreement controllers (subject noun phrases) but are similarly insensitive to the notional number variations of spurious controllers. Notably, verbs and pronouns appear to be equally attracted to the grammatical number of spurious controllers. Experiments on collective agreement in American and British English and on distributive agreement in English and Spanish converge on the same conclusions. To account for these patterns, a psycholinguistic theory of marking and morphing (M&M) proposes separate mechanisms for control and concord that differ in how agreement features are represented and how they are transmitted during sentence formulation. The goal is to explain how number agreement works during language production to create a bridge from number meaning to number morphology.
Agreement, the systematic covariation of linguistic form, poses a
number of problems for grammatical theory. This talk briefly reviews
the following topics, then focuses in more detail on items 2 through
4:
1. Formal model. What is the appropriate formal model for agreement?
On unification-based frameworks, for example, agreement is a side-
effect of the fact that distinct elements of a sentence (e.g. the
subject NP and the subject agreement inflection on the verb) can
encode information about a single formal entity, hence that information
must be consistent.
2. Grammatical vs. semantic agreement. Some agreement is grammatical
(the target registers formal features of the trigger), while some
is semantic (the target registers semantic features of the trigger).
How does the competence grammar negotiate between these two types?
What is the role of defaults?
3. Other subtypes. Grammatical agreement is further divided into
more local and more distant subtypes. This division corresponds to
the varying origins of lexical agreement features of the trigger:
the features involved in more local agreement relations are related
to a noun's form, while those involved in the more distant relations
are related to its meaning.
4. Paradigms. Some apparent problems for agreement result from misanalysis
of person/number paradigms. Specifically, it will be argued that
the category of number should be expelled from the (1st and 2nd)
person paradigm, notwithstanding the traditional cross-classification.
This has significant implications for agreement.
5. Locality. What are the locality conditions on agreement and how
are they to be explained? What about apparent exceptions, so-called
"long distance agreement?"
6. Incorporated pronouns. How can we distinguish agreement affixes
from incorporated pronouns and why does it matter?
7. Discourse. How does pronoun agreement in discourse relate to the
interpretation of pronouns?
8. Variation. How should speaker variation in agreement be understood?
What are the boundaries of the competence grammar and how does grammar
interact with other factors?
Is agreement between sentence elements determined by application of a set
of formal syntactic rules, or by the semantic properties and discourse
functions of the agreeing entities? Conventional linguistic treatments
place agreement within the formal syntax. However, compelling arguments
exist that some or all agreement phenomena are better construed as being a
function of semantic or discourse factors. I will describe a series of
experiments in which we recorded event-related potentials (ERPs) from the
scalp while participants read sentences, some of which contained an
agreement violation. This strategy is motivated by prior work showing
that syntactic and semantic/pragmatic anomalies elicit qualitatively
distinct responses (the P600 and N400 effects, respectively). We have
found that a wide range of agreement-violating phenomena (including
subject-verb number, article-noun number, reflexive-antecedent number and
gender, and pronoun-antecedent number) in two languages (English and
French) elicit robust P600 effects highly similar to those elicited by
syntactic anomalies. Apparently, our participants perceived these
agreement anomalies to be syntactic, rather than semantic, in nature.
One interesting question is how and when this formal system of agreement
becomes "grammaticalized" during language acquisition. We examined this
question by longitudinally studying English-speaking adults as they
progressed through their first year of university French instruction. We
recorded ERPs in three separate sessions (after approximately 1, 4, and 7
months of instruction, respectively) while learners read sentences, some
of which contained a semantic or one of two types of agreement anomalies.
The agreement conditions included one that is highly similar in French and
English (subject-verb number), and one that is not (article-noun number).
Semantically anomalous words elicited an N400 effect at all three ERP
sessions. Subject-verb agreement anomalies elicited an N400-like effect
at the first ERP session and increasingly large P600 effects at the second
and third test sessions. ERPs to well-formed and ill-formed article-noun
agreement did not differ reliably at any of the testing sessions. We
interpret these data as reflecting a diachronic process of
grammaticalization,
but one that is sensitive to L1-L2 similarity.
Agrammatic aphasia, a deficit in syntax following brain damage to the left frontal hemisphere
presents an intriguing pattern of dissociations: People lose their ability to correctly
inflect verbs for tense, use subject pronouns, form relative sentences, produce
complementizers, or construct a well-formed Wh question. Even more striking is the fact that
at the same time they retain their ability to inflect verbs for subject agreement, use object
pronouns, form reduced relatives, produce coordination conjunctions, and form yes/no
questions in some languages. This selective nature of deficit provides a window through which
syntactic and morphological abilities in the brain can be viewed.
A study of syntactic and morpho-syntactic abilities in agrammatism in Hebrew and Palestinian
Arabic will be presented, showing a marked dissociation between verbal agreement and tense
inflection: agrammatic speakers of Hebrew and Palestinian Arabic fail to inflect their verbs
for tense but can still inflect verbs without errors for person, gender and number. This
pattern, together with additional body of evidence regarding the inaccessibility of other
functions of TP and of higher nodes of the syntactic tree can be explained by syntactic tree
pruning, namely, by the assumption that agrammatic aphasics fail to project their tree up to
its highest node. This causes agreement to be intact, whereas tense, which is in the pruned
area of the tree, is impaired. This suggestion thus supports the basic idea of Split
Inflection and entails that subject agreement is checked (or affixed) below the node in which
tense is checked.
Another study of spontaneous recovery of inflection in agrammatism will show that agreement
recovers before tense does, and only at a later stage do subordination and Wh questions
appear.
Finally, a short note will be presented with respect to the inability of individuals with
agrammatism to use their intact knowledge of verbal agreement to interpret sentences that
involve Wh movement.