Computational modelling of structural priming in dialogue

Abstract

Syntactic priming effects, modelled as increase in repetition probability shortly after a use of a syntactic rule, have the potential to improve language processing components. We model priming of syntactic rules in annotated corpora of spoken dialogue, extending previous work that was confined to selected constructions. We find that speakers are more receptive to priming from their interlocutor in task-oriented dialogue than in sponaneous conversation. Low-frequency rules are more likely to show priming.

Automatic Measures to Characterise Verbal Alignment in Human-Agent Interaction

“…In particular, alignment theory predicts the existence of patterns of repetition via a priming mechanism stating that "encountering an utterance that activates a particular representation makes it more likely that the person will subsequently produce an utterance that uses that representation" (Pickering and Garrod, 2004). Thus, DPs tend to reuse lexical as well as syntactic structure (Reitter et al, 2006;Ward and Litman, 2007). One consequence of successful alignment at several levels between DPs is a certain repetitiveness in dialogue and the development of a lexicon of fixed expressions established during dialogue (Pickering and Garrod, 2004).…”

“…(Fusaroli and Tyln, 2016) employ (cross-)recurrence quantification analysis to quantify interactive alignment and interpersonal synergy at the lexical, prosodic and speech/pause levels. (Reitter et al, 2006;Ward and Litman, 2007) focus on regression models to study priming effects within a small window of time in single dialogues. (Stenchikova and Stent, 2007) use a frequencybased approach (Church, 2000) to measure adaptation between dialogues.…”

Sequence Effects in Crowdsourced Annotations

“…We limited our scope to binary and continuous responses, however it is likely that sequence effects are prevalent for multinomial and structured outputs, e.g., in discourse and parsing, where priming is known to have a significant effect (Reitter et al, 2006). Another important question for future work is whether sequence bias is detectable in expert annotators, not just crowd workers.…”

Communication Sequences and Survival Analysis

“…Degree of repetition is a useful proxy measure of engagement, and within a number of studies, comparison of levels of self-repetition and allo-repetition between turns in actual dialogue and randomized counterparts of dialogue provides a means to state when repetition in dialogue exceeds that which one might expect in random base- lines [15,20,19]. Such repetition effects have been shown to correlate with taskoriented success as proxy measures of communication success in dialogue [17,16]. This article examines idealizations alternative specifications of baselines for believable social interaction.…”