Week 2 discussion notes

Discussion points for Bock (1986); Chang, Dell, & Bock (2006); Jaeger & Snider (2013)

Class discussion

Bock

Datives and passive constructions, rules out various correlated factors that can influence the production of a structure — e.g., animacy, lexical frequency, raw probabilities of syntactic structures, by experimentally manipulating the primes people experience and making the task as “implicit” as possible. Bock thinks about the problem in terms of excitation, especially within a neural network framework, and while she doesn’t include probabilities in her paper she does state that probabilities are what are being updated.Thoughts: What algorithms could be used to select a structure with the probability distribution? Note, that Bock (1986) does not directly address the learning mechanism or the algorithm by which a syntactic structure is chosen. It is conceivable that people sample from a probability distribution of syntactic structures (that is non-static and changes throughout the experiment). We can also compute something like an argmax over that distribution. Okay so what are things on the “context” side of $p(\text{PD} | \text{context})$? Ideally, just the presence or absence of a linguistic event, such as whether the primes in these experiments are one structure or another. It is critical to hold meaning constant in these cases. Likewise, all papers will hinge on syntactic alternations

Bock, J. K. (1986). Syntactic persistence in language production. Cognitive Psychology, 18, 355-387. https://doi.org/10.1016/0010-0285(86)90004-6

• Syntactic production is hard: “Selecting among the wide array of syntactic options available for the expression of a message in unplanned speech can create problems that lead to hesitations, errors, and other disruptions (Deese, 1980)” p. 379
  • Repetition effects “may ease the demands of message formulation” and “contribute to fluency”
  • Factors adjacent to frequency as a construct (activation, strength, persistence) especially for more abstract categories probably play a large role in how people solve the problem of syntactic production
  • Syntactic persistence: “The probability of a particular syntactic form being used in the description increased when that form had occurred in the prime”
  • Syntactic in nature: “this effect was not appreciably modified by variations in certain conceptual characteristics of sentences”; “some syntactic processes are organized into a functionally independent subsystem”
  • For this cognitive problem, can we define the input? Output?
    • Input:
    • Output:

• Came out at a time where psycholinguistics was very interested in “syntax first” and the modularity of syntactic processing; Bock describes a new “performance constraint” in language production
  • Despite broader grammatical knowledge, producers of a language will “employ the same syntactic form across successive utterances, suggesting the existence of additional processes that are antagonistic to the productivity of syntax in actual language use”
  • Structural repetition and lexical repetition and conceptual repetition are all often confounded — growing movement in the field to deconfound sentence stimuli
  • “influence of conceptual factors should be confined to an initial mapping from the message to an abstract syntactic representation, after which syntactic processing proceeds without further intervention from message characteristics” which contrasts with
  • “language structures are relatively direct reflections of semantic notions or communicative strategies that are generally important in human thought and social interaction (Bates & MacWhinney, 1982)”
  • “experimental paradigms in which structural variables can be manipulated independently of content variables”

• Strength or activation accounts — “the activation or strengthening of information implies a quasi-neurological energizing, excitation, or threshold reduction that persists over time, increasing the probability that the activated or strengthened information will influence subsequent cognitive processes (Anderson, 1983; Posner, 1978)”
  • Discussing speech errors in which a word appears twice within the same utterance but two syntactically related constructions: “simultaneous activation of two different syntactic structures capable of expressing the same semantic intention”
  • Activation and representational assumptions note the existence of frequency effects in production — sentences may work similarly to words except sentences usually don’t have well-defined frequencies (or e.g., algorithms for estimating the probability of a parse)
  • The set of grammatical sentences is not finite; “even less likely that sentence codes (at the level of sequences of grammatical categories, for example) might be linked to one another to represent shared syntactic properties” —

• Weiner and Labov (1983) and Levelt and Kelter (1982) both show how repetition of a structure is more likely (e.g., prepositional modifiers like ‘at what time do you close?’ or passive vs. active constructions like ‘the church was struck by lightning’)
  • Note the implicit necessity of alternations that are assumed to have equivalent meanings so far for syntactic questions
  • (a) Could be driven by maintenance of a prior sentence in memory (because you often have to repeat or modify what another person said when answering a question in e.g., the Levelt and Kelter 1982 paper)
  • (b) Persistence of communicative intentions or discourse strategies
  • Both cases can lead to greater attention to one’s speak and/or its structural features
  • Condition-action pairings — increase in activation level or strength of the procedures, raising the probability of their subsequent use
    • Some kind of learning function is at work here, but which one?
    • Hypothesis: The procedures responsible for the creation of a sentence’s structure can be activated or strengthened by use

• Domain: Production of English passive/active constructions and prepositional versus double object dative constructions
  • Manipulate prime form on an item that a participant repeats a description of out loud; then they spontaneously describe an image with a similar kind of event
  • Nuisances: Coreference, rhythmic or intonational overlap, no relationship between words or constructs in the prime and target sentences
  • Verified that participants were “not aware” of the manipulations
  • Experiment 1: PD use up 23%, DO use up 8% [similar effects in Experiment 2]

• “linguistic rules as graded rather than fully determinate” … “comparable to that incorporated into the completion model of Bates and MacWhinney (1982; MacWhinney, Bates, & Kliegl, 1984)

• competition model and parallel processing of lexical and syntactic information are both better fits than more modular proposals like those outlined in Garrett (1975, 1980, 1982)

• “pattern of priming effects is inconsistent with a model in which disjoint sets of message features constitute necessary conditions for the use of particular syntactic structures” - p. 382

Chang, Dell, and Bock (2006)

Chang, F., Dell, G. S., & Bock, K. (2006). Becoming syntactic. Psychological Review, 113, 234-272. https://doi.org/10.1037/0033-295X.113.2.234

Jaeger and Snider (2013)

Jaeger, T. F., & Snider, N. E. (2013). Alignment as a consequence of expectation adaptation: Syntactic priming is affected by the prime’s prediction error given both prior and recent experience. Cognition, 127, 57-83. https://doi.org/10.1016/j.cognition.2012.10.013