Comparing Hall Van de Castle Coding and Linguistic Inquiry and Word Count Using Canonical Correlation Analysis

For dream content analysis, automatic quantitative analysis techniques cannot only be faster than traditional hand-coding but also be lower in coding errors and bias caused by humans. Linguistic Inquiry and Word Count (LIWC; Pennebaker et al., 2015) is an automatic technique possibly useful for dream research. To consider suitability of information produced by LIWC for dream studies, this article compares output of the Hall Van de Castle coding system (Hall & Van de Castle, 1966), the traditional and most often used hand-coding system in dream research, and LIWC, with canonical correlation analysis (CCA) using a classic set of dream reports collected by Hall and Van de Castle (1966). Results show good compatibility between outputs of the 2 tools, which indicates good potential and usability of LIWC for dream content analysis. CCA is a multivariate statistical method for measuring the association between 2 sets of variables. Despite its complexity, CCA is able to reveal rich information about relationships among multivariate variables. It has several advantages over univariate methods and can be useful for dream research. This article gives an intuitive and also a more formal introduction of CCA for dream researchers.