Research-Oriented Frameworks for Narrative Analysis

Abstract

This chapter focuses on explaining the significance of research-oriented analytic systems for disordered discourse and discussing various frameworks of assessments in the existing literature. First, the general principles of various measurements on lexical diversity of discourse samples (such as Type-Token Ratio [TTR], moving average TTR, number of different words, or textual lexical diversity) are explained. This is followed by a detailed description of the Quantitative Production Analysis (QPA), including its procedures of eliciting discourse samples, extracting propositional speech, segmenting extracted samples into utterances, and evaluating performances on the lexical content and sentence structure of a narrative sample. Other published procedures to quantify sentential characteristics in a discourse are then discussed, to measure important components of discourse grammar, or to estimate the degree of discourse cohesion and coherence of an oral discourse. Finally, the principles and procedures of natural language processing (NLP) are summarized. Illustrative examples of these analyses are given, with corresponding details explaining the potential clinical implications for each approach.