How Should a Speech Recognizer Work?

Although researchers studying human speech recognition (HSR) and automatic speech recognition (ASR) share a common interest in how information processing systems (human or machine) recognize spoken language, there is little communication between the two disciplines. We suggest that this lack of communication follows largely from the fact that research in these related fields has focused on the mechanics of how speech can be recognized. In Marr's (1982) terms, emphasis has been on the algorithmic and implementational levels rather than on the computational level. In this article, we provide a computational-level analysis of the task of speech recognition, which reveals the close parallels between research concerned with HSR and ASR. We illustrate this relation by presenting a new computational model of human spoken-word recognition, built using techniques from the field of ASR that, in contrast to current existing models of HSR, recognizes words from real speech input.     

How dogs navigate to catch frisbees

Using micro-video cameras attached to the heads of 2 dogs, we examined their optical behavior while catching Frisbees. Our findings reveal that dogs use the same viewer-based navigational heuristics previously found with baseball players (i.e., maintaining the target along a linear optical trajectory, LOT, with optical speed constancy). On trials in which the Frisbee dramatically changed direction, the dog maintained an LOT with speed constancy until it apparently could no longer do so and then simply established a new LOT and optical speed until interception. This work demonstrates the use of simple control mechanisms that utilize invariant geometric properties to accomplish interceptive tasks. It confirms a common interception strategy that extends both across species and to complex target trajectories.     

Children's competence for assent and consent: a review of empirical findings

This narrative review summarizes the empirical literature on children's competence for consent and assent in research and treatment settings. Studies varied widely regarding methodology, particularly in the areas of participant sampling, situational context studied (e.g., psychological versus medical settings), procedures used (e.g., lab-based vs. real-world approaches), and measurement of competence. This review also identifies several fundamental dilemmas underlying approaches to children's informed consent. These dilemmas, including autonomy versus best interest approaches, legal versus psychological or ethical approaches, child- versus family-based approaches, and approaches that emphasize consent versus those that emphasize assent, have implications for the measurement of children's competence and interpretation of findings. Recommendations for future research in the area of children's informed consent include the use of diverse samples and control groups, development of multidimensional and standardized measures of competence, utilization of multidimensional and standardized measures of competence, utilization of observational methods and longitudinal designs, examination of noncognitive aspects of children's competence and comparison of children's competence for treatment and research decisions.     

Effects of teacher-directed versus student-directed instruction on self-management of young children with disabilities

In this study, students worked independently by setting goals, selecting assignments, and recording and evaluating their results after receiving one of two different types of self-management training. During teacher-directed training, the teacher set goals, assigned work, and recorded and evaluated results for students. During student-directed training, students performed those tasks themselves. The results indicated that students engaged in the self-management behaviors more frequently during independent work following student-directed instruction than following teacher-directed instruction.     

An outcome management program for extending advances in choice research into choice opportunities for supported workers with severe multiple disabilities

We evaluated an outcome management program for increasing choice opportunities provided by 2 job coaches for 5 supported workers with severe multiple disabilities in a community job. The program involved specifying and monitoring behavioral outcomes among workers and staff, training staff, and supportive and corrective feedback. Increased choice provision occurred for both job coaches across a 1-year period. Results indicate how outcome management can help translate advances in choice research into routine practice.     

What you get out of memory depends on the question you ask

Following study, participants received 2 tests. The 1st was a recognition test; the 2nd was designed to tap recollection. The objective was to examine performance on Test 1 conditional on Test 2 performance. In Experiment 1, contrary to process dissociation assumptions, exclusion errors better predicted subsequent recollection than did inclusion errors. In Experiments 2 and 3, with alternate questions posed on Test 2, words having high estimates of recollection with one question had high estimates of familiarity with the other question. Results supported the following: (a) the 2-test procedure has considerable potential for elucidating the relationship between recollection and familiarity; (b) there is substantial evidence for dependency between such processes when estimates are obtained using the process dissociation and remember-know procedures; and (c) order of information access appears to depend on the question posed to the memory system.     

The Reduction of Adverse Impact in an Employment Setting Using a Short-Term Memory Test

Research suggests that less adverse impact should result from the use of employment tests assessing memory abilities than is obtained from tests assessing cognitive abilities. To test the application of this principle, a Short-Term Memory test and a Reading Comprehension test were administered as part of a selection battery for entry level police. Subjects were 1423 police officer applicants for a Midwestern city in the U.S. The results for the overall sample showed a smaller standardized difference between the scores of Blacks and Whites on the Short-Term Memory test than on the Reading Comprehension test, a difference which was further reduced when a subsample of Blacks and Whites was created which had been matched on gender, years of education and age.     

Internal consistency and reliability of alternate short forms of the college women's assertion sample: Clinical and research implications

Assertive behavior is most often assessed with self-report or role-play measures. The latter modality is preferred because it provides for the sampling of the structure of behavior and for the consideration of the situational context. MacDonald (1978) has developed such an assessment device but it is limited by the length of time for administration and scoring. Two studies were conducted to reconstruct reliable alternate short forms. The first study describes the selection of items and demonstrates the internal consistency of the alternate forms. The second study demonstrates the alternate form and retest reliability and provides normative statistics. We conclude that reliable alternate short forms have been constructed to be used in research in clinical applications.This research was supported by the Marie Wilson Howells Fund.Alternate short forms of the CWAS may be obtained from the first author.