The influence of prominence cues in 7- to 10-year-olds’ pronoun resolution: Disentangling order of mention, grammatical role, and semantic role

Abstract

In two visual world experiments we disentangled the influence of order of mention (first vs. second mention), grammatical role (subject vs object), and semantic role (proto-agent vs proto-patient) on 7- to 10-year-olds’ real-time interpretation of German pronouns. Children listened to SVO or OVS sentences containing active accusative verbs (küssen “to kiss”) in Experiment 1 (N = 72), or dative object-experiencer verbs (gefallen “to like”) in Experiment 2 (N = 64). This was followed by the personal pronoun er or the demonstrative pronoun der. Interpretive preferences for er were most robust when high prominence cues (first mention, subject, proto-agent) were aligned onto the same entity; and the same applied to der for low prominence cues (second mention, object, proto-patient). These preferences were reduced in conditions where cues were misaligned, and there was evidence that each cue independently influenced performance. Crucially, individual variation in age predicted adult-like weighting preferences for semantic cues (Schumacher, Roberts & Järvikivi, 2017).

The Influence of Focus Marking on Pronoun Resolution in Dialogue Context

“…Therefore, prosody is simply not as stable of a cue for these children, so may be activated relatively less than for adults. Indeed, this fits previous work showing that young children do not initially use cues in the same way as adults ( Pyykkönen et al, 2010 ; Järvikivi et al, 2014 ; Goodrich Smith and Hudson Kam, 2015 ; Blything et al, 2021 ). The overall extent to which individual children resembled adults is likely dependent on a number of factors, ranging from whether they have the required language experience and developed the same robust preferences for established cues (e.g., subjecthood), as well as the processing skills required to appropriately activate and suppress multiple cues ( Järvikivi et al, 2014 ; Hartshorne et al, 2015 ).…”

“…A misalignment of cues may also explain why, unlike adults, prosody did not display a clear effect on children in object focus conditions. In these conditions, prosody was misaligned against established cues (subjectivity, first mention, and agentivity) and, compared to adults, children’s interpretive preferences are weakened by misaligned cues to a greater extent ( Blything et al, 2021 ). Related, whilst sensitivity to linguistic functions of prosodic focus marking starts to appear between 3 and 6 years of age, it is still developing during this age window ( Moore et al, 1993 ; Wells et al, 2004 ; Arnhold et al, 2016 ).…”

“…Studies with flexible word order languages, such as German and Finnish, have allowed for the partial disentangling of these different cues, given that in an OVS word order the first-mentioned entity is instead aligned to the object and typically the patient role. These studies have shown that interpretative preferences appear most robust when grammatical role, order of mention, and semantic cues are aligned (SVO), but are significantly weakened when the cues are put into conflict (OVS), both for adults ( Järvikivi et al, 2005 ; Kaiser and Trueswell, 2008 ; Schumacher et al, 2017 ) and even more clearly, for children ( Blything et al, 2021 ). In fact, it has often been suggested that children differ from adults in terms of the ability to appropriately use (or suppress) multiple cues, which would hinder their ability to interpret the pronoun, though with a potential role for individual differences ( Järvikivi et al, 2014 ; Hartshorne et al, 2015 ).…”

The Importance of Linguistic Factors:HeLikes Subject Referents

“…If order of mention affects the subject preference, we expect more subject choices in SVO order than in OVS order, and more object choices in OVS than in SVO order. As several studies found that referent preferences can be clearer when several effects combine (e.g., Blything et al., 2021; Kaiser & Trueswell, 2008), it is also possible that the subjecthood effect is modulated by information structure and/or word order.…”

“…In order to disentangle grammatical role (subject vs. object) and linear order (first‐ vs. second‐mention), previous research has often investigated pronoun resolution in languages with flexible word order, such as German and Finnish (Blything, Iraola Azpiroz, Allen, Hert, & Järvikivi, 2021; Bouma & Hopp, 2006; Järvikivi, van Gompel, Hyönä, & Bertram, 2005; Järvikivi, van Gompel, & Hyönä, 2017; Kaiser, 2011; Kaiser & Trueswell, 2008; Sauermann & Gagarina, 2017; Schumacher, Dangl, & Uzun, 2016; Schumacher et al., 2017).…”