The chief source of evidence in support of the view that messages can be discriminated on the basis of linguistic features is derived from studies which manipulated the phonetic similarity between two competing messages. For example, Treisman (1964a) reported that when the irrelevant message was presented in an unfamiliar foreign language, rejection of this message was significantly easier than when the irrelevant message was composed of gibberish phonetically similar to English. Moreover, Miller and Isard (1963) supported the concept of linguistic feature analysis by revealing that syntax functions as a discriminable feature in auding.

Morey (1969) suggested a two-process theory of tracking-first the target message must be identified, then it must be followed. This ability to follow the target message through time typically implies that the auder attends to the semantic features of the message. In other words, given the presentation of messages similar in acoustic and linguistic features, the auder may rely upon semantic meaning to track (Sticht, 1972). Experimental evidence has tended to verify this view; Miller and Isard (1963) and Rosenberg and Jarvella (1969) found that subjects selected and devoted their tracking time to semantically meaningful or well-integrated messages-rather than to less meaningful messages.

This feature division (acoustic, linguistic, and semantic) parallels, at least roughly, our distinction of processes in auding. That is, the tracking of acoustic features can be construed as a "listening" issue, because these features of the acoustic display are involved in any listening task. Linguistic features, on the other hand, can be considered as a "language" issue; such features are appropriate to the selection and sequencing of a system of conventionalized signs. Finally, the semantic features of tracking imply "conceptualizing," or the extraction of meaning from the speech message.

We consider that, ordinarily, in auding continuous prose such as a lecture or recorded or broadcast presentations, the processing of acoustic and linguistic information is done automatically by preattentive processes. Automaticity cannot extend, however, to conceptualizing (La Berge and Samuels, 1973). The latter demands focal attention, and is hence subject to the constraints imposed by the operating parameters of focal attention specified by Blumenthal (1972) and stated earlier. Chapter V, Hypothesis 3 considers languaging and conceptualizing processes as factors that limit the rate of auding. It should be kept in mind that rate of focal attending may limit the rate at which conceptualizations can be formed, and hence limit auding performance, as well as reading performance, the topic to be discussed next.