A speech-to-speech, machine translation mediated map task: An exploratory study

Abstract

The aim of this study is to investigate how the language technologies of Automatic Speech Recognition (ASR), Machine Translation (MT), and Text To Speech (TTS) synthesis affect users during an interlingual interaction. In this paper, we describe the prototype system used for the data collection, we give details of the collected data and report the results of a usability test run to assess how the users of the interlingual system evaluate the interactions in a collaborative map task. We use widely adopted usability evaluation measures: ease of use, effectiveness and users satisfaction, and look at both qualitative and quantitative measures. Results indicate that both users taking part in the dialogues (instructions giver and follower) found the system similarly satisfactory in terms of ease of learning, ease of use, and pleasantness, even if they were less satisfied with its effectiveness in supporting the task. Users employed different strategies in order to adapt to the shortcomings of the technology, such as hyper-articulation, and rewording of utterances in relation to error of the ASR. We also report the results of a comparison of the map task in two different settings – one that includes a constant video stream (“video-on”) and one that does not (“no-video.”) Surprisingly, users rated the no-video setting consistently better.