The role of animacy in the real time comprehension of Mandarin Chinese: Evidence from auditory event-related brain potentials
Introduction
Within the cognitive neuroscience of language, animacy-related phenomena have long stood out as a central domain of scientific investigation. At the word level, for example, the recognition of animate vs. inanimate entities has been shown to correlate with different neuroanatomical activation patterns (see, for example, Caramazza and Mahon, 2005, Chao et al., 1999). At the sentence level, animacy distinctions between arguments have been shown to ameliorate the comprehension of non-canonical word orders in both aphasic patients (Caramazza & Zurif, 1976) and non-impaired comprehenders (e.g. Chen, West, Waters, & Caplan, 2006; for similar findings from eye movements, see Traxler, Morris, & Seely, 2002) and to interact with plausibility-based assignments of arguments to thematic roles (e.g. Kim and Osterhout, 2005, Kuperberg et al., 2007). Finally, recent results from functional magnetic resonance imaging (fMRI) suggest that even in the comprehension of simple sentences (e.g. The man took care of the garden vs. The man took care of the director), sentences with an animate actor (subject) and an inanimate undergoer (object) are easier to process than sentences deviating from this pattern in terms of animacy (Grewe et al., 2007).
The undisputed saliency of animacy as a semantic feature of arguments raises the question of whether and how this feature serves to determine aspects of incremental comprehension at the sentence level. This question is closely tied to the issue of which representations are assigned during real time sentence interpretation. As an example, consider what happens when the language comprehension system encounters a sentence-initial noun phrase (NP) such as the teacher. On the one hand, this NP could be assigned a formal analysis (e.g. in terms of integration into a particular structural position or via the assignment of a formal status such as “subject”). On the other hand, the processing system may adopt a somewhat “deeper” interpretation strategy and analyse the teacher as the agent or, more generally, the actor of the event described by the sentence. Both theoretical positions have been argued for in the sentence processing literature (see, for example, Bever, 1974, for the assumption of an Agent–Action–Object strategy for English; and Frazier, 1987, for a structurally-based mechanism). Crucially, the two types of approaches lead to different predictions with respect to the role of the animacy status of the initial NP. From the perspective of a purely structural or grammatical function-based mechanism, animacy should not be expected to influence the analysis assigned, while semantic role (actor)-based approaches should expect a modulating influence of animacy as a feature determining role prototypicality.
If it is indeed the case that initial arguments are preferentially analyzed as agents or actors (as predicted, for example, by Bever’s “canonical sentoid strategy”), an inanimate initial NP should generally lead to increased processing costs because inanimate entities are not ideal agents/actors. Converging evidence for this assumption stems from a study by Weckerly and Kutas (1999), in which event-related brain potentials (ERPs) revealed a centro-parietal negativity (N400) for inanimate vs. animate sentence-initial arguments in English. However, findings from other languages suggest that the disadvantage for inanimate initial NPs is language-specific rather than universal. ERP results in Turkish (Demiral, Schlesewsky, & Bornkessel-Schlesewsky, in press) and German (see Bornkessel & Schlesewsky, 2006) revealed no differences between inanimate and animate initial arguments. This difference between English on the one hand and German and Turkish on the other can be derived within a recent neurocognitive model of cross-linguistic language comprehension, the extended Argument Dependency Model (eADM; Bornkessel & Schlesewsky, 2006). Within the eADM, English is classified as a language in which incremental argument interpretation takes place primarily on the basis of the linear position of the arguments. Thus, the processing system generally endeavors to analyze the first argument it encounters as bearing the highest generalized semantic role (actor). In the case of an inanimate first NP, this leads to a processing conflict, which is expressed in the form of an N400. In languages such as German and Turkish, by contrast, incremental interpretation proceeds on the basis of morphological (case marking) information. Since the mapping between individual cases and particular semantic roles is not as strict as that between position and semantic roles in English, animacy only plays a secondary role in the processing of the first NP in languages of this type.1 Nevertheless, as soon as two or more arguments must be integrated and interpreted relative to one another, animacy becomes important even in languages like German (see Frisch & Schlesewsky, 2001, for electrophysiological; and Grewe et al., 2007, for neuroanatomical evidence). This suggests that, rather than being used as a cue towards the interpretive role of individual arguments, animacy is drawn upon for the establishment of an interpretive hierarchy (“who is acting on whom”) amongst several arguments.
However, what role does animacy play in the incremental comprehension of languages that can draw neither upon position nor upon morphological (case) marking for argument interpretation? A case in point is Mandarin Chinese, a language with virtually no grammatical means of identifying grammatical functions or semantic roles. Should the influence of a semantic factor such as animacy not be expected to be strongest in a language of this type, which lacks other cues for incremental interpretation? The present study aimed to examine this question by investigating the neurophysiological correlates of incremental argument comprehension in Chinese argument–argument–verb constructions. In two auditory ERP experiments, we manipulated the actor-undergoer relation between the first and second argument (active vs. passive) and the animacy of both arguments. Our results show that Chinese behaves similarly to German and Turkish in that it (a) does not show an animacy effect at the position of the initial argument, while (b) animacy plays an important role in the establishment of interpretive relations between two arguments. They further support the cross-linguistic neurocognitive generalization that mapping problems between arguments and thematic/semantic roles are reflected in N400 effects.
Section snippets
Experiment 1
Experiment 1 was designed to investigate the effect of animacy on the real time comprehension of spoken Mandarin Chinese. In particular, we wanted to establish whether animacy plays a crucial role in the interpretation of an initial argument and whether animacy influences the establishment of relations between arguments (“who is acting on whom”). To this end, we employed the experimental design illustrated in Table 1.
The critical sentences in Table 1 instantiate a fully crossed 2 × 2 × 2 design. In
Experiment 2
In our interpretation of the results of Experiment 1, we suggested that the N400 observed at the position of the passive-marker bèi resulted from a mismatch between the construction-specific thematic-pragmatic requirements of the adversity passive (psychological affectedness of the undergoer) and the inanimacy of the first argument. If this is indeed the case, a bèi-passive should generally lead the processing system to predict an animate (human) undergoer. Experiment 2 tested this hypothesis
General discussion
We have presented two auditory ERP experiments on the processing of animacy in active and passive sentences in Mandarin Chinese. While Experiment 1 employed main clauses with an NP-coverb-NP-V structure, Experiment 2 manipulated the temporal order in which the critical information types became available by using relative clauses with a coverb-NP-V-RC-marker-NP structure. The two main findings of both experiments were: (a) an N400 for inanimate undergoers in bèi-passive construction; and (b) an
Conclusions
In two auditory ERP studies of Mandarin Chinese, we were able to show: (a) that the N400 is a robust cross-linguistic correlate of increased processing costs related to argument prominence (the endeavor to establish “who is acting on whom” using a variety of information types); (b) that animacy is processed as a relational rather than as an inherent feature of arguments during online sentence comprehension in Chinese; and (c) that language-specific pragmatic properties that are relevant for
Acknowledgments
The research reported here was supported by grants from the German Research Foundation (BO 2471/2-1/SCHL 544/2-1 and BO 2471/3-1) and was conducted in collaboration with the Clinic for Audiology and Phoniatry (Prof. Manfred Gross) of the Charité Berlin. We are grateful to Katja Bruening for invaluable assistance in data acquisition. We would also like to thank Anna Xian Hong for assistance in the preparation of stimulus materials and Tong Fei, Wang Luming, Hung Yu-Chen and Bai Chen for helpful
References (59)
- et al.
Multi-dimensional contributions to garden path strength: Dissociating phrase structure from case marking
Journal of Memory and Language
(2004) - et al.
Dissociation of algorithmic and heuristic processes in language comprehension: Evidence from aphasia
Brain and Language
(1976) - et al.
Determinants of BOLD signal correlates of processing object-extracted relative clauses
Cortex
(2006) - et al.
A rose by any other name: Long-term memory structure and sentence processing
Journal of Memory and Language
(1999) - et al.
The resolution of case conflicts from a neurophysiological perspective
Cognitive Brain Research
(2005) Linguistic complexity: Locality of syntactic dependencies
Cognition
(1998)- et al.
The role of the posterior superior temporal sulcus in the processing of unmarked transitivity
Neuroimage
(2007) - et al.
The independence of combinatory semantic processing: Evidence from event-related potentials
Journal of Memory and Language
(2005) - et al.
The role of animacy and thematic relationships in processing active English sentence: Evidence from event-related potentials
Brain and Language
(2007) - et al.
The role of structural prediction in rapid syntactic analysis
Brain and Language
(2006)