Development and Validation of an Instrument to ... - Springer Link

14 downloads 299 Views 95KB Size Report
Miller & Swift, 1988; Parks & Roberton, 1998a, 1998b; Rubin & Greene,. 1991; Rubin .... The purposes of this study, therefore, were to (a) develop a conceptual.
Sex Roles, Vol. 42, Nos. 5/6, 2000

Development and Validation of an Instrument to Measure Attitudes Toward Sexist/Nonsexist Language Janet B. Parks1 and Mary Ann Roberton Bowling Green State University

Despite documented, negative effects of sexist language, some people still strongly oppose inclusive language. Their attitudes need study, but existing measures of attitude lack comprehensiveness or validity. This paper reports construction and validation of the 42-item Inventory of Attitudes Toward Sexist/Nonsexist Language (IASNL), which is based on a conceptual framework related to beliefs about language, recognition of sexist language, and willingness to use inclusive language. Four studies tested the IASNL’s content validity (9 content experts, 329 other participants), construct validity, and reliability (31 advocates of inclusive language, 298 additional respondents). Cronbach alphas were high and validity measures were strong. Two short forms of the IASNL were also created—the IASNL-General and the IASNLSport. The IASNL-General is in the Appendix. Sexist language has been defined as ‘‘words, phrases, and expressions that unnecessarily differentiate between females and males or exclude, trivialize, or diminish either gender’’ (Parks & Roberton, 1998a, p. 455). Theoretically, sexist language could have deleterious effects on either women or men, but empirical evidence in American society strongly suggests that the preponderance of negative results accrue to women. Concern about the negative effects of sexist language is grounded in the Sapir–Whorf hypothesis of linguistic relativity, which suggests an affinity between language and culture (Cameron, 1985; Carroll, 1956; Simpson, 1993; Thorne, Kramarae, & Henley, 1983). The strong version of this hypothesis maintains that language 1

To whom correspondence should be addressed at 112 Gertrude M. Eppler Center, School of Human Movement, Sport, and Leisure Studies, Bowling Green State University, Bowling Green, Ohio 43403-0248; e-mail: [email protected]. 415 0360-0025/00/0300-0415$18.00/0  2000 Plenum Publishing Corporation

416

Parks and Roberton

directly causes particular behaviors in members of a culture. The moderate version holds that culture and language are intertwined such that the meanings people ascribe to language affect their realities, their self-concepts, and their world views (Bing, 1992; Lakoff, 1973; Miller & Swift, 1988; Parks & Roberton, 1998a, 1998b). Although research has not supported the strong version (Bing, 1992), evidence for the moderate version has been provided. Studies have revealed that pseudogeneric pronouns and job titles (e.g., he, mailman) (a) imply the superiority of the masculine over the feminine (Gastil, 1990; Hyde, 1984), (b) produce masculine images or confusion in the receivers’ minds (Bodine, 1975; Hamilton, 1988; Khosroshahi, 1989; Kidd, 1971; Martyna, 1978; McConnell & Fazio, 1996; Merritt & Kok, 1995; Moulton, Robinson, & Elias, 1978; Schneider & Hacker, 1973; Stericker, 1981), and (c) have an effect on the self-concepts of women and men (MacKay, 1980; Nilsen, 1977). Scholars have also suggested that negative effects can result from the use of other pseudogenerics such as man and mankind (Ng, 1990; Schneider & Hacker, 1973; Shimanoff, 1977) as well as terms that deny women their adulthood (e.g., girl) or trivialize women’s accomplishments (e.g., Lady Bisons) (Eitzen & Zinn, 1989, 1993; Messner, Duncan, & Jensen, 1993). Based on this documentation as well as the possibility that there are additional, as yet undiscovered, negative effects of sexist language, many academics have suggested that sexist language should be reformed (Blaubergs, 1980; Henley, 1989; Merritt & Kok, 1995; Miller & Swift, 1988; Parks & Roberton, 1998a, 1998b; Rubin & Greene, 1991; Rubin, Greene, & Schneider, 1994). Despite the empirical evidence, language reform is still controversial. The topic of nonsexist or inclusive language continues to evoke emotional reactions from the general populace. For example, Parks and Roberton (1998a) recently found that 53% of a group of contemporary college students opposed some aspects of inclusive language. Twenty-one percent resisted all forms of inclusive language. Interestingly, many of those who trivialized the need for inclusive language wrote the longest diatribes against it (see Parks & Roberton, 1998a, for specific examples). It is clear that more studies of people’s attitudes toward sexist/nonsexist language are needed. To date, studies associated with sexist language have examined either the usage of sexist language or attitudes toward the use of sexist language. While usage can be studied through writing samples, speeches, and other observable behaviors, the study of attitudes toward such language is more challenging. Moreover, not enough is known about the complex question of whether people’s attitudes toward sexist language predict or even correlate with their usage of such language (McMinn, Williams, & McMinn, 1994; Rubin et al., 1994). To be able to examine questions about attitudes

Attitudes Toward Sexist Language

417

as well as the relationship between usage and attitude, a valid and reliable measure of attitude is needed.

ASSESSMENT OF ATTITUDES TOWARD SEXIST/NONSEXIST LANGUAGE Three primary instruments have been used in studies of attitudes toward sexist language. The first was a questionnaire constructed by Henley and Dragun (1983). The second was an adaptation of the Henley and Dragun instrument constructed by Rubin and Greene (1991). The third instrument was created by McMinn et al. (1994).

Language Questionnaires Using a self-report questionnaire, Henley and Dragun (1983) conducted the first survey of attitudes and behavior regarding sex-biased language. Based in part on the work of Bate (1978), Henley and Dragun’s instrument elicited information from respondents in eight categories: (a) degree of interest in sexism in language; (b) methods respondents had used to avoid sexist language; (c) respondents’ impetus for changing to inclusive language; (d) the need for a single sex-neutral pronoun and respondents’ attitudes toward suggested pronouns; (e) situations in which respondents used ‘‘they’’ as a singular pronoun; (f) how often respondents heard alternatives to the male ‘‘generic’’; (g) respondents’ preferences for sex-neutral pronoun systems (e.g., ‘‘per, pers, perself’’; ‘‘s/he’’); and (h) influences on their adoption of one of the sex-neutral pronouns. The Henley and Dragun instrument was subsequently used by Harrigan and Lucic (1988). Rubin and Greene’s (1991) adaptation of the Henley and Dragun (1983) instrument was titled The Language Questionnaire. It comprised six sections: one item assessed respondents’ beliefs about sexist language, one item reflected their level of concern about sexist language, and one item required respondents to rank 10 possible influences for their adoption of nonsexist language; additionally, there were three composite scales that addressed judgments of the degree to which certain statements were sexist, methods respondents used to avoid sexist language, and their willingness to use alternatives to ‘‘he’’ as a generic pronoun. In a sample of 128 undergraduate students and 119 college graduates, the Cronbach alphas for the three composite scales of The Language Questionnaire were .84, .67, and .85, respectively (Rubin & Greene, 1991). In a

418

Parks and Roberton

subsequent study of 88 undergraduate students, Rubin et al. (1994) reported Cronbach alphas of .78, .66, and .89, respectively, for the three scales.

Gender-Specific Language Scale McMinn et al. (1994) constructed the Gender-Specific Language Scale (GSLS), which assessed respondents’ ability to recognize sexist language. The GSLS was a 12-item scale that contained statements with grammatical, spelling, or punctuation errors. Six of the 12 statements also contained gender-specific language. These sexist language errors were based on ‘‘designation errors’’ identified in the sexist language section of the third edition of the Publication Manual of the American Psychological Association (1983). Examples of designation errors included the exclusive use of masculine pronouns and the use of ‘‘generic’’ masculine terms such as ‘‘man,’’ ‘‘chairman,’’ and ‘‘mailman.’’ Respondents to the GSLS were instructed to identify all errors in the statements. Unknown to the participants, their score on the GSLS was the number of correct identifications of sexist language. The psychometric properties of the 12-item GSLS have not been established. The validity of an earlier 18-item version was established by comparing 141 college students’ scores on the GSLS to the number of times they used sexist language on essay questions. The moderate inverse correlation between the two measures (r ⫽ ⫺.42) suggested a slight tendency for respondents who recognized sexist language to avoid its use. In two earlier studies, one with 129 college students and the other with 35 college students, Cronbach alphas for the 18-item GSLS were .81 and .95, respectively.

Need for a New Instrument The Henley and Dragun (1983) questionnaire, the Rubin and Greene (1991) instrument, and the GSLS (McMinn et al., 1994) have been useful in studying various aspects of sexist language. These instruments, however, were not designed to incorporate all aspects of the construct known as ‘‘attitude toward sexist language.’’ The instruments either focused more narrowly on subsets, such as ‘‘recognition,’’ or they were not grounded in an explicit conceptual framework that clearly defined the domain being sampled. McMinn et al. established moderate content validity for the GSLS, but did not address construct or discriminant validity. Information about the validity of the Henley and Dragun instrument and The Language Questionnaire also has not been published. There is a need, therefore, for a

Attitudes Toward Sexist Language

419

valid and reliable measure built on the contributions of these earlier instruments. This measure could promote the systematic, coherent examination of attitudes toward sexist/nonsexist language. The purposes of this study, therefore, were to (a) develop a conceptual framework relative to attitudes toward sexist/nonsexist language, (b) design an instrument that sampled the elements of that framework, and (c) validate the use of that instrument. The remainder of this paper describes the construction of the Inventory of Attitudes Toward Sexist/Nonsexist Language (IASNL), provides evidence of its validity and reliability, and discusses possible uses of the instrument and its shorter forms in research.

DEVELOPMENT OF THE IASNL Conceptual Framework The content validity of an instrument deals with the degree to which items in the instrument sample the entire domain under study (Nunnally & Bernstein, 1994). Before designing the IASNL, we searched the literature on attitudes toward sexist language in order to establish a conceptual framework that would define that domain. We found that studies of attitudes toward sexist language had been conducted from a variety of perspectives, with the preponderance of the research falling into three categories: 1. Respondents’ beliefs, thoughts, or opinions about sexist/nonsexist language (Bate, 1978; Blaubergs, 1980; Eitzen & Zinn, 1993; Harrigan & Lucic, 1988; Henley & Dragun, 1983; Rubin & Greene, 1991; Rubin et al., 1994). 2. Respondents’ ability to recognize sexist language (McMinn et al., 1994; Rubin & Greene, 1991; Rubin et al., 1994); 3. Respondents’ use of inclusive language or their willingness to use inclusive language (Cronin & Jreisat, 1995; Harrigan & Lucic, 1988; Henley & Dragun, 1983; Jacobson & Insko, 1985; McMinn, Lindsay, Hannum, & Troyer, 1990; McMinn, Troyer, Hannum, & Foster, 1991; Nilsen, 1984; Richmond & Dyba, 1982; Rubin & Greene, 1991; Rubin et al., 1994). We inferred that an instrument tapping into all three of these categories would possess content validity and would be useful in measuring attitudes toward sexist/nonsexist language. Consequently, these three areas provided the conceptual foundation for the IASNL.

420

Parks and Roberton

Beliefs About Sexist Language One way to ascertain people’s beliefs about language is to examine arguments they advance either for or against the use of inclusive language. In 1980, Blaubergs identified eight arguments against inclusive language that had been used by scholars during the 1970s. Recently, Parks and Roberton (1998a) verified that the arguments Blaubergs discovered almost 20 years earlier were still being used by contemporary college students to justify sexist language. Therefore, we used these arguments as a conceptual framework for constructing the first section of the IASNL, which assesses respondents’ beliefs about sexist language. Table I presents the eight arguments identified by Blaubergs (1980).

Recognition of Sexist Language As shown in previous research, attitudes toward sexist/nonsexist language can also be manifest in a person’s ability to recognize sexist language. Three categories of sexist language identified by Miller and Swift (1988) and explored in an educational videotape about sexist language (Parks, Harper, & Lopez, 1994) provided the conceptual framework for this section: (a) false generics such as ‘‘he’’ and ‘‘mankind,’’ (b) hierarchical and separatist terms such as ‘‘man and wife’’ and ‘‘waiter’’ or ‘‘waitress,’’ and (c) language that denies women their self-esteem and personal identity, such as referring to adult females as ‘‘girls’’ rather than ‘‘women’’ and underscoring women’s marital status with ‘‘Mrs.’’ or ‘‘Miss.’’ Each recognition item on the IASNL reflected one of these categories. Rubin and Greene’s (1991, 1994) Language Questionnaire contained a number of recognition items that were compatible with our theoretical framework. We based some items in our recognition section on concepts included in their instrument and, with their permission, used three of their items verbatim.

Willingness to Use Inclusive Language Willingness to use inclusive language could be the single most revealing aspect of a person’s attitude toward sexist language. The willingness section of the IASNL was based on the same conceptual framework as the recognition section: false generics, hierarchical or separatist terms, and denigrating language relative to women’s self-esteem or identity (Miller & Swift, 1988; Parks et al., 1994; Rubin & Greene, 1991).

Attitudes Toward Sexist Language

421

Table I. Sample Items from the IASNL Based on Blaubergs’ (1980) Classification of Arguments Against Changing Sexist Language Argument Cross-cultural

Language is a trivial concern

Freedom of speech/unjustified coercion

Sexist language is not sexist

Word etymology

Appeal to authoritya

Change is too difficulta

Historical authenticity

Definition No evidence that cultures using sexist language have more sex discrimination than those using a nonsexist language; no evidence that nonsexist languages result in equal treatment of the sexes Sexist language is trivial compared to more serious injustices in society as well as to other forms of sexism; worrying about sexist language is frivolous in general People who want to change the language are ‘deviants’ who intimidate others with coercive tactics and deprive them of their freedom of speech The language is not really sexist because the users do not have sexist intentions; the problem with these terms is the listener who misinterprets terms as sexist The original meaning of a word is justification for its use The final authorities for the meanings of words are the dictionary, linguists, or people who are important in society or in our lives Most expressions, particularly pronouns, are too deeply ingrained to be changed by individuals; some people fear change, don’t want to break habits, are stubborn, etc. Changing to nonsexist language would require the rewriting of great literature, English idioms, and historical documents

IASNL Item Sexist language is related to sexist treatment of people in society

The elimination of sexist language is an important goal

Teachers who require students to use nonsexist language are unfairly forcing their political views upon their students When people use the term man and wife, the expression is not sexist if the users don’t mean it to be If the original meaning of the word he was person, we should continue to use he to refer to both males and females today Because TV sportscasters refer to female professional golfers as girls, it must be OK The English language will never be changed because it is too deeply ingrained in the culture We should not change the way the English language has traditionally been written and spoken

Note. Adapted from Parks and Roberton (1998a), copyright 1998 by Plenum Publishing Corporation. a Modified slightly from Blaubergs’ (1980) original definitions. See Parks and Roberton (1998a).

422

Parks and Roberton

The Sport Context In addition to identifying three aspects of attitude toward sexist language, researchers have begun to examine language choices relative to specific contexts. Examples of such contexts are occupations (Matheson & Kristiansen, 1987; McMinn et al., 1990, 1991), and sport (Parks & Roberton, 1998b). Sport is a particularly interesting context for the study of sexist language because it is a powerful, traditionally conservative social institution that privileges males over females. The pervasive influence of sport on the values and attitudes of the general population demonstrates its power in American society. One example of this power is the adulation accorded many athletes, both amateur and professional, and the concomitant expectation that athletes should serve as positive role models for youth. The influence of athletes is also reflected in the sizable royalties paid for their endorsements of a multitude of products, ranging from long-distance telephone services to beer to sports bras. An additional example of the influence of sport is the amount of attention devoted to athletic contests and the sport industry by both print and broadcast media. Although individual athletes might exhibit flamboyant or unconventional behavior, sport is generally regarded as a conservative social institution (Eitzen & Sage, 1997). One manifestation of this conservatism is the resistance of sport to changing its traditionally masculine environment. For example, Title IX of the Education Amendments Act was legislation that mandated gender equity in educational programs receiving federal funding. Although this legislation was passed in 1972, most intercollegiate athletics programs still avoid compliance (Richey, 1997). Another example of sport’s conservatism is the homophobia perpetuated by coaches and athletes (Krane & Kane, 1998; Woog, 1998). An example of male privilege in sport is the overrepresentation of male administrators in intercollegiate athletics. In 1996, Acosta and Carpenter reported that in the United States, men constituted 76% of all administrators of women’s athletics programs. Acosta and Carpenter also noted that more women served as college presidents than as athletics directors with authority over both men’s and women’s programs. The language associated with sport also reflects male privilege. Common expressions, such as ‘‘Lady Bulldogs’’ and ‘‘man-to-man’’ defense, diminish and exclude women (Eitzen & Zinn, 1989, 1993). Finally, research has shown that the media contribute to the favored status of men in sport through reporting styles that emphasize the strength and power of males, but the appearance and personal characteristics of females (Bruce, 1998; Duncan & Hasbrook, 1988; Eastman & Billings, 1999; Kane & Parks, 1992; Messner et al., 1993).

Attitudes Toward Sexist Language

423

These characteristics render sport a worthy context in which to examine attitudes toward sexist language. In recognition of the pervasiveness of sport in American society and the need to facilitate the study of sport as a contextual variable in language research, we included sport items in early versions of the IASNL. FORMAT AND SCORING OF THE IASNL Version 1 and all succeeding versions of the IASNL have three sections, each of which has separate directions and wording. To control among respondents for competing interpretations of the term ‘‘sexist language,’’ we presented the Parks and Roberton (1998a) definition of sexist language in a box on each page of the IASNL. We reasoned that if we allowed each respondent to define sexist language in her/his own way, the results of the questionnaire would be uninterpretable. By including the definition on the instrument, we ensured that all respondents were formulating their responses in terms of the same definition. In Version 1, the beliefs section contained 32 items—4 items for each of Blaubergs’ (1980) arguments, 2 of which were placed in a general context and 2 of which were placed in a sport context. Examples of items generated from each of the 8 arguments identified by Blaubergs are shown in Table I. The recognition section included 4 general context items and 4 sport items. Three of the items reflected false generics, 3 used hierarchical or separatist terms, and 2 used language that denigrated women’s identity or self-esteem. The last section of Version 1 assessed respondents’ willingness to use inclusive language. Half of the 12 items in this section were placed in a sport context and half in a general context. Four of the willingness items used false generics (2 general, 2 sport), 4 items used hierarchical terms (2 general, 2 sport), and the remaining 4 items used terms associated with identity or self-esteem (2 general, 2 sport). All items on Version 1 and all subsequent versions of the IASNL were scored on a 5-point Likert-type scale. High scores (4–5) indicated a positive attitude toward inclusive language; low scores (1–2) indicated a negative attitude toward inclusive language. A score of 3 indicated neutrality or uncertainty. Selected items in the beliefs section were reverse-scored. Each respondent’s score on the instrument was the total of all the items. VALIDITY AND RELIABILITY OF THE IASNL We developed Version 1 of the IASNL in 1994–1995. After obtaining permission from the Human Subjects Review Board at our university, we

424

Parks and Roberton

launched a series of pilot studies. In the fall of 1995, we tested Version 1 with 173 students and faculty. Using feedback from the participants on characteristics such as clarity of directions, clarity of individual items, and the length of the instrument, we developed Version 2, pilot-testing it with 156 students, faculty, and personnel managers. Again, we revised the instrument based on feedback from these participants relative to clarity and length, resulting in Version 3.

Content Validity As indicated earlier, initial support for the content validity of the IASNL is based on the fact that we grounded the instrument in a conceptual framework derived from the published research. The framework conceptualized the domain of attitudes toward sexist/nonsexist language as being composed of beliefs, recognition, and willingness. We further examined the content validity of Version 3 of the IASNL with the assistance of nine university professors (six women and three men) who were established scholars in their respective fields. Two of these scholars had published research articles about sexist language in general, three of them had published research about sexist language in sport, and four of them had been identified by their national scholarly association as experts on gender issues. We sent each expert a copy of the IASNL and asked them to identify and comment on inaccurate or ambiguous items, to identify and comment on items that did not appear to measure the construct of interest, and to make any other observations and suggestions they deemed appropriate. Using feedback from these scholars, we reworded and clarified several items and developed Version 4 of the instrument. The format of Version 4 remained the same as Version 3. It contained 52 items that had been placed randomly within their respective sections of the instrument. The section on beliefs about inclusive language contained 32 items, the recognition section had 8 items, and the section on willingness to use inclusive language had 12 items. Half of the 52 items dealt with a general context and half with a sport context.

Construct Validity We then examined the construct validity of the IASNL, that is, the degree to which the instrument measured the psychological attribute of ‘‘attitude toward sexist/nonsexist language’’ (Nunnally & Bernstein, 1994). We sent copies of Version 4 to volunteer site coordinators at six universities

Attitudes Toward Sexist Language

425

in the United States. Each of these volunteers distributed the instrument to students, faculty, and administrators at their home institutions. We also administered the instrument to students, faculty, and administrators at our home institution. The total sample size was 298. The participants were primarily White (86.1%) or African-American (7%), and students (80.1%) or professors (11.1%). The ages of the participants ranged from 18 to 59 years of age. One hundred forty-nine (52.3%) participants were 18, 19, or 20 years old; 44 (15.4%) were 21 or 22 years old; and 92 (32.3%) were 23 years old or older.

Item Analysis We conducted an item analysis on the data generated by the 298 participants from the seven universities. First, we calculated item-to-total correlations (discrimination index) for all 52 items. We then eliminated the items that correlated with the total score at ⬍.40, a more stringent criterion than the correlations of .25 and .20 considered acceptable for discrimination by Ary, Jacobs, and Razavieh (1990) and Rust and Golombok (1989), respectively. This step eliminated 10 items, 7 of which were from the beliefs section and 3 of which were from the willingness section. Five of the deleted items were from the general context, and 5 were from the sport context. This procedure produced Version 5, a 42-item instrument with 21 general items and 21 sport items. Section I (beliefs) contained 25 items; section II (recognition) contained 8 items; section III (willingness) contained 9 items. Table II shows the scores of the 298 participants on the questions that formed Version 5. The average score (3.22) across items was midway on the 1–5 scale and the average standard deviation across items was 1.22. Each of the 42 items had a range of responses that spanned the 1–5 scale. Based on these scores, we concluded that the instrument had an Table II. Means and Standard Deviations of 298 Respondents to the 42-Item IASNL (Version 5) Section

Number of Items

Possible Range

M

SD

Total instrument Beliefs Recognition Willingness Generala Sportb

42 25 8 9 21 21

42–210 25–125 8–40 9–45 21–105 21–105

135.41 79.51 22.95 32.95 66.59 66.82

27.88 16.55 7.26 7.95 14.90 14.04

a

IASNL-G. IASNL-S.

b

426

Parks and Roberton

appropriate facility index (Rust & Golombok, 1989); that is, the items elicited different responses from different respondents. The item-to-total correlations for the 42 items ranged from .394 to .73. Only 1 of the items correlated with the total score at less than .40. This general item correlated with the total at .394, but we retained it in order to preserve the balance of general and sport-related items. The item-tosection-total correlations for the 25 items in the beliefs section ranged from .42 to .70; for the 8 items in the recognition section, the coefficients ranged from .53 to .78; and for the 9 items in the willingness section the coefficients ranged from .57 to .79. For the 21 sport items, the item-to-section-total correlations ranged from .42 to .70, and for the general items, they ranged from .43 to .75. In all cases, including the general and sport sections, each item correlated more highly with its own section total than with the totals of the other sections (Ary et al., 1990).

Exploratory Factor Analysis Although the correlations indicated that the three sections of the IASNL had moderate within-section integrity, there appeared to be considerable overlap, suggesting they were measuring essentially the same construct. The correlation between the section on beliefs and the section on recognition was .75, indicating that they shared 56% of the variance. The correlation between beliefs and willingness was .55; the correlation between recognition and willingness, .54, indicating that both beliefs and recognition shared approximately 30% of the variance with willingness. This overlap motivated us to perform an exploratory factor analysis, using the principal components extraction method followed by varimax rotation. The Kaiser criterion suggested that eight principal components should be rotated. However, it is known that the Kaiser criterion vastly overestimates the number of factors when the data are composed of many variables (items) with low commonalities (Stevens, 1996). These data had both characteristics. The alternative Cattell scree plot of the eigenvalues suggested that the data should be characterized by only one factor. Faced with these conflicts, the investigators decided to rotate the first three principal components, which accounted for 41% of the variance. Each successive principal component individually accounted for only 3% or less additional variance. Table III contains the three rotated factors and the factor loadings. It is clear that the three factors did not represent the test sections of beliefs, recognition, and willingness nor of sport and general context. Indeed, both belief and recognition items were represented in the higher loadings on the first two factors. Interestingly, the willingness section did load most

Attitudes Toward Sexist Language

427

Table III. Factor Patterns for the IASNL (Version 5) Yielded from Varimax Rotation (N ⫽ 298) Item number 12 18 33 7 26 5 27 15 31 1 23 16 2 8 28 9 10 25 13 14 6 21 24 20 22 30 19 29 11 32 3 4 17 40 41 37 35 38 39 42 34 36 Eigenvalues Pecent of variance

Factor 1

Factor 2

Factor 3

0.71279 0.70059 0.69494 0.66881 0.64481 0.64197 0.59703 0.59598 0.57204 0.56799 0.56068 0.49845 0.45219 0.39673 0.33705 0.22570 0.12001 0.24951 0.06184 0.16909 0.21259 0.13575 0.46032 0.19294 0.10112 0.33483 0.35649 0.35093 0.36306 0.22435 0.35707 0.20511 0.17670 0.41030 0.03474 0.28481 0.00153 0.37194 ⫺0.05642 ⫺0.12209 0.30865 0.35048 7.0103 17.00

0.20084 0.14845 0.26456 0.17014 0.15872 0.18842 0.18260 0.30523 0.30913 0.14954 0.36911 0.44333 0.36626 0.16692 0.32041 0.67658 0.61690 0.60883 0.58791 0.58497 0.56311 0.55426 0.55174 0.55121 0.53938 0.42047 0.39813 0.39644 0.37792 0.37457 0.37436 0.37124 0.36837 0.10416 0.18109 0.10973 0.20558 0.06368 0.28410 0.39518 0.16058 0.00347 6.0381 14.38

0.11979 0.23603 0.24673 ⫺0.16025 0.22822 0.10221 0.18961 0.13762 0.24806 ⫺0.02359 0.11304 0.17687 0.28044 0.18400 0.04887 0.18700 0.19916 0.21666 0.09704 0.07110 0.11220 0.00780 0.07686 0.12626 0.19945 0.13745 0.13546 0.18771 0.05155 0.16551 0.13630 0.05505 0.13014 0.70199 0.69358 0.65977 0.65440 0.64649 0.59739 0.55684 0.54886 0.51627 4.3365 10.32

Note: Factor loadings ⬎.298 are significantly different from 0 at 움 ⫽ .01.

428

Parks and Roberton

highly and fairly exclusively on the third factor. The other items did not load strongly on this factor. Sport and general items loaded across all three factors. Analysis of the items loading most highly on factor 1 suggested that factor 1 reflected recognition and beliefs statements that gave specific examples. For instance, the item that loaded most highly on factor 1 used the example of ‘‘man-to-man’’ defense. The highest loadings on this factor all involved items containing specific examples of sexist language. In contrast, factor 2 seemed to reflect beliefs about sexist/nonsexist language in an abstract sense. Of the items loading most strongly on this factor, seven were statements of philosophy that did not contain examples of specific sexist phrases. For example, the item with the highest loading was the philosophical statement, ‘‘The elimination of sexist language is an important goal.’’ Factor 3 items were those which, again, gave specific examples, but also asked readers their willingness to change their own language usage. This factor, which accounted for 10% of the variance, may be an artifact of the wording common to all the items loading on this factor: ‘‘How willing are you . . . ?’’ Such an artifact is called a ‘‘bloated specific’’ in the lexicon of factor analysis (Kline, 1994).

Discriminant Validity Discriminant validity is also an aspect of construct validity (Nunnally & Bernstein, 1994). It indicates the degree to which the IASNL differentiates between groups of people who should have different attitudes toward sexist/ nonsexist language. To establish the discriminant validity of the IASNL, we generated a list of 41 college faculty, administrators, and graduate students who had reputations for being knowledgeable about gender issues. We contacted these individuals and asked them (a) if they were feminists, (b) if they would be willing to be characterized as advocates of changing sexist language to nonsexist forms, and (c) if they would assist us in the construction of the IASNL. We sent the IASNL to the 31 advocates who agreed to participate. All of them completed the instrument and returned it. Most of these advocates were White, female college professors. They were in academic fields such as liberal studies, sociology, kinesiology, social justice education, physical education, political science, sport management, and women’s studies. The ages of the advocates ranged from 35 to 57 years, with a mean age of 46.4 years. We assessed discriminant validity by comparing the mean scores of the 31 advocates of language change with the scores of the 298 participants on Version 5 of the IASNL. We reasoned that if the means for the two

Attitudes Toward Sexist Language

429

groups were at least 1 standard deviation unit apart, we would verify that the IASNL could discriminate between a ‘‘known group’’ (advocates of change) and participants who had not identified themselves as advocates of change. On a 42- to 210-point scale, the mean score for the 31 advocates was 200.42 (SD ⫽ 9.48) and the mean score for the 298 test site participants was 135.41 (SD ⫽ 27.88). A t test showed this difference between means to be statistically significant, t(327) ⫽ 27.66, p ⬍ .001. The magnitude of the difference between the means was 2.43 standard deviation units (Cohen’s [1988] d). The IASNL can clearly differentiate between advocates of change and the general populace.

Reliability Again using the data generated by the 298 participants, we assessed the internal consistency of Version 5 of the IASNL using Cronbach’s coefficient alpha. Because earlier studies had found age and gender differences (Parks & Roberton, 1998b) and regional differences (Nilsen, 1984) in attitudes toward or use of sexist language, we also examined the Cronbach alphas for these subgroups. The alpha for the complete inventory taken by the total sample was .94. Even the separate sections of the IASNL had alphas of .83 and higher. Indeed, the alphas for sections of the IASNL by subgroups of the sample were all over .70, the minimum recommended by Nunnally and Bernstein (1994) for a reliable test (see Table IV).

DISCUSSION Psychometric Properties Version 5 of the IASNL fulfills a need for a valid and reliable instrument that will measure attitudes toward sexist/nonsexist language. Its psychometric properties are excellent. The Cronbach alphas were well over the minimum specified as acceptable by Nunnally and Bernstein (1994), indicating a high degree of internal consistency. Moreover, as shown in Table IV, the alphas remained consistently high across several age groups, genders, and geographical regions. Version 5 of the IASNL also has strong content, construct, and discriminant validity. It was constructed from an explicit conceptual framework, which experts subsequently agreed it sampled well. The items within the instrument correlated well with each other and the total, further contributing to its reliability as well as its construct validity. Lastly, the instrument

b

IASNL-G. IASNL-S.

.94 .91 .83 .85 .89 .88

Total inventory Beliefs Recognition Willingness Generala Sportb a

Sample (N ⫽ 298)

Section .95 .91 .86 .90 .90 .90

South (n ⫽ 102) .93 .90 .79 .80 .88 .86

Midwest (n ⫽ 196) .95 .92 .85 .85 .91 .89

Women (n ⫽ 165) .91 .86 .76 .82 .85 .83

Men (n ⫽ 123)

.92 .89 .77 .80 .86 .85

18–20 Years (n ⫽ 149)

.93 .89 .80 .81 .88 .87

21–22 Years (n ⫽ 44)

.95 .93 .87 .89 .91 .91

23 ⫹ Years (n ⫽ 92)

Table IV. Cronbach’s Coefficients Alpha Based on the Scores of 298 Respondents to the 42-Item IASNL (Version 5)

430 Parks and Roberton

Attitudes Toward Sexist Language

431

has been shown to distinguish between advocates of inclusive language and the general populace. On Version 5, scores of advocates for language change were 2.43 standard deviation units higher than those of 298 students, faculty, and administrators from seven universities. A Cohen’s d of this magnitude is exceptionally large in behavioral science research (Cohen, 1988). In addition to differentiating inclusive language advocates from the general public, the IASNL also differentiates people on the basis of their age and their gender. Using Version 4 of the instrument, Parks and Roberton (1998b) found that 11% of the total variance in scores was accounted for by gender and 12% by age. Men were significantly less supportive of nonsexist language than were women. People 23 years of age and older were significantly more supportive of nonsexist language than were 21- to 22-year-olds, who were, in turn, significantly more supportive than 18- to 20-year-olds.

Structure of the IASNL While arranged in three sections, each of which is worded differently, the IASNL seems to be a unitary measure of the overall construct of ‘‘attitude toward sexist/nonsexist language.’’ With the exception of ‘‘willingness,’’ the three sections of the IASNL did not load separately on three factors in the exploratory factor analysis. Nor did the sport and generalcontext items load separately. This pattern suggested that the beliefs, recognition, sport, and general sections used to create the IASNL are not tapping discrete constructs. In addition, Parks and Roberton (1998b) found that the difference between the sport context and the general context accounted for less than 1% of the variance in respondents’ attitudes toward sexist language. Thus, the sport items on the IASNL seem as ‘‘general’’ as the general items, making the instrument useful in any context. At this point, therefore, the evidence suggests that the IASNL measures one construct, attitude toward sexist/nonsexist language. Whether the ‘‘willingness’’ section will prove to be a separate construct or merely an artifact of wording needs further study. Indeed, since factor analysis can be sample-specific, additional research using new samples, different numbers of factors, and different rotation techniques is warranted.

Short Forms of the IASNL In Version 5, the correlation between the 21 general items and the 21 sport items was .86, suggesting that the two sections of the instrument could

432

Parks and Roberton

be used independently. Given that some researchers may be interested in either the general context or the sport context, but not both, we created two short forms of the IASNL. One, the Inventory of Attitudes Toward Sexist/Nonsexist Language-General (IASNL-G), contains only the 21 general items from Version 5. The item-to-total correlations of the IASNL-G range from .43 to .75. It has a Cronbach alpha of .89 (see Table IV). These 21 general items reflect the same elements of the conceptual framework on which Version 5 was based. The IASNL-G is included in the Appendix. Readers interested in studying sexist language in the sport context can contact the authors to obtain the other short form, the Inventory of Attitudes Toward Sexist/Nonsexist Language-Sport (IASNL-S). This instrument contains the 21 sport-related items from Version 5. The item-to-total correlations for the IASNL-S range from .42 to .70. It has a Cronbach alpha of .88 (see Table IV) and reflects the same elements of the conceptual framework on which Version 5 was based. The full-length, 42-item IASNL is also available from the authors.

Future Research Parks and Roberton (1998b) suggested that variables affecting attitude toward sexist/nonsexist language could be divided into ‘‘person’’ characteristics (e.g., age, gender) and ‘‘contextual’’ characteristics (e.g., occupation, work environment, social institutions). Within the category of person characteristics, future research could use the IASNL to determine whether the age differences reported in several studies (Cronin & Jreisat, 1995; Nilsen, 1984; Parks & Roberton, 1998b; Rubin & Greene, 1991) can be replicated and whether they are due to cohort differences, experience, or learning. Parks and Roberton (1998b), for instance, have hypothesized that age differences might be due to the increased empathy that sometimes accompanies experience. On the other hand, Rubin and Greene (1991) have speculated that the political climate in which a person grows up might produce a cohort effect leading to either activism or complacency regarding sexist language. The question ‘‘What does age represent?’’ (Parks & Roberton, 1998b) is worthy of exploration. In the contextual area, studies could use the IASNL to examine attitudes among people of different occupations (e.g., educators, physicians, attorneys), academic disciplines (e.g., humanities, business, education), or socioeconomic status. Harrigan and Lucic (1988) found that effects of group membership (e.g., National Organization for Women, medical school) were greater than the effects of gender. Subsequently, Rubin and Greene (1991) hypothesized that the effect of group membership might become less pro-

Attitudes Toward Sexist Language

433

nounced as inclusive language enjoys more widespread use. Studies designed to test this hypothesis would be useful. Another line of research in which the IASNL could be used would be experimental studies focused on instructional methods aimed at changing attitudes toward sexist/nonsexist language (McMinn et al., 1994; Rubin et al., 1994). As recommended by Rubin and Greene (1991), such studies could be longitudinal, examining the long-term effects of modeling, teaching, or mandating the use of gender-inclusive language. Lastly, although one study (Rubin et al., 1994) found no relationship between subjects’ attitudes toward sexist language and their use of sexist language, additional studies of these variables are needed. Such studies could be conducted in a predictive format: That is, do attitudes measured on the IASNL predict language behaviors? These questions bring the study of sexist language back to the Sapir–Whorf hypothesis and the nature of the connections among language, attitudes, thoughts, and actions. We are hopeful the IASNL will be useful to scholars who wish to investigate these relationships.

APPENDIX. THE INVENTORY OF ATTITUDES TOWARD SEXIST/NONSEXIST LANGUAGE-GENERAL (IASNL-G) The Inventory of Attitudes Toward Sexist/Nonsexist Language-General (IASNL-G) is included here for use by any interested researchers. The IASNL-G should be used exactly as it appears on the following pages, including the presentation of the operational definition of sexist language on each page.

Scoring All items are scored on a 5-point Likert-type scale. Items 1, 2, 3, 4, 5, 8, and 20 are reverse-scored. High scores (4–5) indicate a positive attitude toward inclusive language; low scores (1–2) indicate a negative attitude toward inclusive language. A score of 3 indicates neutrality or uncertainty. Each respondent’s score on the instrument is the total of all the items.

Interpretation The range of possible total scores on the 21-item IASNL-G is 21–105. Across the 21 items, total scores between 73.6 and 105 reflect a supportive

434

Parks and Roberton

attitude toward nonsexist language; total scores between 21 and 52.5 reflect a negative attitude toward nonsexist language; and total scores between 52.6 and 73.5 reflect a neutral attitude.

Inventory of Attitudes Toward Sexist/Nonsexist Language-General

Please use the following definition in completing this questionnaire: Sexist language includes words, phrases, and expressions that unnecessarily differentiate between females and males or exclude, trivialize, or diminish either gender. SECTION I: For each of the following expressions, choose the descriptor that most closely corresponds with your beliefs about language. 1 ⴝ strongly disagree; 2 ⴝ tend to disagree; 3 ⴝ undecided; 4 ⴝ tend to agree; 5 ⴝ strongly agree 1. Women who think that being called a ‘‘chairman’’ is sexist are misinterpreting the word ‘‘chairman.’’ 2. We should not change the way the English language has traditionally been written and spoken. 3. Worrying about sexist language is a trivial activity. 4. If the original meaning of the word ‘‘he’’ was ‘‘person,’’ we should continue to use ‘‘he’’ to refer to both males and females today. 5. When people use the term ‘‘man and wife,’’ the expression is not sexist if the users don’t mean it to be. 6. The English language will never be changed because it is too deeply ingrained in the culture. 7. The elimination of sexist language is an important goal. 8. Most publication guidelines require newspaper writers to avoid using ethnic and racial slurs. So, these guidelines should also require writers to avoid sexist language. 9. Sexist language is related to sexist treatment of people in society. 10. When teachers talk about the history of the United States, they should change expressions, such as ‘‘our forefathers,’’ to expressions that include women. 11. Teachers who require students to use nonsexist language are unfairly forcing their political views upon their students. 12. Although change is difficult, we still should try to eliminate sexist language.

Attitudes Toward Sexist Language

435

SECTION II: Are the underlined words and phrases in the following sentences sexist? 1 ⴝ not at all sexist; 2 ⴝ probably not sexist; 3 ⴝ undecided; 4 ⴝ somewhat sexist; 5 ⴝ definitely sexist 13. 14. 15. 16.

People should care about all mankind, not just themselves. The belief that frogs will give you warts is just an old wives’ tale. If a child wants to play the piano well, he must practice hard. Alice Jones should be chairman of our committee.

SECTION III: Choose the descriptor that most closely describes you in the following situations. 1 ⴝ very unwilling; 2 ⴝ reluctant; 3 ⴝ undecided; 4 ⴝ somewhat willing; 5 ⴝ very willing 17. When you are referring to a married woman, how willing are you to use the title ‘‘Ms. Smith’’ rather than ‘‘Mrs. Smith’’? 18. How willing are you to use the word ‘‘server’’ rather than ‘‘waiter’’ or ‘‘waitress’’? 19. How willing are you to use the expression ‘‘husband and wife’’ rather than ‘‘man and wife’’? 20. How willing are you to use the term ‘‘camera operator’’ rather than ‘‘cameraman’’? 21. How willing are you to use the title ‘‘flight attendant’’ instead of ‘‘steward’’ or ‘‘stewardess’’? Reminder: Sexist language includes words, phrases, and expressions that unnecessarily differentiate between females and males or exclude, trivialize, or diminish either gender.

ACKNOWLEDGMENTS This study was supported by a grant from the Ohio Association for Health, Physical Education, Recreation and Dance and the Fund for the Study of Sport and Diversity, School of HMSLS, Bowling Green State University. We are grateful to the following individuals for their assistance with data collection: Kate Barrett (University of North CarolinaGreensboro), Nell Faucette (University of South Florida), Peg Nugent (National-Louis University), Brenda Pitts (University of Louisville), Linda Schoonmaker (University of Missouri-Columbia), Eugenia Scott (Butler

436

Parks and Roberton

University), and faculty and students in the School of HMSLS at Bowling Green State University (BGSU). We further acknowledge the valuable contributions of 31 advocates of inclusive language, 9 content experts, and Don Rubin (University of Georgia) and Kathryn Greene (East Carolina University). We are grateful for the help of Nancy Boudreau, Maria Hong, and John Tisak of the Statistical Consulting Center at BGSU. REFERENCES Acosta, R. V., & Carpenter, L. J. (1996). Women in intercollegiate sport: A longitudinal study—Nineteen year update, 1977–1996. Unpublished manuscript, Brooklyn College, Brooklyn, New York. American Psychological Association. (1983). Publication manual (3rd ed.). Washington, DC: Author. Ary, D., Jacobs, L. C., & Razavieh, A. (1990). Introduction to research in education (4th ed.). Chicago: Holt, Rinehart, and Winston. Bate, B. (1978). Changes in pronominal usage in a classroom situation. Psychology of Women Quarterly, 5, 773–779. Bing, J. (1992). Penguins can’t fly and women don’t count: Language and thought. Women and Language, 15(2), 11–14. Blaubergs, M. (1980). An analysis of classic arguments against changing sexist language. Women’s Studies International Quarterly, 3, 135–147. Bodine, A. (1975). Androcentrism in prescriptive grammar: Singular ‘‘they,’’ sex-indefinite ‘‘he,’’ and ‘‘he or she.’’ Language in Society, 4, 129–146. Bruce, T. (1998). Audience frustration and pleasure: Women viewers confront televised women’s basketball. Journal of Sport and Social Issues, 22, 373–397. Cameron, D. (1985). Feminism and linguistic theory. New York: St. Martin’s Press. Carroll, J. B. (Ed.). (1956). Language, thought, and reality: Selected writings of Benjamin Lee Whorf. Cambridge, MA: MIT Press. Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Cronin, C., & Jreisat, S. (1995). Effects of modeling on the use of nonsexist language among high school freshpersons and seniors. Sex Roles, 33, 819–830. Duncan, M. C., & Hasbrook, C. A. (1988). Denial of power in televised women’s sports. Sociology of Sport Journal, 5, 1–21. Eastman, S. T., & Billings, A. C. (1999). Gender parity in the Olympics: Hyping women athletes, favoring men athletes. Journal of Sport and Social Issues, 23, 140–170. Eitzen, D. S., & Sage, G. (1997). Sociology of North American sport (6th ed.). Madison, WI: Brown & Benchmark. Eitzen, D. S., & Zinn, M. B. (1989). The de-athleticization of women: The naming and gender marking of collegiate sports teams. Sociology of Sport Journal, 6, 362–370. Eitzen, D. S., & Zinn, M. B. (1993). The sexist naming of collegiate athletic teams and resistance to change. Journal of Sport and Social Issues, 17, 34–41. Gastil, J. (1990). Generic pronouns and sexist language: The oxymoronic character of masculine generics. Sex Roles, 23, 629–643. Hamilton, M. C. (1988). Using masculine generics: Does generic he increase male bias in the user’s imagery? Sex Roles, 19, 785–798. Harrigan, J. A., & Lucic, K. S. (1988). Attitudes about gender bias in language: A reevaluation. Sex Roles, 19, 129–140. Henley, N. M. (1989). Molehill or mountain? What we know and don’t know about sex bias in language. In M. Crawford & M. Gentry (Eds.), Gender and thought: Psychological perspectives (pp. 59–78). New York: Springer-Verlag.

Attitudes Toward Sexist Language

437

Henley, N. M., & Dragun, D. (1983, August). A survey of attitudes toward changing sex-biased language. Paper presented at the meeting of the American Psychological Association, Anaheim, California. Hyde, J. S. (1984). Children’s understanding of sexist language. Developmental Psychology, 20, 697–706. Jacobson, M. B., & Insko, W. R., Jr. (1985). Use of nonsexist pronouns as a function of one’s feminist orientation. Sex Roles, 13, 1–7. Kane, M. J., & Parks, J. B. (1992). The social construction of gender difference and hierarchy in sport journalism—Very few new twists on very old themes. Women in Sport and Physical Activity Journal, 1, 49–83. Khosroshahi, F. (1989). Penguins don’t care, but women do: A social identity analysis of a Whorfian problem. Language in Society, 18, 505–525. Kidd, V. (1971). A study of the images produced through the use of the male pronoun as the generic. Moments in Contemporary Rhetoric and Communication, 1, 25–30. Kline, P. (1994). An easy guide to factor analysis. London: Routledge. Krane, V., & Kane, M. J. (1998). Psychosocial aspects of sport and exercise. In J. B. Parks, B. R. K. Zanger, & J. Quarterman (Eds.), Contemporary sport management (pp. 33–47). Champaign, IL: Human Kinetics. Lakoff, R. T. (1973). Language and woman’s place. Language in Society, 2, 45–80. MacKay, D. G. (1980). Psychology, prescriptive grammar, and the pronoun problem. American Psychologist, 35, 444–449. Martyna, W. (1978). What does ‘‘he’’ mean? Use of the generic masculine. Journal of Communication, 28, 131–138. Matheson, K., & Kristiansen, C. M. (1987). The effect of sexist attitudes and social structure on the use of sex-biased pronouns. Journal of Social Psychology, 127, 395–398. McConnell, A. R., & Fazio, R. H. (1996). Women as men and people: Effects of gendermarked language. Personality and Social Psychology Bulletin, 22, 1004–1013. McMinn, M. R., Lindsay, S. F., Hannum, L. E., & Troyer, P. K. (1990). Does sexist language reflect personal characteristics? Sex Roles, 23, 389–396. McMinn, M. R., Troyer, P. K., Hannum, L. E., & Foster, J. D. (1991). Teaching nonsexist language to college students. Journal of Experimental Education, 59, 153–161. McMinn, M. R., Williams, P. E., & McMinn, L. C. (1994). Assessing recognition of sexist language: Development and use of the Gender-Specific Language Scale. Sex Roles, 31, 741–755. Merritt, R. D., & Kok, C. J. (1995). Attribution of gender to a gender-unspecified individual: An evaluation of the people ⫽ male hypothesis. Sex Roles, 33, 145–157. Messner, M. A., Duncan, M. C., & Jensen, K. (1993). Separating the men from the girls: The gendered language of televised sports. Gender & Society, 7, 121–137. Miller, C., & Swift, K. (1988). The handbook of nonsexist writing (2nd ed.). New York: Harper & Row. Moulton, J., Robinson, G. M., & Elias, C. (1978). Sex bias in language use: ‘‘Neutral’’ pronouns that aren’t. American Psychologist, 33, 1032–1036. Ng, S. H. (1990). Androcentric coding of man and his in memory by language users. Journal of Experimental Social Psychology, 26, 455–464. Nilsen, A. P. (1977). Sexism in children’s books and elementary teaching materials. In A. P. Nilsen, H. A. Bosmajian, H. L. Gershuny, & J. P. Stanley (Eds.), Sexism and language, Urbana, IL: National Council of Teachers of English. Nilsen, A. P. (1984). Greetings and salutations in a new age. Language and Society, 13, 245–247. Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill. Parks, J. B. (Executive Producer), Harper, M. C. (Script Writer), & Lopez, P. G. (Director). (1994). One person’s struggle with gender-biased language [Videotape]. Bowling Green, OH: WBGU-TV. Parks, J. B., & Roberton, M. A. (1998a). Contemporary arguments against nonsexist language: Blaubergs (1980) revisited. Sex Roles, 39, 445–461.

438

Parks and Roberton

Parks, J. B., & Roberton, M. A. (1998b). Influence of age, gender, and context on attitudes toward sexist/nonsexist language: Is sport a special case? Sex Roles, 38, 477–494. Richey, W. (1997, June 5). How some colleges score parity for women’s athletics. Christian Science Monitor, p. 3. Richmond, V. P., & Dyba, P. (1982). The roots of sexual stereotyping: The teacher as model. Communication Education, 31, 265–273. Rubin, D. L., & Greene, K. L. (1991). Effects of biological and psychological gender, age cohort, and interviewer gender on attitudes toward gender-inclusive/exclusive language. Sex Roles, 24, 391–412. Rubin, D. L., Greene, K., & Schneider, D. (1994). Adopting gender-inclusive language reforms: Diachronic and synchronic variation. Journal of Language and Social Psychology, 13(2), 91–114. Rust, J., & Golombok, S. (1989). Modern psychometrics: The science of psychological assessment. London: Routledge. Schneider, J. W., & Hacker, S. L. (1973). Sex role imagery and use of the generic ‘‘man’’ in introductory texts: A case in the sociology of sociology. American Sociologist, 8, 12–18. Shimanoff, S. B. (1977). Man ⫽ human: Empirical support for the Whorfian hypothesis. Bulletin: Women’s Studies in Communication, 1(2), 21–27. Simpson, P. (1993). Language, ideology, and point of view. New York: Routledge. Stericker, A. (1981). Does this ‘‘he or she’’ business really make a difference? The effect of masculine pronouns as generics on job attitudes. Sex Roles, 7, 637–641. Stevens, J. (1996). Applied multivariate statistics for the social sciences (3rd ed.). Mahwah, NJ: Erlbaum. Thorne, B., Kramarae, C., & Henley, N. (1983). Language, gender, and society: Opening a second decade of research (pp. 7–24). In B. Thorne, C. Kramarae, and N. Henley (Eds.), Language, gender and society, Boston: Heinle & Heinle. Woog, D. (1998). Jocks: True stories of America’s gay male athletes. Los Angeles: Alyson Books.

Suggest Documents