Production and comprehension show divergent constituent order preferences: Evidence from elicited pantomime

Abstract

All natural languages develop devices to communicate who did what to whom. Elicited pantomime provides one model for studying this process, by providing a window into how humans (hearing non-signers) behave in a natural communicative modality (silent gesture) without established conventions from a grammar. Most studies in this paradigm focus on production, although they sometimes make assumptions about how comprehenders would likely behave. Here, we directly assess how naïve speakers of English (Experiments 1 & 2), Korean (Experiment 1), and Turkish (Experiment 2) comprehend pantomimed descriptions of transitive events, which are either semantically reversible (Experiments 1 & 2) or not (Experiment 2). Contrary to previous assumptions, we find no evidence that Person-Person-Action sequences are ambiguous to comprehenders, who simply adopt an agent-first parsing heuristic for all constituent orders. We do find that Person-Action-Person sequences yield the most consistent interpretations, even in native speakers of SOV languages. The full range of behavior in both production and comprehension provides counter-evidence to the notion that producers’ utterances are motivated by the needs of comprehenders. Instead, we argue that production and comprehension are subject to different sets of cognitive pressures, and that the dynamic interaction between these competing pressures can help explain synchronic and diachronic constituent order phenomena in natural human languages, both signed and spoken.

Interpreting Silent Gesture: Cognitive Biases and Rational Inference in Emerging Language Systems

“…The heuristic that we have focused on in this paper (the one that connects SOV to extensional and SVO to intensional events) is different from the one that is described by Hall et al. (), but they are both clearly semantics based, and certainly compatible with one another…”

“…The interpretation of strings in the silent gesture paradigm has received little attention, with the exception of two recent studies: one in which participants are asked to recognize the intended meaning of silent gesture strings in a timed forced-choice setup (Langus & Nespor, 2010), and one in which participants are asked to choose an interpretation for ambiguous reversible events (Hall, Ahn, Mayberry, & Ferreira, 2015). Langus and Nespor (2010) asked adult participants to watch video clips of gesture sequences describing simple transitive events through space in a two-alternative forcedchoice task.…”

“…Langus and Nespor (2010) claim that this effect is due to the fact that in this task, participants disregard their computational system of grammar. Hall et al (2015) focused on the interpretation of reversible events and came to very different conclusions. They showed participants silent gesture strings that were made up of an action and two animate participants, in three possible orders (Action-Participant1-Participant2, Participant1-Action-Participant2, Participant1-Participant2-Action).…”

“…Hall et al. () focused on the interpretation of reversible events and came to very different conclusions. They showed participants silent gesture strings that were made up of an action and two animate participants, in three possible orders (Action‐Participant1‐Participant2, Participant1‐Action‐Participant2, Participant1‐Participant2‐Action).…”

“…Hall et al. () emphasize the difference between silent gesture production and interpretation. They postulate procedural, production‐related constraints for production, and a semantic heuristic (“agent first”) for interpretation.…”