SUMMARY.A sample of 130 secondary school teachers was systematically observed teaching mixed classes varying in age from 11 to 16 years. The OPTIC schedule was employed to record teachers' rates of approving and disapproving responses and the amount of time pupils spent on-task. Although rates of approval were generally higher than disapproval rates, most of the approval was directed at academic pupil behaviours whereas most of the disapproval was for inappropriate social behaviour. Mean on-task behaviour of classes at 80 per cent was higher than that observed in a similar sample of primary classes.INTRODUCTION BEHAVIOURAL approaches to classroom management commonly advocate increasing teacher approval and decreasing disapproval (see, for example, Wheldall and Merrett, 1984), since it has frequently been demonstrated that contingent teacher approval can function as a powerful reinforcer of pupil behaviour. This is true not only for primary aged children but also for young people attending secondary schools (Wheldall and Austin, 1980;Merrett and Blundell, 1982; Wheldall and Merrett, 1987;Merrett and Wheldall, 1987a). Relatively little is known, however, about the ways in which approval and disapproval are normally employed in secondary school classrooms.The earliest study conducted specifically to determine natural rates of teacher approval and disapproval appears to be that of White (1975) in the United States. Under White's guidance 16 observational studies were carried out (employing the Teacher Approval and Disapproval Observation Record) with 104 teachers in classrooms from grades 1 to 12. The findings showed generally that, with the exception of those teaching grades 1 and 2, teachers gave more disapproving than approving comments to their pupils, with overall rates declining as pupils got older. The approval rate for instructional (academic) behaviour was higher in all grades than the rate of disapproval. With reference to managerial (social) behaviour, the rate of disapproval was very much higher than the rate of approval, which was described as "almost non-existent".Heller and White (1975) subsequently observed five teachers of mathematics and five teachers of social studies each teaching a higher ability and a lower ability class in grades 7 to 9 (i.e., lower secondary aged classes). Mean rates of disapproval were again found to exceed mean rates of teacher approval. Moreover, out of the total of 1,105 statements of approval and disapproval recorded, only one was found to be approving of social behaviour. Additionally, it was found that whereas teacher approval did not differ in rate between high and low ability classes, rates of disapproval were much higher in the lower ability classes.In New Zealand, Thomas et al. (1978) observed the classroom behaviour of 10 grade 7 teachers (i.e., classes of 11-to 13-year-old pupils) in the Mangere Guidance Unit, a clinical treatment facility. A time sampling instrument was used to record teacher behaviours and the on-task behaviour level of all 10 pupils. Results si...

No abstract

This paper describes an optimal algorithm using continuous state Hidden Markov Models for solving the HMS decoding problem, which is the problem of recovering an underlying sequence of phonetic units from measurements of smoothly varying acoustic features, thus inverting the speech generation process described by Holmes, Mattingly and Shearme in a well known paper (Speech synthesis by rule, Language and Speech 7 (1964)).

No abstract

Low-dimensional 'bottleneck' features extracted from neural networks have been shown to give phoneme recognition accuracy similar to that obtained with higher-dimensional MFCCs, using GMM-HMM models. Such features have also been shown to preserve well the assumptions of speech trajectory dynamics made by dynamic models of speech such as Continuous-State HMMs. However, little is understood about how networks derive these features and how and whether they can be interpreted in terms of human speech perception and production. We analyse three-dimensional bottleneck features. We show that for vowels, their spatial representation is very close to the familiar F1:F2 vowel quadrilateral. For other classes of phonemes the features can similarly be related to phonetic and acoustic spatial representations presented in the literature. This suggests that these networks derive representations specific to particular phonetic categories, with properties similar to those used by human perception. The representation of the full set of phonemes in the bottleneck space is consistent with a hypothesized comprehensive model of speech perception and also with models of speech perception such as prototype theory.

No abstract

No abstract

Recent advances in automatic speech recognition have used large corpora and powerful computational resources to train complex statistical models from high-dimensional features, to attempt to capture all the variability found in natural speech. Such models are difficult to interpret and may be fragile, and contradict or ignore knowledge of human speech production and perception. We report progress towards phoneme recognition using a model of speech which employs very few parameters and which is more faithful to the dynamics and model of human speech production. Using features generated from a neural network bottleneck layer, we obtain recognition accuracy on TIMIT which compares favourably with traditional models of similar power. We discuss the implications of these results for recognition using natural features such as vocal tract resonances and spectral energies.