General tendencies and attitudes.
Historically, the majority of social scientists appear to utilize a variety of quantitative measures and methods to analyze verbal information. However, a coexistent and growing number of social scientists have also selected from a similarly diverse group of qualitative methods for the analysis of verbal information.
As we point out in the first step of conducting a Quantitative Phenomenology with Raven’s Eye, the selection of a particular method is determined by such things as the theoretical orientation of the researcher, the type and function of the question being asked, and the apparent structure and function of the constructs involved in the research area. Typically, when a research question translates verbal information into numerical information prior to analysis, or asks participants to select a response from a list of word- or phrase-based responses, quantitative approaches are being conducted. When, instead the research question (which is typically open-ended) focuses on the meaning or relations between the concepts being expressed in the verbal response it evokes, social scientists have historically used qualitative research methods.
Increasingly, what has become known as mixed-methods approaches are appearing in the social science literature. In these, researchers typically acquire a variety of verbal information. Those verbal responses historically amenable to quantitative methods (e.g., choosing from a pre-existing list of words, or rank-ordering a pre-existing set of responses) are analyzed statistically, while those responses not readily amenable to numerical comparison (e.g., freely formed natural language statements) are analyzed through a particular qualitative method. Whether mixed or single methods are being employed, therefore, in contemporary social science the qualities of the verbal information acquired tend to predict the type of analysis that will be performed.
Though a few limited quantitative approaches appear in the literature sporadically throughout the 20th Century, by far the most dominant social science attitude toward verbal information in the form of natural language has been to analyze it with qualitative research methods. Indeed, even a cursory review of the relevant social science literature will readily reveal the highly dependent relationship between the analysis of natural language expression, and qualitative data analysis. In other words, if researchers are analyzing natural language in the social sciences, they are quite likely to call what they are doing, “qualitative data analysis,” and if they utilize software to assist them, such software is commonly known as, “qualitative data analysis software,” or, “computer-assisted qualitative data analysis software,” (CAQDAS).
Inspection of the progress in natural language processing and natural language understanding in such fields as cognitive science, information science, computational linguistics, and artificial intelligence, however, reveals alternative and quantitative approaches that proceed from a very different attitude toward natural language data. Such an attitude tends to conceptualize the words that comprise natural language statements as bits of data which, when aggregated and related sufficiently, can be analyzed in an objective and replicable manner through mathematically-based techniques. Because of the focus on sufficiently large samples, researchers with such an attitude often utilize what has become known as Big Data sets, and employ advanced and often specifically crafted software programs to assist them in their analyses. Depending on the particular focus of research, such software programs are generally classified as natural language processing programs, and include such types as sentiment analysis tools, and opinion mining software.
In designing Raven’s Eye, we assumed an expansive attitude, which integrates and subsumes the heretofore disparate perspectives toward natural language held by the social sciences in the one hand, and the cognitive and information sciences in the other. Our expansive attitude acknowledges the validity of both qualitative and quantitative means of understanding natural language in each their own way. As a result, we have developed a truly hybrid method, Quantitative Phenomenology, which combines the advantages of both approaches. Likewise, our software represents the synthesis of qualitative data analysis programs and Big Data natural language processing tools into a product which both facilitates many of the existing functions of these separate classes of software, while also advancing upon them in new, meaningful, and productive ways.