Does Language Type Affect Perceptions of Disability Images? An Experimental Study

The present experimental study examined the impact of language type on perception of disability images with text captions. 204 disability naïve undergraduate students viewed disability images containing one of six disability language captions: disability-first, defiant self-naming, impairment, negative, person-first, and apologetic naming. Participants completed measures of identification, emotion, willingness to help, willingness to include, and perceptions of capabilities and rights. Person-first and apologetic naming did not result in more positive perceptions of disability. Rather, defiant self-naming evoked the most positive emotions and identification, and greater perceived capabilities and willingness to include whereas negative language evoked the most negative perceptions of images. Results suggest that the elimination of negative language and the use of empowering defiant self-naming by people with disabilities, rather than a focus on using person-first and apologetic naming, may be more effective in reducing negative disability stereotypes.

and cultural level (Hall, 1997). Bilingual individuals report thinking differently and having a different sense of self when using separate languages (Pavlenko, 2006) and neuroimaging suggests language plays a role in decision-making (Tan et al., 2008). Since language influences behaviour, it is likely that certain types of language may result in greater willingness to donate money, hire, and befriend others, including PWDs.
Indeed, disability is one area where there is current and considerable effort towards changing both spoken and written discourse to more positive language, though English speaking countries do not necessarily agree on what type of language should be used. For example, in Canada, the term "Mental Retardation", which is still used by the World Health Organization, has been replaced with "Developmental Delay" or "Intellectual Disability". In the UK, the term "Learning Disability" is used in reference to the above group but also to individuals who have non-intellectual learning disabilities, such as dyslexia. In both the US and Canada, the term "Learning Disability" only applies to non-intellectual disabilities. It is unclear, however, whether practices such as these actually reduce disability stereotypes and result in greater inclusion of PWDs into broader society.
It is also unclear whether within a specific culture more positive terms successfully abate negative perceptions of people with disabilities (Dunn & Andrews, 2015). Proponents of positive terms contend that this effort represents an application of the Sapir-Whorf hypothesis where supposedly changing language may change one's world-view, resulting in reduced negative disability stereotypes and greater inclusion. Consequently, influential organizations such as the American Psychological Association (APA) advocate the use of person-first language when referring to individuals with disabilitiesallegedly to reduce bias in psychological discourse. The positive disability language used more recently has indeed been instrumental in providing a clear and unified message by broad disability rights movements. Terms such as "inclusion" in special education have become very powerful even though their meaning and application are often inconsistent (Norwich, 2002). Others, including disability groups, argue that changing language in itself is not substantial enough to change disability perceptions and rather more action, such as education, legislation, and enforcement is needed to advance full citizenship rights of PWDs. As a result, there is considerable debate as to whether international organizations should use the language of the United Nations Convention on the Rights of Persons with Disabilities (CRPD) to ensure that they use consistent disability language (Obosi 2010).
There are various types of disability language and these do not all belong to well defined linguistic categories (see Marx, 1999, p. 137). The only naming type not well defined is what we have coined "apologetic naming". This refers to a positive naming strategy that uses terms with positive connotations or terms that confine negativity to a difference or a challenge, such as "physically challenged". It may even suggest a denial of any inherently negative differences between PWDs and non-disabled persons, as it is based on the idea that everyone is different and everyone has strengths and challenges. A second positive naming strategy is "person-first naming", such as "individual with a physical disability". This language attempts to overcome the possibility that a person's disability receives more attention than his or her personhood. Apologetic and person-first terms have come to replace terms falling under both "disability-first naming", which categorises people based upon their disability identity (e.g., "the blind boy"); and impairment naming, which identifies the type of impairment, such as "physically impaired". Impairment naming assumes a negative medical model approach which focuses on an individual's deficits and how he or she can be "fixed" to overcome them.
Arguably, and non-controversially, "negative naming", which uses offensive terms (e.g., "mad", "crippled" or "deaf and dumb"), should be removed from discourse as such language can only result in negative stereotypes leading to exclusion. A final form of naming is "defiant self-naming", which makes ironic but empowering use of offensive terminology (e.g., "super-crip") by people with disabilities about themselves, whereas the general public may not use these terms lest they be considered ableist.
There is some research exploring public opinion on disability language type. Lynch and Groombridge (1994) found that a third of 300 US state government employees considered person-first and disability-first language equivalent and were uninfluenced by language type when choosing descriptors of hypothetical job applicants. Nevertheless, the majority of the sample preferred person-first naming. Rottenstein (2014) found that 70% of 3000 persons with disabilities preferred to describe themselves as a "person with a disability" compared with only 8% who chose "disabled person". Clearly there are differences in society, often based on political orientation, as well as differences within the disability community as to the benefits of politically correct disability language. A first goal of the present study was to determine whether indeed different language captions that accompany charity images determine perceptions of PWDs by the general public.

Type of Disability and Language Type
Research has shown that people view disability type hierarchically and that the order of disability types, from most to least "acceptable", has remained the same since the 1960's: from limited physical, to auditory, visual, paraplegic, mental, and quadriplegic (Olkin & Howson, 1994;Westbrook et al., 1993;Horne & Ricciardo, 1988;Abroms & Kodera, 1978;Goodman, Hastorf, & Richardson, 1967). Studies also show that when imagining themselves as employers, university students are more willing to hire persons with physical rather than mental disabilities (Gouvier, Systsma-Jordan, & Mayville, 2003;Drehmer & Bordieri, 1985;Stone & Sawatzki, 1980). Another study has shown that the more visible a person's disability, the less likely he or she will be chosen for a job requiring more contact with customers (Gouvier et al., 1991).
This hierarchy is likely determined by perceptions and stereotypes about different disabilities created by the media and charity advertisements as discussed above. Complicating this issue is the fact that not all disability groups have embraced more recent attempts to use more positive language in a manner consistent with the unified message provided by disability rights groups: Members of the Deaf community refer to themselves with a capital "D" as Deaf individuals, and do not endorse person-first language to identify themselves (Tyler, 1993). The National Federation of the Blind rejected www.scholink.org/ojs/index.php/jetss Journal of Education, Teaching and Social Studies Vol. 2, No. 4, 2020 80 Published by SCHOLINK INC.
person-first language in 1993 because they thought it presumed an apology should be made for being blind (Bickford, 2004). The Autism community consider Autism to be part of their personhood, rather than a separate disability and therefore are not opposed to being defined by their disability (What to say 2004). It is possible that the use of different types of language by different disability communities, whether due to differences in historical terms or endorsement of more positive person-first language, may result in less favourable perceptions of particular disability groups.
A second goal of this study was to determine whether disability type and language type interact with one another to produce especially positive and/or negative perceptions for particular combinations of disability and language types when used in charity advertisements.

Gender and Language Type
Men and women use language differently though a clear pattern has yet to emerge (Newman, Groom, Handelman, & Pennebaker, 2008). As such, it is of great interest whether men and women perceive disability language captions differently and if so whether they are similarly affected by differences in language types. It is also of great interest whether the gender depicted in a disability image interacts with type of language caption used to produce different perceptions of people with disabilities. Before reviewing the literature on gender differences in language use a few words about gender differences are noteworthy.
A key relevant area to the present study where gender differences are found is in prosocial behaviour.
In one review, Eagly (2009) concluded that while both sexes are similar in their extensive engagement in prosocial behaviour, female emphasis is more communal and relational whereas male emphasis is more power-oriented. Eagly argues that this difference may be routed in the division of labour between the sexes. While modern families are often egalitarian and both parents work outside of the house, women still do the majority of housework (Lachance-Grzela & Bouchard, 2010).
With respect to the perceived, the literature is less conclusive. Within the context of disability images, women with disabilities may be viewed more negatively (Fine & Asch, 1981) as a result of the compounding of negative stereotypes of both sex and disability. Women are often seen as passive and in need of protection, as are PWDs (Marks, 1999). However, since men are expected to be strong and self-sufficient, if they are not seen as able to fulfil masculine sex-roles prevalent in most cultures, it is possible that they may be subject to more negative evaluations than women. Indeed, Kamenetsky et al. (2016) found that images (without text captions) of men with disabilities elicited greater sadness, anger, disgust, guilt and lower perceptions of capabilities and rights but also, and perhaps consequently, a greater willingness to help.
Numerous studies documented that men and women use language differently. While there are no gender differences in verbal ability (Hyde & Linn, 1988) a seminal text on gender difference in language (see Coates, 2015) discusses the idea of gender as a social construct, and addresses gender differences in conversational practice, same-sex talk, conversational dominance, and children's acquisition of gender-specific language among other topics. Early studies have shown that women tend to be more expressive, supportive, polite, talk more about their families, and use more affective words and tend to discuss psychological states more frequently compared with men (Haas, 1979). Males' language tends to be more direct, succinct, personal, and instrumental, whereas women's language tends to be more indirect, elaborate and affective (Mulac, Bradac, & Gibbons, 2001). Newman et al. (2008), however, argue that at least some scholars contend that there are no meaningful gender differences in language. They cite Bradley (1981) and Weatherall (2002) and suggest that the lack of agreed upon measures and too broad generalizations based on small text samples may be responsible for some of the gender differences obtained. Furthermore, if indeed gender is a social construct that fosters gender differences in language one might expect that a reduction in gender role differences seen in Western, secular and egalitarian millennials (Ng & McGinnis Johnson, 2015) would bring with it reduced gender differences in language.
A third goal of this study was to determine whether sex and language type interact with one another to produce especially positive and/or negative perceptions when used in disability charity advertisements.

Measuring Perceptions of Persons with Disabilities
There are many ways to assess how individuals respond to images of PWDs. Broadly speaking, these may include identification with individuals depicted, emotional reactions to the images, willingness to help and include such individuals, perceptions of their capabilities, as well as whether they are entitled to the same rights accorded to the rest of society. The literature on these constructs has recently been reviewed by Kamenetsky et al. (2016). The following section presents a brief summary.
Identification is a social phenomenon whereby one self-categorizes oneself by affiliation with others (Tajfel, 1982). Social ties and identification are highly related to the possession of similar physical, demographic, and behavioural traits (Kossinets & Watts, 2009;Hoffner & Buchanan, 2005).
Identifying with groups can create an in-group bias where members tend to view their group more favourably than out-groups (Tajfel & Turner, 2004); similarly identifying with persons is associated with a higher likelihood of directing helping behaviour toward them (Fisher & Wakefield, 1998).
Emotions are highly influenced by subjective appraisal and cognition (Bennett & Lowe, 2008;Lazarus, 1991). When the stimulus is a person or an animal, perspective-taking can result in strong emotions (Neumann & Strack, 2000). Images of sadness can thus result in viewer sadness. Helping behaviour can take the form of donation and volunteerism. Familiarity with a charity as well as media depictions of an issue and of the beneficiaries influence the decision to donate (Bendapudi, Singh, & Bendapudi, 1996). Positive attitudes toward disability have been shown to correlate with helping behaviour directed toward them (Carter, Hughes, Copeland, & Breen, 2001;Lauber, Nordt, Falcato, & Rössler, 2002). Negative attitudes, however, caused by images of sadness and need, may mediate helping behaviour that decreases the negative attitudes of the viewer (Schmidt & Weiner, 1988). Social inclusion, defined by Lemay (2006) as societal involvement and acknowledgement of marginalized groups as valued members, can be indicated by the willingness to live near, recognize, befriend and employ persons from said groups (Harth, 1971). In fact, PWDs have rated "being employed" as one of the most important factors of social inclusion (Hall & Kramer, 2009). Nevertheless studies show they are disproportionately under-or unemployed (Burkhauser & Stapleton, 2004;Dyda, 2008;Levy & Hernandez, 2009;Verdonschot, de Witte, Reichrath, Buntinx, & Curfs, 2009); and that employers tend to hold negative stereotypes about them (Fraser et al., 2010). Less favourable perceptions of capabilities are held not only by employers in relation to worker potential, but are also held by the public in relation to success, fortune, partnering, parenthood and friendship (Buljevac, Majdak, & Leutar, 2012). Even clinicians tend to make wrong assumptions of disabled patients' abilities as well as devalue their intelligence and motivation (Iezzoni, 2006). Human or equal rights in its most basic definition tends to include the right to equality and dignity, and to live free from all forms of discrimination (Canadian Human Rights Commission). Research has documented that less favourable perceptions of equal rights prevailed when persons with disabilities received inequitable treatment in education and public services (O'Keeffe, 1993). However, Kamenetsky et al. (2016) provided clear evidence that this may be changing as Canadian first year university students provided very high ratings of equal rights for a broad range of images of PWDs.

Purpose
The purpose of the present study was to experimentally determine whether type of language caption included in charity images determines identification with, emotional responses towards, willingness to help and include, perceived capabilities, and perception of equal rights of the PWD depicted in an image. We were also interested in determining whether sex of the observer and both sex and disability type of the observed interact with language type in determining the above perceptions.

Method
A mixed design was used with viewer sex (male, female) as the between-subjects factor and language (disability-first, defiant self-naming, impairment, negative, person-first, apologetic), disability (physical, vision, hearing) and sex of depicted individuals (male, female) as within-subject factors. Table 1 displays the experimental design. Order represents the order in which images 1-18 were coupled with the six language types. Each participant was presented with one order only.

Participants
Participants were 218 first-year undergraduate students enrolled in a first-year psychology course.
Eleven who had a disability diagnosis were excluded to ensure the sample was naï ve to the study. This was important to ensure that participants' disability perceptions came primarily from the media and not through their own personal experiences. Three for whom either survey or questionnaire data was missing were also excluded. This resulted in a final sample of 204 participants (160 female, 44 male), ranging in age from 16 to 24 (M = 18.5, SD = 1.1). Low male participant numbers are prevalent in this line of research (Clay, 2017). The number of male participants, however, exceeded the general rule of thumb (n>30) regarding the central limit theorem, which implies that a representative sample mean is close to the true mean (Hogg & Tanis, 2009 whether any of their siblings, friends' siblings, friends, and neighbours has a disability, and if so, how many. Total disability experience was calculated by summing these four categories and ranged from 0 to 19 (M = 1.67, SD = 2.45). The majority of participants had minimal disability experience. We were especially interested in recruiting participants who were naï ve to disability language issues so that we could determine whether differences in disability language use have any impact on the "average" young adult.

Materials
Test stimuli included 18 images depicting three males and three females for each of the three disability types (see Figure 2 for example, actual images are withheld for confidentiality). They were picked using the world-wide-web in terms of sex, disability, size, eye direction and age. Demographic    Table 2 displays the disability naming captions, which included 18 descriptive phrases (six language types and within these one for each of the three disabilities). These were presented concurrently below the images. Phrases were written based on the terms used by each language type (defiant self-naming, impairment, apologetic, disability-first, person-first, and negative). Defiant self-naming used prideful and ironic sentences; impairment language used adjective-noun combinations; apologetic naming referred to the disability as a challenge; disability-first presented the descriptor before personhood;

Captions
person-first language presented personhood before the descriptor (once in a simple phrase, and a second time in a following phrase that was more specific in relation to sex and degree of disability); and negative language type used phrases or terms with negative connotations.

Questionnaire
The 19-item test stimuli questionnaire by Kamenetsky et al. (2016) was used. Viewers rated their responses on a scale from 1 (not at all) to 7 (great extent) toward questions relating to six constructs.  Hire "To what extent would you be willing to hire this person to work for you?" Perceptions of capabilities Make friends "Do you think this person has the capability of making friends?" Self-care "Do you think this person has the capability to take care of him/herself?" Get married/child care "Do you think this person has the capability to look after children and/or get married?" Get hired "Do you think this person can get hired to work?" Perceptions of rights Equal rights "Do you think this person has the same right as everyone else?"

Procedure
Viewers individually entered a research room on campus to participate in the study, which they were told was about attitudes towards people with disabilities. Viewers were seated, and upon providing informed consent, completed a demographics survey displayed on a 23-inch high-resolution computer screen. They then viewed and responded to the images presented in one of the six experimental orders in a single 1-hour session. Disability images accompanied by captions were presented on the screen in random order, with each image and its caption being followed by the 19-item questionnaire. The questionnaire was presented one question at a time in the same order and in clear font below each image and its caption. A response using the numerical keypad of a standard keyboard was required before continuing onto the next question. Once viewers completed the entire questionnaire for each of the 18 images, they were debriefed and thanked for participating.

Results
Statistical computing software SAS was used to import and combine individual files of each participant into a master file which included all demographic information, experimental order (1-6), and 18 responses to each of 18 images. Overall means are presented first, followed by main effects and interactions between language and disability type. The latter were obtained from a series of ANOVAs for each of the 19 responses. Effects of sex and its interactions with other factors were also explored using all possible 2-factor ANOVAs (5 series). Post hoc comparisons were done using Tukey's Honest Significant Difference (HSD) tests (Tukey 1949 tests from 19 response variables x 6 series of ANOVAs). We found no significant main effects for sex or interactions between sex of person depicted and language type, as well as between sex of viewer and language type. Only statistically significant results are reported. Table 4 shows the overall means and standard deviations for the response variables. Emotions among viewers were low except for happiness and sadness, which were moderate. Identification was also low.
Willingness to help and include, as well as perceptions of capabilities and rights, were all high with perception of rights being the highest. This pattern replicates previous findings by Kamenetsky et al.
(2016) using different participants and different images -this time accompanied by text captions. A series of 6 (language: disability-first vs. defiant-self naming vs. impairment vs. negative vs. person-first vs. apologetic) x 3 (disability: hearing impairment vs. physical impairment vs. visual impairment) ANOVAs were conducted. For each category of response variables (emotions, identification and willingness, and perceptions), we address the main effects of language type, followed by disability type, and the interactions between them.
A language main effect was observed for all emotions but fear and contempt.

Table 5. Main Effects of Language Type for Emotions
Note. Boldfaced means are significantly different than all means in the same column.
A language main effect was also observed with respect to identification and willingness to help and include (see Table 6). DSN compared to other language types elicited higher identification (2.13 vs.

Table 6. Main Effects of Language Type for Identification and Willingness to Help and Include
Note. Boldfaced means are significantly different than all means in the same column.
A language main effect was also observed with respect to perceptions of capabilities and equal rights (see Table 7). DSN elicited higher perceived capability for self-care (5.66) than disability-first (5.41) and person-first (5.21); get married/child care (5.55) than disability-first (5.29) and person-first (5.17); and get hired (5.26) than person-first (4.96). Person-first compared to impairment naming elicited lower perceived capabilities for self-care (5.21 vs. 5.50), get married/child care (5.17 vs. 5.43), and get hired (4.96 vs. 5.23); and when compared to apologetic naming, lower perceived capability to get married/child care (5.17 vs. 5.42). Negative naming compared to almost all language types elicited the lowest perceived capabilities (e.g., make friends with 5.79 vs. 5.99 and higher). There was no disability type main effect for emotion or identification. There was, however, for willingness to help and include. Hearing impairment elicited lower willingness to volunteer (M = 5.76, F(5,203) = 15.9, Tukey's HSD = 0.13) than both physical (5.90) and visual impairment (5.93). Vision impairment elicited lower willingness to hire (M = 4.39, F(5,203) = 69.6, Tukey's HSD = 0.17) than both physical (4.77) and hearing impairment (4.91).
There was a disability type main effect for perceptions of capabilities and equal rights. Hearing impairment elicited greater perceived capabilities (for self-care, get married/child care, get hired) and equal rights than both physical and visual impairment (see Table 8 for bolded means); and lower perceived capability to make friends (5.93) compared with physical impairment (6.10). Visual impairment elicited lower perceived capabilities for get married/child care (4.98) than physical (5.19) and hearing impairment (5.78); and for get hired (4.73) than physical (5.05) and hearing impairment (5.44).

Table 8. Main Effects of Disability Type on Perceptions of Capabilities and Equal Rights
Note. Boldfaced means are significantly different than all means in the same column.

Language Types
It is clear that when people are referred to in negative terms others will perceive them more negatively.
Indeed terms such as "mad" and "dumb" have been removed from discourse about PWDs and this study confirms that such removal reduces negative perceptions of PWDs. We show that other language types as compared to negative language elicit greater perceived capabilities of PWDs, greater willingness to hire them and less negative emotions in those who perceive PWDs. What is even more interesting is the fact that more positive language, such as person-first naming, designed to replace negative terms, has not emerged as a superior way to communicate about PWDs since, for the most part, it is not associated with more positive perceptions of PWDs. What is most fascinating is that at least some disability groups never really believed that positive language would be useful in reducing stereotypes and promoting inclusion and in some cases even felt that it may be detrimental to their cause (Deafness, Tyler, 1993;Visual Impairment, Bickford, 2004;and Autism, What to say 2004). It is often broader groups, such as the American Psychological Association and progressive political organizations that promote the use of such language.
Especially important is the finding that defiant self-naming (DSN) captioning produces the most positive perceptions of people with disabilities, including the greatest identification and greatest willingness to include (befriend and hire). This result is actually not surprising. DSN implies that PWDs feel empowered, confident and have self-determination. In a large meta-analysis Algozzine, Browder, Karvonen, Test, and Wood (2001) found that self-determination can be taught and learned, and can make a difference in the lives of individuals with disabilities. Further, people who experience positive emotions are more likely to flourish (Fredrickson, 2001) and subjective well-being and empowerment are highly correlated (Diener & Biswas-Diener, 2005). It is quite clear, therefore, that empowering PWDs, encouraging them to use DSN and having others refer to PWDs unapologetically and with no pity is a more promising strategy for inclusion and reducing negative stereotypes compared with other language types. It appears, therefore, that it is society's duty to remove negative language and to empower PWDs so that they feel proud of who they are and as a result communicate this confidently. Other attempts, such as using person-first naming, appear to be less likely to make a difference in how PWDs are perceived.
But it's quite clear from the results of the present study that differences between language types in all perceptions of PWD are relatively small. As it turns out, perceptions are difficult to change despite the use of positive naming strategies; and identification with PWDs remains very low regardless of language type. Indeed, Kamenetsky et al. (2016) found the same general pattern of responses including low ratings of identification in a similarly multicultural sample when no language captions were provided at all. It is possible that as Peters (2010) and others have proposed -there is a "disability culture" and the mainstream tends to consider its "members" as "the other" in the same way they may consider individuals from another racial or religious group as very different from themselves -

Interactions between Disability Types and Language Types
It appears that people with hearing impairments, in general, are perceived as more capable and thus in less need of support and that those with visual impairments are perceived as least capable. It is well known that stereotyped perceptions are related to the type of disability someone has (Yuker, 1994).
These stereotypes are problematic and do not necessarily align with reality. They likely represent the general public's perception that not seeing is "worse than" not hearing (Owoeye, Ologe, & Akande, 2007). Indeed, previous research has shown that people believe that individuals with visual impairments are less capable for work than those who have physical amputations (Strohmer, Grand, & Purcell, 1984); that employers believe that it is harder for a person with a visual impairment to gain employment than someone who has diabetes, thalassemia or renal insufficiency (Zissi, Rontos, Papageorgiou, Pierrakou & Chtouris 2007); and that 53% of respondents thought that individuals who are visually impaired could do as well a job as people without disabilities, compared with 58% in reference to individuals who have hearing impairments and 68% for people who have paralyzed legs (Bowman, 1987). It appears that such perceptions are resistant to change despite progressive legislation, language, and education policy designed to eradicate such stereotypes. In reality, of course, whether people with certain types of disabilities compared to others are more needy or less able to work is much more complex and depends on the etiology, the nature and severity of the disability, the type of job in question, and the extent to which assistive technological devices are provided by employers to enable individuals with various types of disabilities to do their jobs.
The interactions between language and disability types should not draw any particular conclusions other than the fact that they exist. Notable findings were that the use of DSN (i.e., Because I can't see you, I don't judge you superficially, I judge based on what I see inside you) for individuals who have visual impairments resulted in a large spike in identification. The use of negative language (i.e., Dumb/Mute) for individuals who have hearing impairments resulted in a sharp decrease in perceptions of ability to make friends, get hired, get married and willingness of the perceived to hire such individuals. We propose that the particular sentence for DSN (equivalent to "I don't judge a book by its cover") may have resonated with our viewers and resulted in high identification; and that the combination of negative language with hearing impairment yielded more negative perceptions due to the double-meaning and offensiveness of the word "Dumb" (as compared with "Blind as a bat" and "Crippled" for visual and physical impairments, respectively). The difference in words used is strikingly apparent for negative hearing language compared with, for example, disability-first (Hearing: "The Deaf"; Vision: "The Blind"; and Physical: "The Handicapped"). While this interaction could be the result of the specific words used, type of language used to describe different disabilities varies and is inconsistent. Clearly, however, this finding demonstrates the degree to which mere words can negatively affect our perceptions of specific disability types and with this the need to focus on the consequences of negative language separately for different disabilities.

Interactions between Sex of Viewer/Depicted and Language Types
The fact that such interactions were not obtained suggests that both males and females are similarly affected (or unaffected) by language type used in captions of charity advertisements. The effect of language type used is also similar regardless of whether males or females are depicted in charity images. The lack of findings in this area is consistent with reduced gender role differences (Ng & McGinnis Johnson, 2015) among the current egalitarian cohort of millennials (most participants were 18-year-old first-year university students) and supports the notion that gender differences in language are socially constructed within the context of disability charity advertisements.

Limitations
One limitation of this study is that linguistically, the six types of captions used are different from one Images of people with physical disabilities depicted individuals looking directly at the camera. As such, the images of people with physical disabilities were the only images in which direct eye contact with the viewer was possible. Because attitudes toward PWDs seem to be influenced to a great extent by appearance and social skills (Yuker, 1994), it is possible that the lack of direct eye contact for people who have hearing or visual impairments may have influenced our results. Indeed, eye contact has been shown to be a positive determinant in the decision to hire (Amalfitano & Kalt, 1977), and yet the hearing disability images in this study elicited a greater willingness to hire and perception of capabilities.
A possible explanation exists: one study has shown that eye contact with negative verbal content produces negative evaluation (Ellsworth & Carlsmith, 1968); our disability captions could be considered to be verbal content (especially in the case of DSN, the captions of which are actual quotes); and were likely viewed by our participants as negative due to the negative connotations of and stereotypes toward disability. Thus the lack of eye contact in images of people with hearing impairments may have resulted in less negative evaluations than those for images of people with physical or visual impairments. Perhaps as such, images of people with hearing impairments did not elicit significantly more negative perceptions.

Implications
The important implications for constituents of society with theoretical and/or practical interest in disability language use (sociolinguists, professionals, disability advocates, charitable organizations and the media) are the following: 1. The use of positive language (i.e., person-first and apologetic naming) is not likely to produce the utopian barrier-free and inclusive society we all desire. Rather, the elimination of negative language coupled with the empowerment of PWDs to communicate in a confident manner displaying high self-esteem and pride in one's identity (i.e., by using DSN) are more likely to create a more inclusive society.

2.
Most of the differences in perception and attitudes due to language use, albeit statistically significant, are of small magnitude. That being said, such small differences may have practical consequences over many interactions with PWDs. It is ironic that person-first language, a type of language designed to have practical consequences and adopted by influential organizations like the American Psychological Association, in fact lowers perceived capabilities compared to the language types it was intended to replace (impairment and disability-first).

3.
Changing public opinion on disability and other important societal issues is a slow process.
Setting long-term goals and devising methods, including language use plans, based on sound empirical science are more likely to produce the desired changes compared with idealistic approaches that have not been tested.
In conclusion, our results suggest that positive disability language may not deserve the "positive" appellation as neither person-first nor apologetic naming result in any substantial increase in perceptions or attitudes toward PWDs when compared with other forms of naming. Negative language, however, evokes comparably more negative emotions, less willingness to hire, and lower perceptions of capabilities and should be removed. DSN is best at evoking positive emotions and identification, while producing greater perceptions of capabilities and willingness to help and include. But overall, type of language used has little impact on the perception of PWDs by the general public. Further studies could incorporate one or more of many variables (e.g., ethnicity, skin colour, and sexual orientation) to determine if and how they may interact with language type in determining perceptions of charity images depicting people with disabilities. Such knowledge may contribute to the development of more effective methods of communication about disability issues -methods that promote support and inclusion rather than pity.