Semantic Habits Measure

1. Abstract

The Semantic Habits Measure is a specialized psychometric instrument developed by Jum C. Nunnally, Ronald L. Flaugher, and William F. Hodges in 1963 at Vanderbilt University. Designed to evaluate individual differences in the preferred modes through which individuals establish verbal and conceptual associations, the instrument operationalizes semantic processing as enduring cognitive tendencies rather than mere linguistic ability. The tool comprises 143 forced-choice, binary-association items that systematically pit four distinct modes of semantic reaction against one another: positive evaluative associations (E-plus), negative evaluative associations (E-minus), sensuous denotative attributes (D), and taxonomic categorization (C). These response tendencies are combined across five sub-pairings to yield three primary composite scales: the E-plus Scale (derived from E-plus vs. D and E-plus vs. C pairings), the E-minus Scale (derived from E-minus vs. D and E-minus vs. C pairings), and the C-D Balance Scale (contrasting taxonomic categorization against sensory-denotative descriptions). Psychometric evaluations conducted with undergraduate cohorts have demonstrated strong internal consistency, with each of the three major scales exhibiting reliability coefficients hovering around .80. By constraining associative responses to controlled pairwise selections, the instrument bypasses traditional free-association coding ambiguities and provides an empirical method to map how human beings habitually filter their semantic worlds through evaluative, descriptive, or structural cognitive prisms.

2. Keywords

Semantic Habits Measure, Jum C. Nunnally, psychometrics, word association, semantic relations, cognitive style, evaluative tendency, categorization, denotative attributes, forced-choice format, verbal mediation, internal consistency.

3. Authors

The Semantic Habits Measure was conceptualized and developed by a research team in the Department of Psychology at Vanderbilt University:

• Jum C. Nunnally, Ph.D. (1927–1982): A foundational figure in modern psychometrics, Professor and Chair of the Department of Psychology at Vanderbilt University. Nunnally is globally recognized for his seminal textbook, Psychometric Theory (first published in 1967), which established standard protocols for reliability estimation, scale construction, and factor analysis.
• Ronald L. Flaugher, Ph.D.: Research psychologist associated with Vanderbilt University and subsequently a prominent psychometrician at the Educational Testing Service (ETS), focusing on test construction, cognitive fairness, and predictive validity.
• William F. Hodges, Ph.D.: Clinical and research psychologist at Vanderbilt University, known for his work in experimental clinical psychology, state-trait anxiety measurement, and personality assessment.

4. Purpose

The primary purpose of the Semantic Habits Measure is to evaluate idiosyncratic, stable tendencies in how individuals connect concepts, process lexical stimuli, and organize their semantic representations. Historically, psychological research into verbal habits leaned heavily on unstructured free-association techniques, such as the Kent-Rosanoff Word Association Test. While free association provides rich qualitative data, it suffers from severe psychometric limitations, including non-standardized response distributions, labor-intensive scoring, and confounding by response speed, lexical fluency, and idiosyncratic vocabulary levels.

To overcome these methodological shortcomings, Nunnally and colleagues designed an objective, binary-choice association task that directly isolates the qualitative nature of conceptual links. When encountering a stimulus word (e.g., an object, an event, or an abstract entity), a person can theoretically relate to it across numerous semantic dimensions. For instance, an apple can be conceptualized in terms of an affective judgment (e.g., “delicious”), a concrete perceptual property (e.g., “red”), or its overarching hierarchical category (e.g., “fruit”). The Semantic Habits Measure was created to determine whether individuals demonstrate pervasive, generalized biases toward specific semantic modalities across varying contexts.

In research contexts, the instrument serves as a critical bridge between cognitive psychology, linguistics, and personality theory. It enables researchers to investigate whether cognitive styles—such as an inclination to categorize versus an inclination to focus on immediate perceptual qualities—correlate with broader intellectual dispositions, cognitive complexity, or aesthetic appreciation. In clinical and personality research, quantifying an individual’s chronic tendency toward positive or negative evaluations (E-plus vs. E-minus) offers a non-obtrusive, performance-based index of affective bias, depression vulnerability, or general evaluative mindset that minimizes the social desirability artifacts common in self-report inventories.

5. Psychological Construct

The instrument operationalizes the psychological construct of semantic habits—defined as persistent, individualized response hierarchies that govern lexical and conceptual retrieval during verbal mediation. Rather than assessing semantic knowledge or verbal intelligence (the capacity to understand meanings), the measure captures stylistic preference: when multiple meaningful relations are concurrently available, which associative pathway does an individual spontaneously follow?

The measure isolates four fundamental semantic modes, organized into structured contrasts:

• Positive Evaluative Tendency (E-plus): This mode reflects a habitual disposition to connect concepts via favorable emotional, moral, or aesthetic valence. When presented with a neutral or complex stimulus, an individual with high E-plus tendencies immediately privileges its beneficial, pleasant, or laudable characteristics. For example, when encountering the word “summer”, an E-plus response selects “glorious” over a purely descriptive or taxonomic alternative.
• Negative Evaluative Tendency (E-minus): This mode represents the inverse affective disposition—a chronic cognitive inclination to link stimuli with unfavorable, hazardous, unpleasant, or critical attributes. For instance, in response to the stimulus “winter”, an E-minus oriented respondent prefers the association “bleak” rather than a neutral categorization such as “season” or a sensory property like “cold”.
• Sensuous Denotative Tendency (D): This mode reflects a perceptual, sensorimotor cognitive style. Individuals displaying high D tendencies consistently associate concepts with concrete, perceptual, and descriptive qualities discernible by the physical senses (e.g., texture, color, shape, temperature, sound). Given the stimulus “stone”, a D-habit individual prioritizes “rough” or “heavy” rather than a formal classification (e.g., “mineral”) or an affective appraisal (e.g., “burdensome”).
• Taxonomic Categorization Tendency (C): This mode captures an analytical, logical, and hierarchical cognitive strategy. It is marked by the spontaneous categorization of stimuli into formal superordinate classes, family groupings, or abstract relational structures. For the stimulus “robin”, a C-oriented respondent chooses “bird”; for “triangle”, they select “shape”, ignoring both aesthetic descriptions and sensory impressions.

By contrasting these four modes in controlled pairs, the instrument constructs three overarching scores: (1) the E-plus Scale, aggregating preferences for positive evaluations against denotative and categorical options; (2) the E-minus Scale, aggregating preferences for negative evaluations against non-evaluative options; and (3) the C-D Balance Scale, which directly contrasts structural categorization (C) with immediate sensory attributes (D). On the C-D Balance Scale, higher scores represent a preference for abstract taxonomic thinking, whereas lower scores reflect a perceptual, sensory-oriented cognitive style.

6. Theoretical Framework

The Semantic Habits Measure is rooted in the intersection of Charles E. Osgood’s semantic differential theory and neo-behaviorist verbal mediation paradigms popular in the mid-20th century. Osgood, Suci, and Tannenbaum (1957) demonstrated through extensive factor analytic work that the human semantic space is predominantly organized along three primary dimensions: Evaluation (good-bad), Potency (strong-weak), and Activity (active-passive), with the Evaluative dimension consistently accounting for the largest share of variance.

Nunnally and his colleagues integrated Osgood’s dimensional model with the psychological tradition of associationism. They posited that when an external linguistic sign is decoded, it activates an internal representational mediation process ($r_m$). While semantic differential rating scales force respondents to locate a single concept along predetermined bipolar rating continua, natural cognitive processing involves an internal competition among disparate mediating responses. An individual presented with a stimulus word does not merely judge its position on a rating scale; they experience competing response inclinations. If internal mediation habits differ systematically across individuals, these differences should manifest as consistent choices when individuals are forced to select between two valid, competing semantic associations.

Furthermore, the scale incorporates developmental and cognitive style theories, specifically those articulated by Jean Piaget and Jerome Bruner regarding the transition from concrete perceptual processing to formal categorical abstraction. In cognitive developmental theory, young children initially classify objects based on physical, perceptual, and sensuous attributes (iconic representation), only later developing the capacity for hierarchical, formal categorization (symbolic representation). Nunnally, Flaugher, and Hodges theorized that even among mature, intelligent adults, residual individual differences persist in the degree to which individuals utilize perceptual-denotative (D) versus abstract-categorical (C) processing channels. By combining Osgood’s affective evaluation framework with Bruner’s structural representation levels, the authors constructed a unified theoretical system for quantifying stylistic verbal preferences.

7. Validity

The initial validation of the Semantic Habits Measure focused primarily on content validity, construct coherence, and experimental convergent-discriminant validation across undergraduate cohorts at Vanderbilt University:

• Content and Face Validity: The authors established content validity through rigorous item-selection procedures. Stimulus words and paired alternative associations were developed and refined to ensure that each alternative cleanly exemplified only one intended semantic mode (E-plus, E-minus, D, or C). Items were filtered to eliminate ambiguous alternatives that could be interpreted as simultaneously evaluative and categorical, ensuring high construct purity across the binary contrasts.
• Construct and Convergent Validity: In early experimental investigations, the E-plus and E-minus scales correlated meaningfully with personality dimensions tapping affective dispositions. Tendencies toward E-minus associations were found to relate positively to measures of neuroticism, social anxiety, and depressive symptomatology, demonstrating that an elevated E-minus habit reflects an underlying cognitive vulnerability or negative cognitive schema. Conversely, the C-D Balance Scale demonstrated convergent correlations with cognitive style measures, such as field independence-dependence and abstract reasoning tasks. Respondents exhibiting high C scores consistently favored analytical and hierarchical problem-solving strategies over sensory-bound reasoning.
• Discriminant Validity: The authors demonstrated that individual differences on the Semantic Habits Measure were largely distinct from general verbal intelligence. Performance on the 143 items was not driven by vocabulary knowledge, because the words utilized across all items were intentionally selected from high-frequency lexical bands. Individuals with identical scores on standardized verbal aptitude tests differed substantially in their placement on the E-plus, E-minus, and C-D scales, establishing that the measure assesses cognitive preference rather than intellectual capacity.

8. Reliability

Psychometric evaluations reported by Nunnally, Flaugher, and Hodges (1963) established that despite the forced-choice, binary nature of individual items, the aggregated composite scales exhibit strong internal consistency:

• Internal Consistency: Reliability analyses conducted on samples of undergraduate students yielded internal consistency estimates (calculated via split-half methods corrected by the Spearman-Brown prophecy formula and Kuder-Richardson formulas) of approximately .80 across the three major scales: the E-plus Scale, the E-minus Scale, and the C-D Balance Scale. These coefficients are robust for non-cognitive stylistic measures and meet standard psychometric thresholds for basic research applications.
• Item-Total Homogeneity: Item-analysis procedures confirmed that items within each functional sub-pairing (e.g., E-plus vs. D, E-minus vs. C) exhibited positive and statistically significant point-biserial correlations with their respective composite scale totals, verifying that individual binary items effectively contribute to a common latent trait.
• Test-Retest Stability: Subsequent laboratory investigations into cognitive styles using the Nunnally paradigm reported short-term test-retest correlations ranging between .72 and .81 over intervals of two to four weeks, indicating that semantic habits represent relatively enduring personal dispositions rather than transient, state-dependent verbal fluctuations.

9. Factor Analysis

In the original 1963 monograph, classical exploratory factor analysis (EFA) was not directly computed on the raw 143 binary items, primarily due to the computational constraints of the early 1960s and the statistical challenges inherent to factor-analyzing ipsative or forced-choice item matrices. When binary-choice items directly pit two constructs against one another within the same item stem, negative dependency artifacts (ipsativity) can distort classical Pearson product-moment correlation matrices, artificially producing negative eigenvalues or spurious bipolar factors.

Instead, the structural integrity of the instrument was verified through structural contrast matrices and subscale correlation analyses:

• Subscale Structural Intercorrelations: Nunnally and colleagues examined the correlational matrix among the five basic sub-pairings (E-plus vs. D, E-plus vs. C, E-minus vs. D, E-minus vs. C, and C vs. D). The empirical correlations confirmed the convergent validity of combining the evaluative subscales: the E-plus vs. D subscale correlated highly with the E-plus vs. C subscale, justifying their summation into a single comprehensive E-plus Scale. Similarly, the two negative evaluative sub-pairings demonstrated convergent alignment into the unified E-minus Scale.
• Independence of Affective and Non-Affective Dimensions: Structural analyses revealed that the C-D Balance Scale functioned largely orthogonally to the two evaluative scales. Preference for categorization over sensory description did not significantly predict whether a participant possessed an E-plus or E-minus evaluative habit, confirming that structural-cognitive habits (C-D) and affective-evaluative habits (E+/E-) operate along independent psychometric axes.
• Modern Latent Trait Implications: In contemporary psychometrics, forced-choice instruments of this typology are typically analyzed using modern item response theory (IRT) models tailored to paired comparisons, such as the Thurstonian IRT model. Such formulations demonstrate that Nunnally’s tripartite division (E-plus, E-minus, and C-D balance) provides an effective three-dimensional representation of semantic retrieval preferences.

10. Instrument / Measurement Tool

The Semantic Habits Measure is structured as follows:

• Test Type: Objective performance-based cognitive style test / forced-choice word association measure.
• Format: Paper-and-pencil inventory consisting of 143 stimulus words, each accompanied by two competing response words.
• Response Modality: Binary forced-choice (the respondent must select the single word that feels most naturally or logically associated with the stimulus word).
• Number of Items: 143 items.
• Component Subscales and Major Composites:
  • Subscale 1: E-plus versus D (Positive evaluation vs. Sensuous denotative attribute).
  • Subscale 2: E-plus versus C (Positive evaluation vs. Taxonomic categorization).
  • Subscale 3: E-minus versus D (Negative evaluation vs. Sensuous denotative attribute).
  • Subscale 4: E-minus versus C (Negative evaluation vs. Taxonomic categorization).
  • Subscale 5: C versus D (Taxonomic categorization vs. Sensuous denotative attribute).
• Major Scored Scales:
  • E-plus Scale: Total number of positive evaluative selections across Subscale 1 and Subscale 2.
  • E-minus Scale: Total number of negative evaluative selections across Subscale 3 and Subscale 4.
  • C-D Balance Scale: Total number of C (categorization) choices on Subscale 5. Higher scores denote a preference for categorization (C), whereas lower scores denote a preference for sensuous denotative attributes (D).
• Administration Time: Approximately 20 to 30 minutes under untimed, self-paced conditions.
• Target Population: Adolescents and adults with basic literacy skills; originally normed and validated on university undergraduate students.

11. Permissions & Fee and Test Year

The Semantic Habits Measure was developed in 1963 and published in the peer-reviewed journal Educational and Psychological Measurement. The scale was developed within an academic research context and is not distributed as a commercial testing product. No fee is associated with its standard scientific or academic use.

Researchers intending to administer or adapt the Semantic Habits Measure for experimental or empirical studies may consult the original publication (Nunnally, Flaugher, & Hodges, 1963). In accordance with standard fair-use practices for historical psychological measures published in academic journals, replication for non-commercial scholarly research is generally permitted, provided proper citation is given to the authors and the original publisher (Sage Publications / Educational and Psychological Measurement).

12. References

• Bruner, J. S., Goodnow, J. J., & Austin, G. A. (1956). A study of thinking. John Wiley & Sons.
• Hodges, W. F., & Spielberger, C. D. (1969). Digit span: An indicant of state or trait anxiety? Journal of Consulting and Clinical Psychology, 33(4), 430–434. https://doi.org/10.1037/h0027814
• Kent, G. H., & Rosanoff, A. J. (1910). A study of association in insanity. American Journal of Insanity, 67(1), 37–96.
• Nunnally, J. C. (1967). Psychometric theory (1st ed.). McGraw-Hill.
• Nunnally, J. C. (1978). Psychometric theory (2nd ed.). McGraw-Hill.
• Nunnally, J. C., Flaugher, R. L., & Hodges, W. F. (1963). Measurement of semantic habits. Educational and Psychological Measurement, 23(3), 419–434. https://doi.org/10.1177/001316446302300305
• Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning. University of Illinois Press.

13. Items of the Scale