S. Oviatt. Predicting spoken disfluencies during human-computer interaction. Computer Speech and Language, 9:19--35, 1995.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....affects DF distribution. Previous studies have shown that DF rates and the frequency and distribution of particular types of DFs vary according to the scenario and task details. Furthermore, longer, more spontaneous utterances tend to be more disfluent than briefer, more structured utterances [9, 13]. Moreover, individual predispositions are important. It has been shown that some speakers are consistently more disfluent than others [2, 13] Other factors, such as planning difficulties, speech rate, confidence, social relationships and gender have also been discussed in conjunction with DFs ....

....disfluent at all. Individual variation thus exceeds most other kinds of factors in explaining DF rates. In our data, factors such as gender, age and computer skill had no effect on DF rates. Figure 3. The average DF word rates for turns with five to nine words. Dialogue State According to Oviatt [9], the structure of the dialogue affects the manner in which users interact with a spoken or multimodal dialogue system. A system that employs an unconstrained format will encourage its users to produce utterances with higher information per utterance ratio than users who are prompted for more ....

[Article contains additional citation context not shown here]

Oviatt, S.L. 1995. Predicting spoken disfluencies during human-- computer interaction.ComputerSpeechandLanguage, 9, pp.19--35.

....contributions to this problem. As mentioned earlier here, the skills contributing to the designs of auditory displays range from computer science and psychology to the auditory and music engineering disciplines. Multimodal Input and Output Studies in our lab, as well as the research of others [8], have demonstrated that users relish the chance to be able to use gestures and speech in combination with computing tasks, including those on the Web. Given the current state of speech recognition systems and the breadth of speech commands available to users for any given task, the combination of ....

....and speech in combination with computing tasks, including those on the Web. Given the current state of speech recognition systems and the breadth of speech commands available to users for any given task, the combination of speech and gesture is a strong and complimentary HCI partnership. As Oviatt [8] has shown, the combination of speech and gestures during HCI actually reduces disfluencies in speech while improving accuracy of both the speech and gestural recognition systems. Understanding the design issues involved in user interfaces requires expertise in speech recognition, linguistics, ....

Oviatt, S. Predicting spoken disfluencies during human-computer interaction. Computer Speech and Language, 9 (1995), 19--35.

....on the difficulties we have getting users to behave as they need to when interacting with perfectly adequate spoken dialogue systems. Users repeat themselves needlessly, mistake when it is their turn to speak, and otherwise behave in ways that make dialogue systems less likely to function well [20]. It is in situations just like these in life that the non verbal modalities come in to play: in noisy situations, humans depend on access to more than one modality [26] This leads us to the third reason we might wish to add the non verbal modalities to dialogue systems. While humans have long ....

Oviatt, S. L., "Predicting spoken disfluencies during human-computer interaction," Computer Speech and Language, vol. 9, pp. 19-35, 1995.

....our interactions with others all in real time. It is thus hardly surprising that we sometimes need to pause, or to repair our previous speech. Rates of disfluency per word in spontaneous English speech vary from under 1 for constrained human computer dialog, to 5 10 for natural conversations [12, 16, 4, 19]. There is also considerable variation across speaking environments in the relative rates of particular disfluency types [12, 19] Such distributional differences reflect differences in cognitive demands, and in managing interaction in conversation [9, 2] While considerable past work has focused ....

Oviatt, S. L. 1995. Predicting spoken disfluencies during humancomputer interaction. Computer Speech and Language, 9, 19--35.

.... this corpus consists of interactive problem solving speech between young (pre university) adults and since the task domain involves navigating through a map (i.e. the extensive use of spatial expressions) one would expect that this corpus would contain a relatively large number of disfluencies [Oviatt 95] and indeed it does. 7.1 Aborted constituents To start with, we look at the occurances of aborted constituents in the corpus. Tables 1 and 2 show the number of aborted major constituents in the corpus. Table 1: All phrasal clausal categories, ignoring self nesting category Total Aborted ....

Oviatt,S.L., "Predicting spoken disfluencies during human-computer interaction", Computer Speech and Language, 9, 1995.

....that Givers should produce more disfluencies. However, there is also a possible confound between role and utterance length. Givers tend to produce longer utterances than Followers. Some previous research has suggested that longer utterances have higher disfluency rates than shorter utterances [5]. Hence Givers might have higher disfluency rates simply because they tend to produce longer utterances. 1.3.3. Familiarity. As stated above, each speaker took part in dialogues with one partner with whom they were familiar, as well as with a partner with whom they were not familiar. Familiarity ....

Oviatt, S. 1995. Predicting spoken disfluencies during humancomputer interaction. Computer Speech and Language 9, 9-15.

....but probably stem from a number of factors including poor articulation, increased coarticulation, highly variable speaking rate, and various types of disfluency such as hesititations, false starts and corrections. Some of the linguistic aspects of spontaneous speech have been studied in some detail[84] and a number of specific issues have been addressed such as identifying repairs in speech[43] modeling non speech interjections [94] detecting hesitations [79] and modelling the timing patterns of disfluent speech[80] However much of this work is at a very early stage. Practical ....

Oviatt S. Predicting Spoken Disfluencies during Human-Computer Interaction. Computer Speech Language, Vol 9, No 1, pp19-36, 1995

No context found.

Oviatt, S. L. Predicting spoken disfluencies during human-computer interaction, Computer Speech and Language, 1995, 9, 1, 19-35.

....features of children s spoken conversations with an animated character. For example, disfluencies are widely recognized as presenting a major challenge and source of failure for current spoken language systems. Although they have been analyzed quite extensively in adult speech to computers ([6], 5] they have yet to be evaluated in adult speech to animated characters, or in children s speech. Previous research has shown that disfluencies are landmark features of spoken language [6] although they occur at consistently higher rates in adult interpersonal speech than in human computer ....

....spoken language systems. Although they have been analyzed quite extensively in adult speech to computers ( 6] 5] they have yet to be evaluated in adult speech to animated characters, or in children s speech. Previous research has shown that disfluencies are landmark features of spoken language [6], although they occur at consistently higher rates in adult interpersonal speech than in human computer speech [5] During adult human computer communication, the rate of disfluencies bears a positive linear relation with utterance length, which accounts for 80 of their variance ( 6] 5] Recent ....

[Article contains additional citation context not shown here]

Oviatt, S. L. 1995. "Predicting Spoken Disfluencies during Human-Computer Interaction." Comp. Speech and Lang. 9(1): 19-35.

No context found.

S. Oviatt. Predicting spoken disfluencies during human-computer interaction. Computer Speech and Language, 9:19--35, 1995.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC