Collaborative Commentary: Opening Up Spoken Language Databases

We define collaborative commentary as the involvement of a research community in the interpretive annotation of electronic records. The goal of this process is the evaluation of competing theoretical claims. The process requires commentators to link their comments and related evidentiary materials to specific segments of either transcripts or electronic media. Here, we examine current work in the construction of technical methods for facilitating collaborative commentary through browser technology. To illustrate the relevance of this approach, we examine seven spoken language database projects that have reached a level of web-based publication that makes them good candidates as targets of collaborative commentary technology. For each database, we show how collaborative commentary can advance the relevant research agendas. Collaborative Commentary We define collaborative commentary as the involvement of a research community in the interpretive annotation of electronic records. The goal of this process is the evaluation of competing theoretical claims. The process requires commentators to link their comments and related evidentiary materials to specific segments of either transcripts or electronic media. Consider the following real-life example. Brian MacWhinney, a researcher in the field of child language, wants to explore evidence for the neo-Vygotskyan claim (Nelson, 1998) that word meanings are shaped through communicative interactions. While browsing through online media at the CHILDES (childes.psy.cmu.edu) site, he locates several instances of videos of mother-child book reading in the Julie and Rollins corpora. In these interactions, mothers help children turn the pages and name the animals or objects in the pictures. In some cases, children call the pictures by the wrong name. In accord with his theoretical views on the logical problem of language acquisition, MacWhinney (in press) believes that mothers will use these errors as opportunities to provide corrective positive feedback. For example, if the child calls a bear a “doggie,” the mother should respond, “no, that’s a bear, not a doggie.” MacWhinney would be particularly happy if the child subsequently engaged in self-correction and said “a bear.” Pursuing this idea, he locates 26 segments in these corpora that are relevant to the position he is advocating. To initiate the process of collaborative commentary, MacWhinney writes up a short summary of his analysis. He wants to make this analysis available in three ways. First, he wants to post his claim to some discipline-based commentary space on the web. Second, he wants to make sure others who view the relevant segments from the Julie and Rollins corpora are able to see that he has provided detailed interpretive commentary regarding at least 26 specific segments. Finally, he hopes to receive feedback from other researchers regarding his interpretations and arguments.