Agreement in cognition, discourse, and syntax

Strangeness of agreement
Does this resemble the common linguistic understanding of the term “agreement”?

Agreement as formal control
“There is <…> a strong intuition, captured in the controller-target terminology, that agreement is asymmetric” (Corbett 2006: 115)
Psycholinguistics: inflectional or control theory of agreement

Origin of the dominant linguistic usage
Hermann Paul, 1880
Prinzipien der Sprachgeschichte,
chapter “On concord”
“die Tendenz Wörter, die in einer Beziehung zueinander stehen <…> in formelle Übereinstimmung miteinander zu setzen. Hierher gehört die Kongruenz in Genus, Numerus, Kasus, Person, wie sie zwischen einem Subst. und einem dazu gehörigen Präd. oder Attribut oder einem dasselbe vertretenden Pron. oder Adj. besteht <…> ”
Principles of the
history of language,
edition 1891
“There exists a tendency to place words related in a way <…> in formal correspondence with each other. Thus is explained the concord in gender, number, case, and person, which subsists between a substantive and its predicate or attribute, or a pronoun or adjective representing the latter <…>”

Formal control agreement is derivative from parallel agreement
”Den Ausgangspunkt für die Entstehung der Kongruenz haben solche Fälle gebildet, in denen die formelle Übereinstimmung eines Wortes mit einem andern nicht durch Rücksichtnahme auf dasselbe herbeigeführt, sondern nur durch die Gleichheit der Beziehung bedingt ist.”
“The starting-point for the origin of concord was afforded by cases in which the formal correspondence of a word with another was produced not by any regard for the latter, but merely by the identity of their relation.”

Formal control agreement terminology in Paul 1880
”Namentlich entsteht eine Verlegenheit des Sprechenden da, wo eine grammatische Kongruenz zwischen zwei Satzteilen dem Sinne nach nicht möglich ist und dazu ein dritter Satzteil tritt, von dem man gewohnt ist, dass er mit beiden kongruiert. Man muss sich für einen von den beiden entscheiden <…>”
“The speaker is especially apt to feel perplexity in cases where a grammatical concord is from the sense impossible, and a third clause comes in which custom has led us to expect to agree with both. We have to decide in favor of one or the other <…>”
It was my orders
Das sind zwei verschiedene Dinge.”

Formal control-style understanding of agreement
Formal control-style understanding dominates in modern linguistics and psycholinguistics
This has a consequence: desire to narrow down the notion of agreement
Kibrik 2011 – narrow, syntactic understanding of agreement
In the domain of argument-predicate agreement, primarily the Germanic pattern, most remote from discourse reference
However, in the context of this workshop I allow a broader, discourse-oriented understanding of the term “agreement”
In order to do that we will need to lift the formal-control requirement towards agreement

Agreement and reference
Agreement has much in common with reduced reference
Person agreement on the verb goes back to reduced reference (pronouns) (Paul 1880/1891: 348-349; Siewierska 2004)
The same often applies to attributive agreement
Russian
bel-yj < běl-ъ=jь lit. ‘white he’
white-M.Sg.Nom white-M.Sg.Nom=3M.Sg.Nom
bel-aja < běl-a=ja lit. ‘white she’
white-F.Sg.Nom white-F.Sg.Nom=3F.Sg.Nom
Classic agreement features are all referential: person, number, gender

Terminology
(person agreement)
broader agreement
broadest agreement

Terminology
(person agreement)
reduced reference
extended reduced reference

Reduced reference and agreement
In the broadest understanding of both, the extent of the included phenomena may almost coincide
There are some unusual agreement features (see Corbett 2006 on tense agreement, also cf. Paul 1880), but let us focus on major features
But the notions still remain distinct
Reduced reference is a functional notion: the process of rendering activated referents in discourse
Agreement is a linguist’s observation about the covariance of discourse constituents

Reference: the process of mentioning mental entities (referents) in discourse by means of referential expressions
The Victorian house that Ms. Johnson is inspecting has been deemed unsafe by town officials. But she asks a workman toting the bricks from the lawn to give her a boost through an open first-floor window. Once inside, she spends nearly four hours Ø measuring and diagramming each room in the 80-year-old house, Ø gathering enough information to Ø estimate what it would cost to rebuild it. She snaps photos of the buckled floors and the plaster that has fallen away from the walls.

Referential choice
Activation in working memory => reduced referential device. Else use a full device
E.g. if the referent ‘Ms. Johnson’ is highly activated, use a pronoun
How are different referential expressions, such as the eight mentions of ‘Ms. Johnson’, related to each other?
Clearly no formal control (different syntactic domains)
One can speak about agreement between them (in person, number, gender), but
such agreement is clearly an epiphenomenon of the individual mappings “referent ? referential expression”
referential expressions just happen to be in agreement or concord with each other

Syntactic anaphora?
Reference and referential choice are fundamentally discourse-based, cognitively-driven processes
Is there something like syntactic anaphora?
A mother and her child NP
I gave John his ticket Clause
I promised John to give him his ticket Closely connected clauses
To account for such syntactic usages, one can still employ a full-scale cognitively based explanation
But it may be sometimes more economical to account for syntactic usages with the help of simple and automatic rules
Including in terms of formal control from the antecedent
Antecedent functions as a placeholder, formal representative of the usual cognitive controller
Syntactic anaphora is grammaticalization or routinization of the more general process of discourse-based reduced reference

Discourse use of broader agreement (bound pronouns)
Latin (Horace, Satires 1.5: 65 ff.)
Cicirrus, Sarmentus
rogaba-t denique cur umquam fugisse-t,
ask.Impf-3Sg finally why sometime flee.Plpf.Conj-3Sg
cui satis una farr-is libra fore-t,
who.Dat enough one flour-Gen.Sg pound be.Impf.Conj-
3Sg
‘Finally he [=Cicirrus] asked why he [= Sarmentus] had ever fled, to whom one pound of flour would have been enough’
Bound tenacious pronouns

Polypersonal broader agreement (Navajo)
wónáásóó shį́į́ bimá hadah ha-b-í-ˀ-ch’-íí-yil
finally Ptcl his.mother down up.out-3.Obl-against-
Pref-4.Nom-Pfv-push
‘Finally, it appears, his mother pushed him out (of the nest)’
ts’ídá shį́į́ naˀahóóhai b-a-ˀ-í-ltsood
just Ptcl chicken 3.Obl-to-Indef.Acc-Pfv-were.fed
‘Probably at that time the chickens were fed’ (lit. ‘ something
was fed to the chickens’)
The more a language has of broad agreement, the less that looks like narrow agreement

Broader agreement
Clearly the same principles of operation as in more familiar reduced reference by free pronouns
Control from the cognitive system
Formal control treatment is ruled out (distinct syntactic domains)
Parallel referential mapping leads to parallel agreement
Related approaches
Agreement and anaphora – Bosch 1983, Barlow 1992
Semantic agreement – Dowty and Jacobson 1989
Constraint approach – Pollard and Sag 1994, Vigliocco et al. 1996, Vigliocco and Hartsuiker 2005
Important terms: unification, reconciliation of features, maximalism, notional agreement

Cooccurrence does not mean cause-effect or control relationship
Controller-target relationship?

Narrow agreement
Such as Germanic verbal person agreement
Clearly related to broad agreement
Cf. German 3Sg present –t still identical to Latin (cognate)
Can be viewed as grammaticalization of the discourse pattern (both diachronic and synchronic)
The narrower the domain, the more appropriate is the formal control approach
Agreement Hierarchy (Corbett 1979, 2006; cf. Eberhard et al. 2006)
attributive > predicate > relative pronoun > personal pronoun
increasing contribution of semantic factors

Formal control view of agreement
CONTROLLER TARGET
art nouveau
sie kommen
?? ??? rogabat
person number gender ………

Discourse-based, parallel agreement
R
art nouveau
sie kommen
quaerebat rogabat
person number gender ………

Grammaticalization of discourse-based agreement
R
CONTROLLER TARGET
art nouveau
sie kommen
rogabat
person number gender ………

Disagreement
But even in the narrow agreement there are multiple difficulties and mismatches
Because of parallel, independent mapping from the cognitive structure?
Errors (?)
In a conversational corpus I says occurs up to 50% of the time (Biber et al. 1999: 191)
Attraction or proximity effect
the key to the cabinets were missing (Bock and Middleton 2011)
“Committee contexts”

Inconsistency
Turkic person agreement
Tuvan
[men] kel­di­m
I come­Past­1Sg
‘I came’
[men] kel­gen=men
I come­Pf=1Sg
‘I have come’

Absence of explicit controller
Russian
Ja voz’m-u krasn-uju
I.Nom take.Pfv-Nonpast.1Sg red-F.Acc.Sg
‘I will take the red one’
mašina
(Fem.)
‘car’

Pulaar-Fulfulde
Detailed gender system allows easy substantivization of adjectives and participles into nouns (Koval 2006)
agreement suffix
gender suffix
on noun

First and second person problem
Even hard-core syntacticians usually do not consider 1, 2 person reference a case of anaphora (formal control from the antecedent)
John lost his wallet ANAPHORA
I lost my wallet DEIXIS
John lost my wallet DEIXIS
Are we more inclined to see agreement in Germanic 1, 2 person verbal inflection?
Ich sprech-e AGREEMENT OR DEIXIS?
Could this be an intuitive borderline between “reference as such” and “agreement as such”?
Each pronominal element is produced independently

Multiple agreement marking
Persistent indication of an activated referent in a clause
Particularly gender, sometimes in unexpected loci
Tariana (Aikhenvald 2000: 204 )
ha-dapana pa-dapana na-tape-dapana na-ya-dapana
Dem.Inan-Cl_house one-Cl_house 3Pl-medicine-Cl_house 3Pl-Poss-Cl_house
hanu-dapana heku na-ni-ni-dapana-mahka
big-Cl_house wood 3Pl-make-Topadv-Cl_house-Recpast.Nvis
‘This one big hospital of theirs has been made of wood’
Possibly, the overprotective strategy of reference (Kibrik 2011) entrenched in grammar
Or “spreading activation”

Conclusions
In terms of the extent of relevant evidence, broadly understood agreement is close to broadly understood reduced reference
The broad understanding of agreement makes us lift the formal control view
Manifestation of referential features in discourse is controlled by the cognitive structure: mapping
Observed identity of features on constituents is a result of this cognitive mapping: parallel agreement

Conclusions
Syntactic (narrow) agreement, compatible with the formal control view, is grammaticalization of the more general discourse-cognitive process
The tighter the constituent, the more likely is such grammaticalization, and this explains the Agreement Hierarchy
Frequent mismatches can be explained by independent mapping onto different constituents
These mismatches and difficulties betray the derivative character of agreement
Agreement phenomena are a periphery of the underlying process of discourse reference

Acknowledgements
Mira Bergelson
Olga Fedorova
Diana Forker
Geoffrey Haig
Antonina Koval
Hermann Paul

Thank you for your attention