An enriched set of features of nuclear cataract identified by multidimensional scaling.

Development of an improved system for visual classification of cataracts requires a three-step procedure: first, to identify the full range of visible features of cataracts; second, to develop and test scales for the visual assessment of each feature; and third, to establish the epidemiological or clinical validity of each scale for cataract classification. This paper focuses on the first step, applying a powerful psychometric technique for identifying the visible features of nuclear cataracts. New visual features of nuclear cataract were identified using the psychometric procedure of multidimensional scaling (MDS). Each of 5 observers independently examined pairings of slitlamp photographs of 24 cases of pure nuclear cataract, making two different ratings of dissimilarity of each of the 276 possible pairs. The two dissimilarity ratings were, first, of nuclear color and, second, of nuclear structure. MDS analysis of the dissimilarity ratings of nuclear color revealed two major visual features underlying the judgments: one a combination of hue and saturation, and the other brightness. Analysis of the ratings of nuclear structure identified a total of nine features: one distinguishing between immature and mature cataracts, four describing features of the immature cataracts (aspect ratio, background haze, clarity of the embryonal nucleus, and clarity of the outer nuclear shell), and four describing features of the mature cataracts (opalescence, aspect ratio, color of the nucleus, and symmetry). We conclude that there are many more systematic distinctions to be made in the appearance of nuclear cataracts than are now recognized in clinical practice.

Speech quality evaluation using "phoneme-specific" sentences.

A new approach is described for the design of speech materials used in subjective speech quality evaluation. Speech sounds are classified by their acoustic properties, and sentences are composed so as to concentrate all sounds with similar properties within one sentence. As a test of the method, subjective quality data were collected, using both a rank ordering and a rating task, from a set of 12 linear predictive vocoders, whose parameters were chosen so as to equate their bit rates at 2600 bps. The results show that the method can reliably reveal small differences in quality, and also yields information that can be of diagnostic help in determining the causes of quality degradation by a particular vocoder. A set of phoneme-specific sentences is appended.

Is periodicity detection central?

What periodicity is perceived when two different segments of noise, E and F, are alternated dichotically between the ears (Left: EFEFE..; Right FEFEF..)? Earlier work has suggested that for nonfusible signals, preliminary auditory processing is performed separately for the two ears before the results are combined. If so, the perceived period should correspond to the combined lengths of the two segments, since this is the signal that iterates in one ear EF. But if periodicity is detected centrally, the perceived period should be equal to the length of one of the alternated segments (E, F), since each segment presented to one ear is immediately repeated in the other. Subjects compared an alternated sequence with two dichotic iterated sequences. (Left: CCC...; Right: DDD...) in an AXB paradigm, and segment duration was varied between 25 and 400 ms. In one comparison sequence, the iterated segments were equal in duration to the alternated segments (C = D = E = F), and in the other they were twice as long (C = D = E + F). The results show that periodicity is detected separately for the two ears at the shorter durations (nonfusible signals), but becomes central when the segments become long enough for structure to be heard within each segment (i.e, longer than 140-200 ms).

Better spectrograms from children's speech. A research note.

A major problem in the spectrographic analysis of children's speech is the poor resolution of formants, which is the result of the widely spaced harmonics of the high fundamental frequency. An attempt has been made to bypass this problem by exciting a child's vocal tract with an artificial larynx, using a fundamental frequency appropriate to a man. This method has promise for tracking formants in children's speech.