Brief ArticleThe role of prosody in the interpretation of structural ambiguities: A study of anticipatory eye movements
Introduction
It has been established that listeners can use prosody to resolve syntactic attachment ambiguities such as early and late closure (e.g. Kjeelgaard and Speer, 1999, Schafer et al., 2000) and prepositional phrase attachments (e.g. Pynte and Prieur, 1996, Snedeker and Trueswell, 2003). The distribution of prosodic phrase boundaries and their realisation (e.g. boundary tones, final lengthening, pausing), as well as the presence or absence of pitch accents, have been shown to guide listeners' interpretation of attachment ambiguities. In this paper, we further investigate whether prosody can override strong syntactic preferences to resolve grammatical function assignment ambiguities in German.
German uses morphological case to mark grammatical function. Although four cases can be clearly distinguished, the system often features syncretism: In many NPs, nominative and accusative case share surface form. While in general, case marking is available for disambiguation, it often fails to discriminate between the two least oblique, and therefore most frequent, grammatical functions (Keenan & Comrie, 1977), namely Subject (nominative case) and direct Object (accusative case). Furthermore, German is a language with relatively free constituent order. The initial position in matrix declaratives observes very few restrictions regarding the kind of constituent it can host, which includes Subjects, Objects, as well as modifiers. As a result, both Subject-Verb-Object (SVO) and Object-Verb-Subject (OVS) surface orders are possible, and case marking on the Subject and Object may be ambiguous. In addition to the clear Subject-first preference (Hemforth, 1993), incorrect initial interpretation of constituent order ambiguity typically results in a much stronger garden-path effect than modifier attachment ambiguities. One possible reason for this is that reanalysis from an SVO to an OVS structure entails a complete reassignment of the verb's roles to both arguments.
In this study, we examine whether prosody can fill the functional gap arising from a combination of syncretism and free constituent order in German. We investigate whether, in the absence of unambiguous morphological and configurational information, prosody can influence the assignment of grammatical function. To assess this we observed anticipatory eye movements of listeners while they were inspecting a related scene. The timing and pattern of eye fixations to visually displayed referents can be used to draw inferences about spoken-language comprehension (Tanenhaus, Spivey-Knowlton, Eberhard, & Sedivy, 1995). Not only has it repeatedly been shown that referents in a scene are identified as soon as they are referred to in an utterance, there are several studies revealing that they can be identified prior to their mention. Kamide, Scheepers, and Altmann (2003), for example, have shown that unambiguous case marking, combined with verb selectional information, leads to post-verbal anticipatory eye movements in German SVO and OVS sentences. That is, immediately following the verb, and before hearing the second argument, listeners were able to use case to assign the appropriate grammatical function to the first argument and combine this with the semantics of the verb, resulting in increased anticipatory fixations to the appropriate second argument.
A recent eye-movement study by Knoeferle and colleagues (Knoeferle, Crocker, Scheepers, & Pickering, 2005), investigated the interpretation of German SVO and OVS sentences with case-ambiguous initial NPs. Structural disambiguation took place only at a second NP that was clearly case marked as either nominative (Subject) or accusative (Object). In their scenes, however, depicted actions were potentially able to resolve the ambiguity as soon as the verb was encountered. Their findings again revealed anticipatory post-verbal eye movements to the appropriate second argument, indicating that listeners were able to use depicted events to resolve the ambiguity, and assign grammatical functions appropriately.
In the present study, we build upon the studies of Kamide, Scheepers, and Altmann, (2003) and Knoeferle et al. (2005), by considering the interpretation of German SVO and OVS structures with sentence-initial NPs that were ambiguously case marked as nominative or accusative. Morphosyntactic disambiguation of grammatical functions took place at the second NP that was clearly case marked as either accusative or nominative. Scenes accompanying the sentences showed plausible Patients and Agents for the referent of the first NP in relation to a given action.1 Actions were not depicted. Thus, even though the scene presented potential referents it could not help with disambiguating thematic roles. In contrast with previous studies, however, prosodic cues could potentially help listeners resolve the temporary SVO/OVS ambiguity.
Post-verbal anticipatory eye movements to the appropriate Patient or Agent were expected to inform us about listeners' interpretation of the sentence-initial NP: looks to the Patient imply the initial NP was interpreted as Subject (and therefore Agent), whereas looks to the Agent imply the initial NP was interpreted as Object (and therefore Patient). If intonation patterns influence the interpretation of grammatical functions, then listeners in our study should anticipate more often the Patient when the sentence begins with an SVO-type intonation, and the Agent if it begins with an OVS-type intonation.
Section snippets
Method
Participants. Thirty-two native speakers of German, all students at Saarland University, took part in the experiment for monetary compensation. They had not participated in the pre-testing of the materials, and they had normal or corrected-to-normal vision and normal hearing.
Materials. Sixteen German nouns referring to animates were chosen as ambiguous characters (e.g. Katze, ‘cat’), which could serve as either the Agent or the Patient of a given action (e.g. jagen, ‘chasing’). For each
Results and discussion
One scene was displayed distortedly during the experiment and had to be removed from analysis. For the remaining 15 items, we computed proportions of inspections launched during the presentation of the verb and the adverb. The two regions following the ambiguous first NP are critical because during that time eye movements to anticipated Patients or Agents occur before the disambiguating second NP refers to them. We report hierarchical log-linear models with the two factors character type (two
General discussion
In the present study we have shown that in the absence of clear morphological and configurational information, prosody can influence the assignment of grammatical function in German. When presented with temporarily ambiguous SVO and OVS sentences, prosodic cues provided by our speaker modulated listeners' Subject and Object assignment: Listeners interpreted case-ambiguous NP1s more often as Subject and expected an Object as upcoming argument only when sentence beginnings carried what we refer
Acknowledgements
This research was funded by SFB 378 ‘ALPHA’ to the first and third author, awarded by the German Research Foundation DFG. We are grateful to Berthold Crysmann for discussion of syntactic issues and for comments on an earlier draft.
References (21)
- et al.
The role of discourse context in the processing of a flexible word-order language
Cognition
(2004) - et al.
The influence of the immediate visual context on incremental thematic role-assignment: evidence from eye movements in depicted events
Cognition
(2005) - et al.
Using prosody to avoid ambiguity: Effects of speaker awareness and referential context
Journal of Memory and Language
(2003) - et al.
Now you see it, now you don't: Mediating the mapping between language and the visual world
- Baumann, S., and Grice, M. (in press). The intonation of accessibility. Special issue on Intonation and Pragmatics....
- et al.
Accenting accessible Information
In Proceedings of Speech Prosody
(2004) Discourse, Consciousness, and time
(1984)Saccadic reaction time: Implications for reading, dyslexia and visual cognition
- et al.
Deutsche Intonation und GToBI
Linguistische Berichte
(2002) - et al.
German intonation in autosegmental-metrical phonology