The Measurement of Meaning

by Charles E. Osgood, George J Suci, Percy Tannenbaum

 In this pioneering study, the authors
      deal with the nature and theory of meaning and present a new, objective
      method for its measurement which they call the semantic differential.
      This instrument is not a specific test, but rather a general technique of
      measurement that can be adapted to a wide variety of problems in such areas
      as clinical psychology, social psychology, linguistics, mass communications,
      esthetics, and political science. The core of the book is the authors' description,
      application, and evaluation of this important tool and its far-reaching
      implications for empirical research.

Genres: Language

354 pages, Paperback

First published September 1, 1967

Reviews

[Rahim Hashim · Rating 5 out of 5]

I was prompted to Charles E. Osgood's foundational work from James Somers Dec 2022 New Yorker piece, "The Science of Mind Reading," and needless to say, I was not disappointed.

Osgood was a pioneer in cognitive science and linguistics in the 1950-80s, conducting experiments which asked a very core question of human language: how are concepts are related to one another, and how does that drive the way we think? His experiments involved asking college students to rate tens of words on fifty different scales (i.e., rate the word RIVER on a scale of 1-7, where 1 is cold, and 7 is hot). He then performed dimensionality reduction techniques to quantify how closely and far apart words were from each other in "semantic space," observing three axes that captured most information about the data: 1) evaluative 2) potency and 3) activity. What it meant was that for each word, not just included in his dataset, but in all of the English language, we can generate a simple x,y,z coordinate system that allows us to map and visualize how closely and/or far apart it is from any other word.

The experiments captured in the book, as well as Osgood's interpretations of the results, were incredibly clear and well-structured, reminding me that there was in fact a time in which scientists wrote specifically with the goal in mind to have a broader audience understand the work.As a non-expert in statistics and linguistics, I felt comfortable in the math and analyses techniques introduced, and learned a ton about a field I had little-to-no experience in.

The results of Osgood's work have never been more applicable, with large language models and text to image algorithms compressing multi-billion parameters of human language and image data omnipresent, and only growing in importance as they become a part of our everyday routines. I highly recommend reading the book, as well as James' article and other commentary (see: Osgood, Psychological Bulletin, 1952; Caroll, Language, 1949).

[Nan Diliyanova · Rating 4 out of 5]

Read this for my PhD. It's curious how some experimental designs and analyses are still up to date so many decades later, while others are outdated as duck.

Nice one.