HISTORY PAGE INDEX

Codifying Connotative Meaning: The Technical Problem

For hundreds of years, lexicographers have systematically codified the denotative meanings of words. Collections of definitions evolved into dictionaries, thesauruses and related denotative language reference tools. However, connotative meaning has eluded such codification or "pinning down":

Denotation and connotation. . . represent two vocabularies using the same set of words. The denotative usage presents the thing in its essential and objective meaning. The connotative usage presents it enriched by associations and feelings which, though not susceptible of being pinned down, are nonetheless real. (Weaver, R. M. A Rhetoric and Composition Handbook. New York, NY: William Morrow & Company, 1974)

For example, a dictionary provides the following denotative meaning for the word pub:

"a building providing alcoholic drinks for consumption on the premises." (Oxford Dictionary)

However, the word pub simultaneously evokes a host of emotional connotations, such as merriment, pleasure, cheerfulness, perhaps some sadness, and so on. Similarly, words such as summer, love, and melody carry a variety of positive emotional connotative associations for most people, while words such as amputate, sleaze, and holocaust have decidedly negative emotional connotations. In all cases, the associated connotations for any given word or phrase are not systematically accessible using any existing language reference resource or tool.

[To Index]

The reason for the absence of codification of connotative meaning is that, while words readily evoke emotional connotations, the converse is not true: emotional connotations are not easily codified using words.

Unlike denotative meaning, connotative or affective meaning does not naturally lend itself to systematic word-symbol codification. Emotions are felt, not thought. Lexicographers cannot employ their usual methods to codify the relationship between a word and its associated connotative content.

People have known for thousands of years that words convey connotative meaning. The problem of "pinning down" the emotional or connotative meanings of words is technical in nature, not conceptual.

The development of emotional or connotative language tools had to await two main technical developments that did not come along until the 20^th Century:

1.  The development of accurate psychological
     measurement techniques, the science of which
     began to mature only a few decades ago, led
     by psychologists such as Likert, Thurstone,
     Osgood, and others.

2.  The development of computers, especially the kind
     of high capacity, very fast personal computers that
     only became available in the 1990s.

[To Index]

Towards a Solution: Attitude Measurement

People adopt attitudes, hold opinions and express emotions with varying levels of intensity. To study "attitude" and measure its intensity, researchers typically use rating scales. Statistical techniques founded upon mathematical proofs such as the Central Limit Theorem are used to analyze numerically-scaled data classified into various "levels of measurement." (Four levels of measurement corresponding to increasing levels of precision are typically identified in the literature: nominal, ordinal, interval, and ratio.) Statistics used in data analysis may include factor analysis, sample variance, standard error of the mean, and multinomial probability distribution.

In the 1920s and 1930s, psychologists L. L. Thurstone and Rensis Likert pioneered attitude-measurement scaling methodologies. Thurstone’s contribution was a type of "differentiated" scale, a scale that would determine a person’s attitude (towards an object or concept) along a continuum with acceptable accuracy.

Likert, a psychologist investigating corporate management styles, devised a "summated" scale, a type of attitude scale that allowed for the summation and averaging of scaled responses. An example of a Likert scale is this familiar sequence:

strongly agree

somewhat agree

neither agree nor disagree

somewhat disagree

strongly disagree

There are many variations on this theme, but in general, Likert’s method involved attaching numbers to levels of meaning. The Likert-type scale proved to be very robust, and continues to be widely used today.

[To Index]

Linking Words with Attitude: Osgood's "Semantic Space"

The scale types devised by Thurstone, Likert and others did not explicitly connect scaled measurement with the connotative meanings of words. Charles Osgood, an American psychologist, is credited with this breakthrough. In the early 1950s, Osgood and colleagues used Roget's Thesaurus to help construct bipolar scales based on semantic opposites, such as "good-bad", "soft-hard", "fast-slow," "clean-dirty," "valuable-worthless," "fair-unfair," and so on. Osgood called these scales "semantic differential" scales because they differentiated attitudinal intensity based on a person’s subjective understanding of the connotative meanings of words.

Osgood et al were able to use computer technology (newly available at the time), and a statistical technique called factor analysis, to explore large amounts of data provided by numerous "judges" (i.e., ordinary people recruited to express their attitudes), who evaluated numerous words and phrases on numerous semantic differential scales. The outcome was Osgood’s discovery of "semantic space"—the existence of three measurable underlying attitudinal dimensions that everyone uses to evaluate everything in their social environment, regardless of language or culture. These three dimensions are Evaluation, Power, and Activity, known as EPA. Subsequent experimentation by many investigators around the world confirmed the reality of semantic space and its cross-cultural validity (Japan, Scandinavia, Germany, Ireland etc.)

Osgood’s team had, in effect, discovered a methodology for quantifying connotative semantic meaning. In 1957, Osgood and colleagues published their findings in a landmark volume, The Measurement of Meaning. (Osgood, Suci, and Tannenbaum, 1957)

[To Index]

A vast psychological literature on semantic differential and semantic space has developed over more than 50 years. Along with Likert scaling, Osgood’s semantic differential has been in continuous use since the method was first published, and is considered a mainstay of measurement of connotative meaning in psychology and other social sciences.

The Evaluation, Potency, and Activity (EPA) structure in subjective responses is one of the best documented facts in social science, and an elaborate technology has developed for measuring EPA responses on "semantic differential scales." (Heise, D. R. "Affect Control Theory and Impression Formation" in Borgatta, E. and Borgatta, M. (Eds.). Encyclopedia of Sociology, Vol. 1. New York, NY: Macmillan, 1992)

Here is a brief overview of semantic differential technique:

The semantic differential...is a method for measuring the meaning of an object to an individual. It may also be thought of as a series of attitude scales. The subject is asked to rate a given concept (for example, Irish, Republican, wife, me as I am) on a series of seven-point bipolar rating scales. Any concept—whether it is a political issue, a person, an institution, a work of art—can be rated . . . Subgroups of the scales can be summed up to yield scores that are interpreted as indicating the individual’s position on three underlying dimensions of attitude toward the object being rated. These dimensions have been identified by using factor-analytic procedures (factor analysis is a [statistical] method of finding the common element or elements that underlie a set of measures) in examining the responses of many individuals concerning many concepts or objects. It has been found that . . . three subgroups measure the following three dimensions of attitude: (1) the individual’s evaluation of the object or concept being rated, corresponding to the favorable-unfavorable dimension in more traditional attitude scales; (2) the individual’s perception of the potency or power of the object or concept; and (3) the individual’s perception of the activity of the object or concept. (Kidder, L. M. Research Methods in Social Relations. New York: Holt, Rinehart & Winston, 1981)

[To Index]

Of the three dimensions of semantic space, Evaluation proved to be the most important. Evaluation is also known as the "connotative" or "affective" dimension.

Of these dimensions, the one most heavily weighted in people’s judgments is evaluation. Osgood has recommended using it as the prime indicator of attitude toward the object. Clearly it is an affective dimension. (Oskamp, S. Attitudes and Opinions. Englewood Cliffs, NJ: Prentice-Hall, 1977)

"Affective" or "affect" is the term psychologists use when referring to emotion, or more specifically, the emotion associated with an idea or set of ideas.

The problem with the semantic differential technique is that it does not distinguish beyond a single evaluative continuum, with positive attitude at one end of the scale through to negative attitude at the other end. That is, it does not actually identify any individual emotions.

[To Index]

Evolution of Connotative Intelligence™ Technology

In the late 1970s, Wayne Chase, a graduate in English and psychology with training and job experience in statistics, research design, and computing, became intrigued with Osgood’s findings and various applications of the semantic differential. Reviewing the literature, he found that several psychologists had compiled short "semantic differential dictionaries" of 500 to 1,500 words with EPA scores for each word (known in the literature as semantic "atlases" because they are analogous to "maps" of semantic space). Chase reasoned that, since these semantic atlases are actually miniature connotative dictionaries, it should be possible to develop a similar connotative dictionary for the entire English language. This would be the connotative equivalent of the familiar "denotative dictionary," the ordinary dictionary that all of us are familiar with. A connotative dictionary would be of tremendous interest to creative writers, students and others who regularly consult dictionaries and thesauruses.

In the early 1980s, Chase began to organize a primitive connotative database on paper. However, without access to a computer, the work was extremely slow and tedious. Mainframe computers were only available to those affiliated with corporations, government, or academic institutions. Early personal computers were prohibitively expensive, and software for them was crude and complicated.

In 1987, Chase acquired an IBM PC clone and began to computerize the fledgling database using the DOS application dBASE. His plan was to eventually export the connotative database to a desktop publishing program and create a connotative dictionary in book form. This would be the first full-language Connotative Intelligence-based reference work in any language.

[To Index]

As the database grew, Chase began to get a better handle on the scope of the work required to complete the connotative dictionary. Software would have certain advantages over print, although both would be feasible. Software could scan and analyze the connotative content of an entire piece of written work. At first, Chase called this application the "Emotional Impact Analyst." By the late 1990s, it became the "Emotional Meaning and Impact" analyst, or "EMMI." This "EMMI" application would be the connotative equivalent of the denotative "grammar checker" (which analyzes the grammatical content of an entire written piece). Analyzing connotative content might, for the first time, provide writers with a method of predicting the subjective or emotional impact of a piece of writing on an audience. A writer would be able to tailor the emotional message inherent in his or her writing before presenting it to an audience. The audience could be anyone, from readers of novels to viewers of television drama to consumers of advertising messages.

Chase continued working on the connotative database through the 1990s, even though mainstream personal computers of the day (up to about 1996) would have been unable to handle fully-developed EMMI software, had it existed at the time. Hard drives were too small and processors too slow. Observing the rapid development of computer hardware, Chase was confident that it would not be long before personal computers affordable to everyone would have sufficient disk capacity and speed to handle EMMI.

In April, 1998, Chase sought help at the University of British Columbia and Simon Fraser University. Academic staff and students in English, psychology, and linguistics joined the EMMI project and made it possible to prove the concept on a substantial scale through a multi-phase pilot study.