Computer analysis of talk-silence sequences: The FIASSCO system

Abstract

The study of simple talk and silence indices that characterize conversation is limited by the costly, "labor-intensive" character of data collection and analysis. In the face of results demonstrating the significance of these data in interpersonal judgments (Hayes & Meltzer, 1972; Lustig, Note I), more efficient collection, storage, and analysis methods are required. This report describes a hardware and software system, FIASSCO, that collects, stores, and analyzes two-person separate-channel audio-recorded conversations for various indices of talk and silence. Data output are both continuous and discrete measures in time sequence. Further, data on the validity and reliability of FIASSCO output are provided along with sample analyses of computer results.This report presents a computer-assisted method of coding and analyzing sound-silence patterns in dyadic relationships. Specifically. it details the procedures used at the University of Wisconsin's Center for Communication Research for automatically recording, digitizing, storing, and analyzing the sound-silence patterns that are found in dyadic communication. These procedures have been developed as an integrated research approach by utilizing the potential of the center's PDP·12 computer. A discussion of how the various programs analyze a sample dyad and a reliability test of these analyses are presented. THEORY AND RATIONALEIt is reasonable to ask if the intensive study of simple, naked behaviors such as talk and silence is likely to offer any practical insight into interpersonal communication situations. An extensive review of the literature on simple talk and silence indices by Lustig (Note 1) produces the following: Talkative persons are more productive (Norfleet, 1948), more task oriented

The validation of automatic methods for measuring conversational speech activity

“…It seems clear that a 300-msec sampling interval (Cappella & Streibel, 1979;Jaffe & Feldstein, 1970) provides insufficient resolution to make the distinctions perceived by our criterial scorer. Our 300-msec condition detected 20% too few speaker switches and produced larger mean durations of talkspurts and pauses as a consequence of missing brief events.…”

“…This rule was applied for each successive 300-msec segment, with no running averaging, erasing, or filling. This condition was included because similar approaches appear in the literature (Cappella & Streibel, 1979;Casotta et al, 1964).…”

“…An observer then reads the beginnings and ends of speech events from these records, which ideally include time reference marks. The automatic acoustic techniques (Brady, 1965;Cappella & Streibel, 1979;Cassotta, Feldstein, & Jaffe, 1964;Phillips, Treurniet, Tigges, & Lewis, Note 2) work with the electrical form of the speech signal, processing it in real time and making an on-off decision' at each successive sampling interval.…”

Automatic mediation in group problem solving

“…The experimental system, like the procedure for resolving conflicting requests to speak, permitted participants to control the sequence of experimental events. A discussion of how the experiment is administered and how the resulting data are checked for accuracy and aggregated by a protocol analysis program is presented.Recent reports by Brown (1979) and Cappella and Streibel (1979) have presen ted automated procedures for analyzing voice chronography data. Their reports emphasize the utility of such talk-silence data for the study of diadic and multiperson communication.…”

“…Recent reports by Brown (1979) and Cappella and Streibel (1979) have presen ted automated procedures for analyzing voice chronography data. Their reports emphasize the utility of such talk-silence data for the study of diadic and multiperson communication.…”