Branigan et al. Do participants of a conversation syntactically converge?
Introduction • Research has already shown that speakers co-ordinate on the semantic and lexical levels • Maze study: participants converged on particular types of descriptions/lexical expressions • I.e. ‘path between 2 points’/‘box’
Co-ordination • Co-ordination: Observable convergence in participants linguistic behavior without necessarily being intentional • Helps listeners to correctly understand a speaker’s meaning • Helps speaker by decreasing computational load. • Cuts down on ambiguity
Bock • Speakers can use 2 different grammatical structures to describe one thing. • ‘x verbing y to the z’ or ‘x verbing z to the y’ • Use co-ordination of grammatical form • Adjacent utterance = local syntactic consistency, not arising from the repetition of words
Bock cont. • Speakers alternately repeated sentences and described pictures • Syntactic form of picture description repeated preceding sentence • Syntactic priming/persistence: Single speakers tend to repeat syntactic structures-can’t be explained non-syntactically, unconscious
Bock and Loebell • Syntactic priming = activation • Activation of procedure doesn’t disappear immediately, later production of the form is facilitated • Procedures associated with production and comprehension are different • Comprehension is word-by-word, production is selection of a word appearing later in an utterance determines selection of a previous word.
All of this leads to… • The hypothesis: Syntactic priming arises from residual activation of syntactic information common to production and comprehension.
The experiment • One participant, one confederate • Use confederate scripting in a dialogue game, alternating between describing pictures and selecting pictures • Manipulate the confederate form, see if participant produces co-ordinated target description • It has been shown that there is greater magnitude when the same verb was repeated between the prime and target descriptions
Method • 24 Participants • 2 sets of 48 cards that depict actions • 12 cards are ditransitive actions involving an agent, a patient and a beneficiary • 36 cards are transitive actions involving an agent and a patient-filler • 2 cards per verb • Depictions are easily recognizable/nameable • Verb is printed under the picture
Method cont. • There is a Subject’s Description Set, which is the target set, and a Confederate’s Description Set, which is the priming set • Ordered pairs with one priming to one target • 2 pairings-same verb between the priming and target, different verb between the priming and target
Method cont. • 4 scripts, each containing a description of the priming card • Per script, half of the priming cards are prepositional object descriptions. Other half are double object descriptions. • Experimental item: the confederate’s scripted description of a prime card, plus the subject’s target card paired with it
Method cont. • So…. There are 4 versions of each item: same verb, P.O. description same verb, D.O. description different verb, P.O. description different verb, D.O. description • Prime type: P.O. vs. D.O. • Verb Identity: Same verb vs. different verb
Procedure • The Subject’s Description Set is in a box, in random order (at least 2 fillers between each target), and there is an empty selection box on a table in front of the subject • Subject also has a Selection Set of cards, which are the Confederate’s Description Set with 24 additional distracter cards-1 distracter per verb • Confederate Setup is the same, but confederate also has scripts specifying the description to use for each prime card
Procedure cont. • Divider between the subject and the confederate so that they can’t see each other’s cards • The experimenter tells them that they are investigating how well people can communicate when they can’t see each other • Describe card and pick card that matches description • Subject could ask for repetition, but nothing else
Procedure cont. • Before experiment there was a practice session with 4 cards • The confederate always went first, so confederate’s description of a prime card always preceded the subjects description of the target card
Procedure cont. • Recorded on audiotape/transcribed • Coded first response the subject produced • 3 target responses with wrong verb were excluded • 285 remaining response coded • P.O. if patient of action immediately followed the verb, and was followed by the preposition ‘to’ and the beneficiary • D.O. if beneficiary immediately followed verb and was followed by patient action
Results cont. • Subjects had a tendency to produce target descriptions in the same syntactic form as the prime description • The effect was stronger when the verb was the same in both instances • 55% when verb was the same • 26% when verb was different
Discussion • Not associated with different discourse registers • Use of one form over the other can’t be due to rhetorical effects • Prime and target cards had different entities, repetition of the verb can’t account for producing certain phrases • Priming of ‘to’ can’t explain the magnitude of the results
Discussion cont. • Results show that speakers are sensitive to the characteristics of the dialogue, specifically, the linguistic behavior of the other participant • Syntactic co-ordination may be equal to the syntactic priming effect • Prior processing of a particular structure can facilitate later use of that structure
Discussion cont. • Results also show that there are shared syntactic representations underlying comprehension and production • Encoded as a part of lexical entities that are accessed during comprehension and production • Goes against Bock and Loebell’s account of syntactic priming
Discussion cont. • This model • There are nodes representing the base form of a verb linked to nodes representing grammatical features, which are linked to nodes representing combinatorial possibilities • Activation of a combinatorial node doesn’t decay immediately, so later use is facilitated • Stronger effects when the same verb appears in prime and target • Similar to Levelt et al.
Discussion cont. • Syntactic co-ordination occurs when the speaker and the listener activate shared syntactic information • Establish common syntactic ground, just like establishing common semantic ground • This can occur in natural dialogue