ABSTRACT

All pairs of names generated by the individual names of nine historically important psychologists were submitted as queries to the Google search engine. The resulting page counts were used to generate similarity/dissimilarity indices that were submitted to both cluster analysis and multidimensional scaling. Both of the analyses separated the names into three distinct clusters that were easily associated with three historically important schools of psychology. The purpose
of the study was to examine the idea that the world-wide-web contains latent structures of the sort made familiar awhile ago by Charles Osgood. Earlier related data, gathered by the author in the last three or four years, is summarized and presented as further evidence that Osgood meaning may be latent in the world-wide-web. Questions regarding appropriate indices of similarity/dissimilarity and problems of the reliability and validity of these procedures and their results are discussed. Evidence is presented for all of these qualities in the results of this study.
Finally, it is demonstrated that at least one of Osgood’s connotative Semantic Differential factors is hidden in the structure of the world-wide-web.

INTRODUCTION
With the amount of discussion that has been generated lately about Google and the, so called, semantic web or Semweb (for example, Ford, 2002a, b), I am surprised that some psychologist has not noticed and remarked publicly that the coarse structure of the world-wide-web may be hiding semantic structures in which psychologists have, in the past, shown great interest. Here I am using the term "semantic structure" in an old fashioned sense that, I think, would warm Charles Osgood’s heart (Osgood et al., pp. 25-124 and elsewhere). Events like the following sometimes occur during a Google search: (a) The search term, {+Freud +Jung}, finds 6760 documents. (b)The search term, {+Freud
+Rembrandt}, co-occurs in only 645 documents. [1] Having noticed several of these occurrences, I am further surprised that our hypothetical psychologist wasn’t moved to infer that those numbers might index the similarity of meaning of the co-occurring concepts (again using Osgood’s sense of meaning).
At least a couple of computer scientists (Cilibrasi and Vitynai, 2005) have made this inference and I suggest that much of their paper is likely to be both accessible and interesting to psychologists.
Osgood is remembered for, at last, three significant contributions to the behavioral/social sciences. He developed a mediational theory of meaning based on conditioning (Osgood, Suci & Tannenbaum, 1957, pp. 5-9). He developed a construct, the "semantic space," where the meaning of a concept is represented as a vector in a space spanned by an unknown (but discoverable) number of dimensions of meaning (Osgood, Suci & Tannenbaum, 1957, p. 25). And he developed a method, the Semantic Differential, for discovering the location, or meaning, of the concepts within the semantic space (Osgood, Suci & Tannenbaum, pp. 18-30, and elsewhere). The mediational theory is rarely mentioned now. However, the spatial model and the measurement method are alive in several areas of behavioral/social science research. I did a Google search for the term +"Semantic Differential" and found 149,000 references. The Google Scholar service (Google website, n.d) classified 920 of these as scholarly works published between 2004 and 2006.

continua >>>>>