This paper describes three psycholinguistic experiments on anaphoric resolution during sentence comprehension in Japanese. In Japanese language, there is a postposition “wa” which signifies “topic” and “ga” which signifies “(grammatical) subject”. Experiment 1 investigated the influence of the difference between “wa” and “ga” to assign ambiguous pronouns. In a self-paced reading paradigm, reading times were longer subject noun phrase than object noun phrase, irrespective of the difference of postposition, and there were more assignment to the topic noun phrase (Topic-NP: “NP+wa”) than subject noun phase (Subject-NP: “NP+ga”). In a probe recognition task, reaction times (RTs) were faster subject (S-probe) than object (O-probe), and RTs for S-probe were faster Topic-NP sentence than Subject-NP sentence. Overall, the influence of topic was observed, thus suggesting that the topic assignment strategy (TAS) were utilized during the assignment of pronouns in Japanese. Experiment 2 investigated the influence of another heuristic strategies which have been proposed to account for the assignment of pronouns in sentences in English: the subject assignment strategy (SAS) and the parallel function strategy (PFS). Furthermore, Experiment 3 investigated to distinguish between the heuristic strategies: TAS, SAS, and PFS. In both Experiment 2 and 3, there was a strong preference assigning a pronoun to the preceding Topic-NP, thus showing TAS was predominantly used over other heuristic strategies. These findings support that Japanese has a linguistic nature of a topic prominent language.

View full abstract

In this paper, we propose a cognitive process model of spoken dialogue. In order to make an interactive dialogue system, we need two management processes: one is an understanding process which manages the subprocess of utterance understanding through response generation; the other is a dialogue management process which aggregates the utterances to the discourse segment and manages focus and intentions of dialogue. Furthermore, in applying the model to spoken dialogue systems, we have to deal with misunderstandings caused by speech recognition errors. Our model specifies the cognitive process of whole dialogue understanding process and stipulates the interaction between understanding process and dialogue management process. We also specify the recovering method from communication errors into this cognitive process. Therefore, our model is suitable for implementing cooperative spoken dialogue systems.

View full abstract

This paper attempts to elucidate the nature of ‘emotive sentences’ like Maa, ureshii! ‘Wow, how happy (Iam)!’ in which the speaker involuntarily expresses his emotion, with special focus on the manifestation of experiencer subjects.
The grammatical person of the experiencer subject that an emotive predicate takes is constrained by the mood of the sentence in which it appears. On the one hand, if an emotive predicate appears in a sentence in ‘the mood of the speaker's statement’, the person of its subject is only pragmatically controlled. While the first person is most often found, other persons are also allowed in certain pragmatic contexts. On the other hand, the first person is the only possibility in sentences in ‘the mood of involuntary expression of an emotion’.
Furthermore, several syntactic tests reveal that predicates of sentences in the latter mood do not syntactically manifest either the experiencer or the source of an emotion. In other words, emotive predicates solely constitute sentences in this particular grammatical mood.
Put differently, since neither the experiencer nor the source of an emotion is realized in a one-word emotive sentence, the values of these implied arguments must be inferred from the context of utterance of that sentence.Consequently, the experiencer is automatically understood as the speaker, and the source as a concurring event in the same context.

View full abstract

This paper addresses the status of deixis and anaphoricity in Japanese demonstrative system. It has traditionally been claimed that the Japanese demostrative prefixes ko, so, and a all have both deictic and anaphoric usages. I shall argue that the core notion of ko, and a are deictic, in the sense that the expressions with these demonstrative prefixes make direct reference to an entity whose existence is recognized by the speaker prior to the discourse session in question, while that of so is not. I will show that even in cases in which they are treated as anaphoric both ko and a are ‘deictic’, in the sense given. In the case of a-series, the entity referred to is located in the ‘episodic memory’, whereas in the case of ko-series, the relevant entity is a‘discourse topic’. The contrast between the ‘episodic memory’ and the‘discourse topic’ can be characterized in terms of‘remote’ vs.‘close’, which are the key notions in characterizing the deictic usages of a and ko, thereby suggesting that the non-deictic usages of a-/ko-series are derived from their deictic usages. I further argue that the various usages of the so-series demonstratives can be best described by assuming that so, unlike a and ko, does not refer an object by direct reference. I demonstrate in particular that the so-series cannot be used to directly refer to an entity whose existence is assumed by the speaker, while it can be used to refer to an entity that is introduced into the discourse by some linguistic expressions. The claim that so is non-‘deictic’ in nature is further supported by the observation that the so-series, but not a-/ko-series, allows a usage that cannot be expressed by‘deixis’, such as the distributive interpretation.

View full abstract

This paper shows that there is a direct connection between dialogues and belief sharing. This connection is shown by proving a correspondence between observational equivalence between dialogues and epistemic bisimulation between Hyper-Discourse Representation Structures (Hyper-DRS) constructed from dialogues. An observational equivalence between dialogues is defined by a kind of similarity of resulting shared beliefs of the dialogues. The theory of Hyper-DRSs is defined by extending Kamp's Discourse Representation Theory (Kamp and Reyle 1993), a formal semantics of discourse or a dynamic semantics, which is exploiting the theory of hypersets (Aczel 1987;Barwise and Moss 1996) in order to solve the problem of the definition, formation and revision of circular objects like shared beliefs.

View full abstract

In this paper, we show our corpus for emergent dialogues. Emergent dialogues have two features, improvisation and creativity. Improvisation means we have to achieve our aim while we are looking for an action required on the spot each other during a dialogue. Although we try to have a dialogue based on a prior plan about the content of the dialogue, we would fail the dialogue. Creativity means we sometimes conceive a new idea after we hear our partner's utterance. To convey the idea to our partner have influence on our partner's thought. These two features are important in human spontaneous dialogues. Our aim is to construct a dialogue management model for emergent dialogue, and to implement it on a computer. We conducted an experiment using tasks in order to collect emergent dialogues. The solutions of the tasks and their correct answers are not clear. We used two experimental settings, “visual and audio” and “audio only”. We report features in relation to how turn-taking were operated and how agreement expressions were used to come to mutual agreement in the dialogue.

View full abstract