Journal of Speech and Hearing Disorders, Volume 54, 482-499, November 1989 TOWARD PHONETIC INTELLIGIBILITY TESTING IN DYSARTHRIA RAY D. KENT GARY WEISMER JANE F. KENT University of Wisconsin-Madison JOHN C. ROSENBEK Williams S. Middleton Memorial Veterans Hospital, Madison, WI The measurement of intelligibility in dysarthric individuals is a major concern in clinical assessment and management and in research on dysarthria. The measurement objective is complicated by the fact that intelligibility is not an absolute quantity but rather a relative quantity that depends on variables such as test material, personnel, training, test procedures, and state of the speaker. This paper reviews sealing procedures and item identification tests as they have been applied to dysarthrie speech. Based in part on previous studies of speech of the hearing impaired, a profile has been designed to direct research on the acoustic or physiologic correlates of dysarthric intelligibility impairment. In addition, a word intelligibility test is proposed for use with dysarthric speakers. This test is designed to examine 19 acoustic-phonetic contrasts that are likely to (a) be sensitive to dysarthric impairment and (b) contribute significantly to speech intelligibility. Preliminary data from a sample of subjects with amyotrophic lateral sclerosis are presented to illustrate the use of this test in the phonetic interpretation of intelligibility impairment. KEY WORDS: intelligibility, dysarthria, articulation, speech The present paper treats the assessment of intelligibil- ity in dysarthric speakers. First, the concept of speaker intelligibility is reviewed, and it is concluded that an intelligibility score cannot be interpreted in a unitary way but must be evaluated relative to the specific conditions under which the intelligibility data were collected. The paper then describes previously published tests and/or materials that have been designed specifically for intelli- gibility testing in dysarthria. It is concluded that these tests and materials are not optimal for explaining intelli- gibility deficits because (a) the test materials (i.e., words and sentences) were not constructed to control several factors that contribute to variability in intelligibility scores and (b) the tests were not designed to provide much more than an index of severity. A remedy to some of these problems is proposed in the form of a word-level intelligibility test wherein specific phonetic contrasts are incorporated into minimal-pair sets. The phonetic con- trasts are selected to be sensitive to the articulatory problems experienced by dysarthric speakers and also have the advantage of well-defined correlates at the acoustic level of analysis. It is argued that the latter point is crucial because the test cannot only index the severity of the dysarthria but may also serve to identify the locus of the intelligibility deficit. In this sense the test may contribute to an explanation of intelligibility deficits. In the current state, the word test described in this paper is not advocated as a clinically proven procedure but rather as a research instrument. Our ultimate goal, however, is to develop procedures that are useful in the treatment of persons with neurogenic intelligibility deficits. Nor is it assumed that any single test can satisfy every purpose in intelligibility testing. Rather, the goal is to identify tests that can provide specific types of information on intelli- gibility impairment. Intelligibility is only one measure, although an impor- tant one, of a dysarthric individual's communicative abil- ity. A speech disorder such as dysarthria also may be defined in terms of "naturalness," "acceptability," or "bizarreness." However, the focus of this paper is on intelligibility, and we will not consider these other di- mensions further in this report. Intelligibility and many other dimensions of speech may be considered together under an even broader category that can be termed communicative competence or determinability. Connolly (1986) used the latter term to refer to the degree of communicative success a speaker can attain. He noted that lack of intelligibility is only one possible cause of a failure of communication. Even an intelligible speaker can fail in communication if other aspects of the message le.g., word choice, syntactic structure) are inappropriate; but intelligibility per se is a limiting factor on almost any Communicative exchange. The Concept of Speaker Intelligibility The conventional evaluation of a communication sys- tem involves measurement of the intelligibility of speech transmitted by the system. The intelligibility measure- ment typically is accomplished "'by counting the number of discrete speech units correctly recognized by a listen- er" (Flanagan, 1972, p. 311). Many intelligibility testing methods were developed by communication engineers whose desire it was to secure the best possible commu- nication between speakers and listeners in a specified environment. Specialists in hearing disorders also con- tributed to the development of intelligibility tests and often capitalized on the work done by communication engineers. A third specialty concerned with speech intel- © 1989, American Speech-Language-Hearing Association 482 0022-4677/89/5404-0482501.00/0 Downloaded From: http://jshd.pubs.asha.org/ on 01/16/2014 Terms of Use: http://asha.org/terms