Does pre-planning explain why predictability affects reference production?

Abstract

How does thematic role predictability affect reference production? This study tests a planning facilitation hypothesis – that the predictability effect on reference form can be explained in terms of the time course of utterance planning. In a discourse production task, participants viewed two sequential event pictures, listened to a description of the first picture (depicting a transfer event between two characters), and then provided a description of the second picture (continuing with one thematic role character, either goal or source). We replicated previous findings that goal continuations lead to more reduced forms of reference and shorter latency to begin speaking than source continuations. Additionally, we tracked speakers’ eye movements in two periods of utterance planning, early vs. late. We found that 1) early pre-planning supports the use of reduced forms but is not affected by thematic role; 2) thematic role only affects late planning; and 3) in contrast with our hypothesis, planning does not account for predictability effects on reduced forms. We then speculate that discourse connectedness drives the thematic role predictability effect on reference form choice.

A systematic evaluation of factors affecting referring expression choice in passage completion tasks

“…This study, as well as its replication in experiment 2, did find a predictability effect on pronominalization. Furthermore, we note predictability effects on referring expression choice (pronouns and null pronouns) in studies that use a picture description task in English ( Zerkle and Arnold, 2019 ) and Spanish ( Medina Fetterman et al, 2022 ).…”

“… Arnold and Zerkle (2019) discuss the issue of a possible effect of cost on referring expression production in detail and review related literature. In fact, there is mixed evidence: no effects of production difficulty on pronominalization have been found in studies measuring speech onset latencies ( Rosa and Arnold, 2017 , Zerkle and Arnold, 2019 ), while effects of cognitive load on pronominalization have been reported in other studies, with some studies showing an increase in pronoun production with cognitive load ( Vogels et al, 2015 ), and others showing a decrease ( Arnold and Griffin, 2007 ). In sum, Arnold and Zerkle (2019) argue that while there is some support for predictability on pronoun production, there is little evidence for an efficiency-based account.…”

“…Speakers' reductions of redundant elements have primarily been investigated using two research paradigms: one which attempts to correlate referent predictability in corpora with referring expression choice; and one which uses passage completion, with tightly controlled prompts, to determine whether likelihood of referent mention correlates with referring expression choice. Recently, studies using a picture description task have also been used ( Zerkle and Arnold, 2019 , Medina Fetterman et al, 2022 ). Based on work following the corpus-based paradigm, there is evidence both for ( Resnik, 1996 , Kravtchenko, 2014 , Tily & Piantadosi, 2009 , indirectly) and against ( Modi et al, 2017 ) the UID hypothesis, which is difficult to reconcile given substantial differences in corpus size, text type, predominant varieties of referring expressions seen in the corpus, and relatively weak results in the positive studies.…”

“…Similarly, there is evidence both for ( Arnold, 2001 , Bott & Solstad, 2023 , Medina Fetterman et al, 2022 , Rosa and Arnold, 2017 , Vogels, 2019 , Weatherford and Arnold, 2021 , Zerkle and Arnold, 2019 ) and against ( Fukumura and van Gompel, 2010 , Rohde and Kehler, 2014 , Holler and Suckow, 2016 ) the same hypothesis in the passage completion and picture description literature – similarly difficult to reconcile given subtle differences in stimulus design, the verb type used in prompts, task type, referring expression ambiguity, and the degree to which the task simulated natural discourse and/or dialogue.…”

“…Many studies that have used a passage completion task have done so in a non-interactive setting where the participant is presented with passages and is asked to complete them, see Table 1 . An exception to this is a series of more recent studies that have aimed at creating more naturalistic and interactive settings ( Rosa and Arnold, 2017 , Zerkle and Arnold, 2019 , Weatherford and Arnold, 2021 , Medina Fetterman et al, 2022 ). Rosa and Arnold (2017) posit that the use of a written passage completion paradigm, with tightly controlled stimuli devoid of any discourse context, divorces the task from natural discourse production.…”