In this paper, we present a Conditional Random Field based approach for automatic detection of edit disfluencies in a conversational telephone corpus in French. We define disfluency patterns using both linguistic and acoustic features to perform disfluency detection. Two related tasks are considered : the first task aims at detecting the disfluent speech portion proper or reparandum, i.e. the p...

A common tool for improving the performance quality of natural language processing systems is the use of contextual information for disambiguation. Here I describe the use of a finite state machine (FSM) to disambiguate speech acts in a machine translation system. The FSM has two layers that model, respectively, the global and local structures found in naturally-occurring conversations. The FSM...

This paper proposes a dialogue control method for completing a task with a short dialogue even when user utterances include out-of-vocabulary (OOV) words. When the user utterance includes an OOV word, conventional methods try to complete a task by acquiring an understanding of the OOV word through dialogue. However, completing a task does not always require that the system understands the OOV w...

speaking assessment is still construed as a complicated, under-researched process from the vantage point of tasks and rater characteristics. the present study aimed at investigating if and how english major and none english major teachers differ in their perception of the construct of oral proficiency while assessing learners’ l2 oral proficiency.  to this end, 38 male and female non-native efl...

Dialogue systems typically follow a rigid pace of interaction where the system waits until the user has finished speaking before producing a response. Interpreting user utterances before they are completed allows a system to display more sophisticated conversational behavior, such as rapid turn-taking and appropriate use of backchannels and interruptions. We demonstrate a natural language under...

The dependency of conversational utterances on themode of dialogue is analyzed. A speech corpus of 800 speakers collected under three different modes, i.e., talking to a human operator, an WOZ system and an ASR system, is used for analysis. Some characteristics such as sentence complexity loudness of the voice and speaking-rate are found to be significantly different among the dialogue modes. L...

In this paper, the simulated annealing Q-learning (SA-Q) algorithm is adopted to automatically learn the optimal dialogue strategy of a spoken dialogue system. Several simulations and experiments considering different user behaviors and speech recognizer performance are conducted to verify the effectiveness of the SA-Q learning approach. Moreover, the automatically learned strategy is applied t...

In recent years statistical dialogue systems have gained significant attention due to their potential to be more robust to speech recognition errors. However, these systems must also be robust to changes in user behaviour caused by cognitive loading. In this paper, a statistical dialogue system providing restaurant information is evaluated in a set-up where the subjects used a driving simulator...

We propose a model for dialogue between non task oriented agents, based on the dissonance and relevance theories. Non task oriented agents are studied as a model for non expert agents, as opposed to task oriented agents, in order to provide models for social science simulations. Dialogue between non task oriented agents can not be modelled like task oriented dialogue because no criterium is pro...