Learning Computerese

Abstract

This article introduces a new construct coined as Computer User Learning Aptitude (CULA). To establish construct validity, CULA is embedded in a nomological network that extends the technology acceptance model (TAM). Specifically, CULA is posited to affect perceived usefulness and perceived ease of use, the two underlying TAM constructs. Furthermore, we examine several antecedents of CULA by relying on the second language learning literature. These include computer anxiety, tolerance of ambiguity, and risk taking. Conceptualization of CULA is based on the observation that computer systems use language as communication between the computer and the user, making system usage significantly dependent on the ability of the individual to learn the language. We posit that learning to communicate with computer technology is akin to learning a second language, that is, a language learned after the first language(s) or native language(s), and is referred to as computerese. The proposed construct, CULA, measures the aptitude of an individual to learn computerese, and it is specified as a second-order variable. It includes measures of three critical facets of computerese pertaining to general hardware/software, programming, and the Internet. Significant relationships are found between computer anxiety, tolerance of ambiguity, and taking risk with CULA, as well as between CULA and TAM constructs.

Keywords

computer user learning aptitude second language learning tolerance of ambiguity computer anxiety perceived usefulness perceived ease of use

Introduction

Using computers involves a language for communication, such as the command language used for directing computer functions and technical jargon sometimes referred to as computerese (Johnson, 1983), each having its own syntax and vocabulary. While interfaces have moved from more rudimentary beginnings such as DOS-based systems to GUI (graphic user interface)-based systems and some computer vocabulary became mainstream jargon, the seemingly never-ending changes to the computer interface languages persist as new program versions and new processes are introduced. For example, to import a file into a spreadsheet an understanding of the term delimited is typically needed. Is the file comma delimited or tab delimited? What kind of file extension is it—txt, doc, docx, cvs, xml, html? The user must learn the vocabulary and the syntax for entering the correct sequence of commands to successfully complete the task. Universal usability is a commendable and desirable goal (Schneiderman & Hochheiser, 2001); however, frustration with issues that involve language, such as unclear error messages and unusable features, continue to affect user emotion and productivity (Lazar, Jones, & Shneiderman, 2006). Language is clearly a part of the process of using a computer system; however, little research has been conducted to examine whether a user’s aptitude to learn a pertinent language relates to system use.

Much of the extant research related to system use relies on the technology acceptance model (TAM; Davis, Bagozzi, & Warshaw, 1989) and its extensions, which frequently include what are termed external variables (EVs), such as system characteristics or user characteristics. Within the realm of TAM, the effects of the EVs on behavioral intentions are mediated by two cognitive constructs representing internal beliefs and are identified as perceived usefulness (PU) and perceived ease of use (PEOU; Davis et al., 1989). Thus, the EVs “provide the bridge between internal beliefs and intentions represented in TAM and the various individual differences, situational constraints and managerially controllable interventions impinging on behavior” (Davis et al., 1989, p. 988). Davis et al. (1989) advocated that external factors such as user characteristics be the foci in future research, supporting ongoing investigations such as this study.

The relationships between the TAM constructs–system usage (SU) explained by measuring behavioral intentions (BI), and BI explained by PU and PEOU, have been well researched and documented (e.g., Shih, 2006). However, researchers continue to specify and test alternative EVs, as advocated by Davis et al. (1989), in the hope of developing a more robust model focusing on contexts such as eCRM (Wu & Wu, 2005), eLearning (Martinez-Torres et al., 2008), and explanatory variables such as computer self-efficacy (Shih, 2006), trust, fairness (Chiu, Lin, & Hsu, 2009), and awareness (Van Schaik, Radford, & Hogg, 2010). Interestingly, with some notable exceptions (e.g., Agarwal & Karahanna, 2000), there is a paucity of studies that deal with the different aspects of the perceptions inherent in the key constructs of PU and PEOU. The concept of perception has been typically treated as a “black box,” as if all users have the ability to interpret and understand the technology equally, even though differences in cognitive abilities are recognized. “Users differ in their . . . cognitive abilities . . . these differences in . . . cognitive abilities can be expected to play an important role in influencing . . . usage” (Deng, Doll, & Troung, 2004, p. 399).

Although this perceptual process is central to TAM, being recognized as the cognitive schema leading to the assessment of usefulness and ease of use, little is known about whether individual perception differences might act as a bridge or whether they might act as a barrier in explaining usefulness and ease of use. We argue that a better understanding of the cognitive process is vitally important as we try to deal with the ever-evolving computer technology and the effects of these changes on end users’ technology acceptance. Specifically, we posit that differences in aptitude to learn computerese (i.e., CULA) contribute significantly toward PU and PEOU. Computerese has been defined as “the specialized vocabulary and jargon used by people who work with computers; the system of linguistic signs or symbols considered in the abstract” (Computerese, 1994, p. 806). We expand on that definition by including vocabulary and jargon used in reference to or working with computer technology, such as any command language or interface needed to complete a computer-based task.

Computer user learning aptitude derives its theoretical basis from Bandura’s (1986) work, which recognized that the basis for cognitive competencies is promoted by psychological and social processes, the educational learning theory, and the literature on the success in secondary language learning. It also relies on the multiple intelligences educational learning theory including a proposal for a digital intelligence by Antonio Battro (Gardner, 1983, 2011). We postulate that individuals who have the aptitude to learn computerese will have a greater propensity to perceive the technology as being both easier to use and useful compared with those individuals who have difficulties learning computerese. We further posit that learning computerese is analogous to learning a second language, since any computer technology has its own vocabulary. That is a language that must be understandable prior to an individual determining whether the technology is useful or easy to use.

This study focuses on the development of the proposed Computer User Learning Aptitude (CULA) construct within the context of the TAM. The rest of the article is arranged as follows. In the next section, we present the theoretical foundations for the proposed CULA construct and argue for its relevancy to TAM constructs. Then, the measurement model is examined followed by an assessment of the aforementioned nomological network via a structural model.

Theoretical Background

Two prevalent constructs in the technology acceptance literature, PEOU and PU, are considered as “self-reported indicants of system usage” (Davis, 1989, p. 320). PEOU is defined as “the degree to which a person believes that using a particular system would be free from effort,” and PU is defined as “the degree to which a person believes that using a particular system would enhance his or her job performance” (Davis, 1989, p. 320). Many studies have examined antecedents of PEOU and PU, such as trust (Chiu et al., 2009; Gefen, Karahanna, & Straub, 2003), or integrate theories, such as the flow theory (Koufaris, 2002). However, aside from a few notable exceptions, such as Agarwal and Karahanna’s (2000) extension of TAM using cognitive absorption, few have paid attention to important individual cognitive differences. However, recognizing cognitive differences is nothing new.

Instructors of commercial and academic computer classes are well aware of the variations in student ability. . . . One student may learn . . . with relative ease, while another student with similar background and education struggles with the same material. (Evans & Simkin, 1989, p. 1322)

With an ever-increasing reliance on technology in every aspect of business, government, and personal life, the importance of understanding technology literacy and technology acceptance grows. Lack of competencies related to technology is seen as a true handicap.

Arguing that end-user learning of a new computer technology is a cognitive process and similar to second language learning, we turn to the literature on learning, particularly second language learning, to identify the factors that explain individual differences in learning ability.

Computer User Learning Aptitude

Aptitude as related to computer technology has typically been studied from the perspective of mathematical proficiency (Howell, Vincent, & Gay, 1967; Konvalina, Woleman, & Stephens, 1983), while aptitude in a learning context has been identified as a much broader and general concept and is

a present concern in virtually all goal-oriented human activities; whenever one thinks about the antecedents of observed individual differences in some valued goal attainment, some concept of aptitude is needed. . . . Aptitudes are initial states of persons that influence later developments . . . that are not merely correlates of learning, but rather propaedeutic to (i.e. needed as preparation for) learning in the particular situation at hand. (Snow, 1992, p. 6)

Snow (1992) linked valued goal attainment with individual differences in aptitude and learning ability. Learning ability manifests as a later development reflecting a user’s initial state (aptitude). Here, we argue that aptitude, as related to information technology, should be viewed from the broader perspective as propaeduetic to learning, particularly in the context of new technology where the ability or aptitude to learn will influence perceptions about the usefulness and ease of use of the introduced technology.

Aptitude is an individual difference, such that each individual possesses his/her own level of aptitude. The role of individual differences is considered important for learning where “Psychologically, aptitude is whatever makes a person ready to learn rapidly in a particular situation (or, more generally to make effective use of a particular environment)” (Cronbach & Snow, 1977, p. 107). Thus, the process of learning is considered to be based to a large degree on individual differences in the respective aptitude. We argue that CULA reflects differences in individual learning aptitudes and as such should not be taken for granted as being homogeneous. Users have individual aptitudes and, thus, learn differentially. A consequence is that users that lack the ability to learn the necessary language may not be able to see how easy it is to use the system, let alone appreciate the usefulness of the system.

Learning about computer technology results in computer literacy, an extension of traditional literacy, which has become increasingly important as competitive pressures highlight the fact that businesses can no longer afford managers who are technologically illiterate. However, the term traditional literacy is not limited to those who can read (called literates) from those who cannot (called illiterates). A theoretically rich view of literacy defines it as a continuous, multidimensional indicator of proficiency in using written language, with its higher levels reflecting an ability to draw logical inferences and think critically (Wallendorf, 2001). Applying this definition, computer literacy has a range of competence, with the higher range reflecting a more elevated ability than the lower range. Cognitive ability, that is, aptitude, and learning are again linked.

Computer literacy has been approached as a humanized communication process exemplified by terminology used in the literature from the earliest use of computers. Examples include the development of different dialogue modes through an adaptable user interface (Kantorowitz & Sudarsky, 1989). In other words, the communication process is mediated through the use of computerese, that is, a language used to communicate or interact with a computer. While computerese is not a natural language, it does fall within the realm of a number of language definitions much alike the American Sign Language. It “is expression in words, . . . a system of symbols” (Raschen, 1928). Thus, as an artificial or constructed language it supports man–machine communication just as programming languages support such communication. “A language is a social tool; . . . A real language is characterized by the practical use within a certain group of people” (Zemanek, 1966, p. 112). Computerese is practiced by a variety of groups of people with some groups, such as the information technology professionals, being more fluent than other groups, such as novice users.

We argue that learning computerese is analogous to learning a second language by the user rather than the first or native language(s). The distinction between first and second language learning is based on the psychological or cognitive approaches to language learning that view the differences as based on general cognitive maturity of second language learners relative to first language learners. First language learners are acquiring knowledge of language and knowledge of the world simultaneously, whereas second language learners already have knowledge of their world before trying to learn to communicate with a different set of symbols and vocabulary (Towell & Hawkins, 1994). Aptitude may then be construed as a predictor of literacy, as literacy is the specific focus of learning symbols and utilizing that knowledge. Thus, user aptitude is part of the learning mechanism for cognitive competencies required to develop computer literacy. The higher the user language–learning aptitude, the more an individual can draw logical inferences and think critically. An example of this relationship in the IT realm has been demonstrated in a study by Szajna and MacKay (1995), wherein computing aptitude is shown to be related to learning performance in software training.

We conceptualize CULA as a multidimensional construct based on the fact that computer systems are multidimensional tools with many facets for the user to learn. These include, for example, hardware, software, programming, and applications such as the Internet. Conceptual development began with a review of the IT literature to determine if there were existing measures for constructs that were similar to the focal CULA construct. Previous computer aptitude measurements have been multidimensional but focused mainly on math and programming abilities (Szajna, 1994). Even when the measurement focus was on nonprogramming computer aptitude, the domain of abilities measured tends to be multidimensional. For instance, the Computer Aptitude, Literacy and Interest Profile instrument measures abilities such as logical reasoning, alphabetic and numeric sequencing, alphanumeric translation, general quantitative and visiospatial (Szajna, 1994). None of the previous studies examined learning aptitude related to computerese. It is here that the CULA contributes to the body of knowledge regarding learning computerese and subsequent technology use.

A related construct that has also been used as an external variable is computer self-efficacy (CSE; Venkatesh & Davis, 1996). CSE “refers to a judgment of one’s capability to use a computer” (Compeau & Higgins, 1995, p. 191). One can argue that CULA is similar to CSE. Based on Bandura’s (1986) definition of self-efficacy, which “is concerned not with the skills one has but with the judgments of what one can do with whatever skills one possesses” (p. 391), we distinguish CSE from CULA in that the former focuses on one’s perceived ability to perform computer-related tasks, whereas the latter focuses on one’s perceived ability to learn computer skills. We postulate that CULA reflects beliefs about the aptitude of an individual to acquire knowledge, which precedes skill application or performance.

Hypotheses Development

The distinction between CSE and CULA becomes more evident when comparing the constructs’ antecedents. The antecedents for CSE are distinct from those of CULA with the exception of (computer) anxiety (CA). Examples of antecedents for CSE include prior experience, behavioral modeling/observation (Johnson & Marakas, 2000), encouragement by others, and other’s use and support (Compeau & Higgins, 1995). Proposed antecedents, adapted from a study on second language success (Ehrman & Oxford, 1995), for CULA include tolerance of ambiguity, propensity to take risk, and anxiety.

Computer Anxiety

Language anxiety and computer anxiety (CA) share similar characteristics and both have been studied extensively (e.g., Fakun, 2009; Horwitz, 2000; Horwitz, Horwitz, & Cope, 1986; Marcoulides, 1989). While a comprehensive review of these studies is beyond the scope of this article, several observations are appropriate. Language anxiety is typically considered as “a form of performance anxiety” (Ehrman & Oxford 1995, p. 68) and branded as a debilitating type of anxiety (Horwitz et al., 1986). Individuals that have high levels of language anxiety are less likely to appreciate the usefulness of the language and will find the language difficult to use. Similarly, CA may have adverse effects on the potential usage of a computer system.

The literature also identifies two types of anxiety: (a) state anxiety caused by a specific situation and (b) trait anxiety referring to a certain disposition where one has a propensity to worry (Mikulincer, Kedem, & Paz, 1990). Research findings have also indicated that math and test anxiety may be related to CA (Desai, 2001). Heinssen, Glass, and Knight (1987) reflected on the potential origins of anxiety when they developed the Computer Anxiety Rating Scale CARS, noting that “cognitive factors [thus] appear to play an important role in computer anxiety” (p. 57). Tobias (1979) argued that the effects of anxiety on behavior and performance are mediated by cognitive processes. Thus, cognitive factors such as learning aptitude are important dimensions to investigate in order to gain a valid understanding of the impact of CA. Studies that incorporated CA into TAM have found a negative relationship between CA and PEOU (Venkatesh & Davis, 2000), though the potential mediating effect of CULA has not been articulated. We argue that CA should be treated as an antecedent to CULA because CA ought to be viewed from a state-based anxiety perspective (anxiety that is caused by a specific situation) rather than a trait-based anxiety perspective (where one has a propensity to worry in general) and will be experienced before any knowledge of a specific technology is attained or processed. We argue that when a user has computer-related “performance anxiety,” the mere anticipation of learning and using a new computer system will negatively affect the user’s ability to learn. This line of discussion leads us to the following research hypothesis:

Hypothesis 1: CA has a significant negative effect on CULA.

Tolerance of Ambiguity and Propensity to Take Risk

Tolerance of ambiguity (TA) and its related concept of risk taking (RT) may be conceptualized as enablers of learning a language, as demonstrated by studies that find “[t]hose who can tolerate ambiguity in language learning are more likely to take risks; and risk-taking is an essential for progress” (Ehrman & Oxford 1995, p. 69). A study that developed a tolerance of ambiguity scale specifically for second language learning found that “students high in tolerance of ambiguity did not mind speaking even when they were unsure of possessing the correct language tools” (Ely, 1989, p. 442). This finding supports the conclusion that “we should become aware of (and take into account) underlying . . . affective variables [such as tolerance of ambiguity] which might tend to inhibit or promote . . . strategy development” (Ely, 1989, p. 443). Other evidence regarding language learning suggests that a high level of tolerance of ambiguity was tied to multilingualism indicating that “multilingualism makes individuals more at ease with ambiguity . . . [or] a higher level of TA can also strengthen an individual’s inclination to become multilingual” (Dewaele & Wei, 2013, p. 231). Similarly, TA is a potentially important explanatory variable with regard to interactions with technology, the language involved in communicating with technology, and promoting individual strategy development for learning to speak the language of a given technology.

In TAM studies, propensity to take risk has been researched indirectly through the construct of personal innovativeness in information technology (PIIT; Agarwal & Karahanna, 2000; Agarwal & Prasad, 1999). PIIT may be a function of individuals’ tolerance for RT (Bommer & Jalajas, 1999). Research has also supported the belief that attitudes toward both risk and ambiguity are important determinants of decision making (Ghosh & Ray, 1992). Learning a new language or a new technology necessarily involves taking risk because the individual has to change habits and invest in something that is nonfamiliar. Individuals with low tolerance for risk are more resistant in adopting a new language or technology, and this will have an adverse effect on their learning aptitude. This can be attributed to the shortage of motivation and subsequent lack of effort. On the other hand, individuals with a high propensity to take risk are in general more confident, motivated, and will thus be more willing to invest in their learning aptitude.

The importance of TA and RT for learning and decision making suggests that they will also be significant to technology acceptance and usage, since learning, which is theorized to be a basis for beliefs and decision making, is part of the process of accepting new technology and using it. This line of discussion leads us to the following research hypotheses:

Hypothesis 2: TA has a significant positive effect on CULA.

Hypothesis 3: RT has a significant positive effect on CULA.

Computer User Learning Aptitude

The proposed CULA construct, which refers to “one’s perception about being able or having the capacity to learn computerese (and, by extension, to be able to dialogue successfully in the use of computer technology),” is based on the concept of aptitude as a foundation for learning (Snow, 1992). While a construct related to computer aptitude has been developed and examined in the context of computer programming (Howell et al., 1967), nonprogramming computer performance (Szajna, 1994), and teacher retraining success (Roszkowski, Devlin, Snelbecker, Aiken, & Jacobsohn, 1988), it has not been operationalized from a learning ability logic. We argue that CULA is relevant to the technology acceptance research based on two observations. First, since PU and PEOU are cognitive constructs, the most influential EVs would be those that support the cognitive process. Thus, there is a need to focus on individually distinct dimensions of cognition. Second, there are a number of assertions by TAM researchers that learning, particularly through training processes, is an important construct within the TAM process (Davis et al., 1989; Venkatesh & Davis, 1996; Venkatesh, Speier, & Morris, 2002).

Users dialogue or “communicate” with computer technology via languages that are theoretically processed in a similar manner as any human language. For example, there are syntactical rules and unique vocabularies, and nuances or meanings of words associated with computer technologies. Individuals with a low CULA score may have a difficult time interacting with a new technology because of a language barrier. The individuals will find the technology rather challenging to use and thus perceive that the technology itself is difficult to use. On the other hand, individuals with high learning aptitude scores will feel at ease operating the new technology because they can easily converse with the technology. Furthermore, individuals that speak the language of the technology will be more prone to find the technology useful because they have a better understanding of its characteristics, its idiosyncratic attributes, and its capabilities. In other words, these individuals will possess a better understanding of what the technology offers. This will result in a perception of usefulness of the new technology.

While it may be argued that interactions with computer technologies are becoming more naturalized because they resemble a native language (particularly for English speakers) and are more intuitive due to extensive use of graphical user interfaces (GUI), there are still rules and vocabularies that are unique to each computer system and even each program. For instance, while a “new” spreadsheet software package might be technically superior to the one being used, a notable advantage of sticking with the current package is that the users know how to communicate with the current package. Specifically, the users know the vocabulary and how to specifically dialogue with the current software (Ingardia, 2003). Prior research has recognized this as a barrier of technology acceptance, and it can be attributed to conscious resistance due to switching costs (Kim & Kankanhalli, 2009) as well as inertia because of transaction costs (Polites & Karanhanna, 2012). Thus, learning aptitude will significantly influence PEOU and PU of a new technology due to the substantive transition costs in terms of effort required to learn the new technology. CULA is akin to language aptitude in that it is distinguishable from general intelligence and that “[a]ptitude as a concept corresponds to the notion that . . . the individual may be thought of as possessing some current state of capability of learning” (Carroll, 1981, p. 84). The greater the CULA, the higher the PU and ease-of-use for the technology because the user finds is easier to converse with the technology and requires lower switching costs. The preceding discussion identifies CULA as an important external variable resulting in the following hypotheses:

Hypothesis 4a: CULA has a significant positive effect on PU.

Hypothesis 4b: CULA has a significant positive effect on PEOU.

Perceived Usefulness and Perceived Ease of Use

Although the main focus of this research is the conceptualization and measurement development of CULA, behavioral intention (BI) is the dependent variable of interest. Therefore, it is important to examine the relationships of PEOU and PU with BI. Behavioral Intention has been used as a proxy for actual SU due to difficulties associated with observing actual SU behavior (Anderson & Agarwal, 2010; Johnston & Warkentin, 2010) and the significant, consistent relationship between intension and actual usage demonstrated through numerous research studies (Venkatesh, Morris, Davis, & Davis, 2003). The relationship between the determinants PEOU and PU and the dependent variable BI have been extensively theorized and tested in the IT literature (e.g., Venkatesh et al., 2003), leading us to the following hypotheses:

Hypothesis 5a: PU has a significant positive effect on BI.

Hypothesis 5b: PEOU has a significant positive effect on BI.

The hypothesized relationships between the main construct, CULA, and its antecedents, as well as the relationships to PU and PEOU, are depicted in Figure 1. In the next section, we describe the operationalization of all constructs.

Figure 1.

Research framework.

Research Methodology

Operationalization of Constructs

The scales were developed based on the extant literature. A preliminary set of items for each construct along with definitions was generated and presented for evaluation to nine academicians from various disciplines. The purpose was to enhance content and face validity. The academics had an opportunity to comment on the adequacy of the measures and to suggest further refinements to the instrument. Measurement of the items was based on a 5-point Likert-type scale where 1 = strongly disagree and 5 = strongly agree for computer anxiety and tolerance of ambiguity, 1 = not at all and 5 = very much for risk-taking propensity and computer user learning aptitude, and a 7-point scale where 1 = extremely likely and 7 = extremely unlikely for perceived usefulness, perceived ease of use and behavioral intention.

Items used to operationalize CA (Table 1) were adopted from Heinssen et al. (1987), who developed a frequently cited instrument known as the Computer Anxiety Rating Scale (CARS). CARS measurement consists of two factors: (a) low anxiety toward computer use or confidence and (b) high anxiety toward computer use or fear (Miller & Rainer, 1995). Measures include, for instance, “I look forward to using a computer for my classes” (CA4), to measure how much enjoyment versus anxiety is acknowledged, and “I hesitate to use a computer for fear of making mistakes that I cannot correct” (CA2), which reflects a performance aspect of using a computer. Indicator CA4 is a reflector of low/confidence anxiety, whereas CA2 is an example of measuring high/fear anxiety.

Table 1.

Measurement Model.

Construct	Item no.	Item	Completely std. loading	t Value
Computer Anxiety—Fear	CA1	It scares me to think that I could cause the computer to destroy a large amount of information by hitting the wrong key.	0.62***	14.01
	CA2	I hesitate to use a computer for fear of making mistakes that I cannot correct.	0.81***	23.83
	CA3	I have avoided computers because they are unfamiliar and somewhat intimidating to me.	0.72***	18.89
Computer Anxiety—Confidence	CA4	I look forward to using a computer for my classes.	0.81***	21.95
	CA5	The challenge of learning about computers is exciting.	0.71***	18.22
	CA6	I am confident that I can learn computer skills.	0.65***	14.18
Risk Taking	RT1	I participate in extreme sports.	0.37***	6.24
	RT2	I like to live on the edge.	0.63***	14.16
	RT3	I would be willing to invest in a risky stock if I thought it would appreciate in price soon.	0.63***	15.54
	RT4	I would describe myself as a risk taker.	0.92***	23.86
Tolerance Ambiguity—Understanding	TA1	It bothers me that sometimes the teacher uses vague terms about the computer, even though I understand the general idea.	0.53***	11.80
	TA2	When I am using a new software, I don’t like the fact that I can’t always accomplish exactly what I want to because the commands are not clear.	0.74***	23.86
	TA3	It bothers me that even when I study the computer applications, some of them are hard to understand.	0.80***	31.34
	TA4	I find it kind of frustrating that it’s sometimes hard to pin down the meaning or function of some computer commands.	0.85***	38.11
	TA5	I don’t like the fact that it’s often impossible to find commands in one application that work the exact same way in another.	0.70***	20.91
Tolerance Ambiguity—Philosophy	TA6	Nothing gets accomplished in this world unless you stick to some basic explicit rules.	0.63***	9.55
Tolerance Ambiguity—Philosophy	TA7	Usually, the more clearly defined rules a society has, the better of it is.	0.81***	10.80
Tolerance Ambiguity—Problem-Solving	TA8	A problem has little attraction for me if I don’t think it has a solution.	0.66***	9.85
Tolerance Ambiguity—Problem-Solving	TA9	I don’t like to work on a problem unless there is a possibility of coming out with a clear-cut and unambiguous answer.	0.57***	8.90
Computer User Learning Aptitude—Programming	CULA1	I am able to write my own software program without problems.	0.85***	28.13
Computer User Learning Aptitude—Programming	CULA2	I can learn new programming languages without a lot of effort.	0.88***	29.91
Computer User Learning Aptitude—General Hardware/Software	CULA3	I have the ability to learn new software easily.	0.62***	15.51
	CULA4	I can learn new hardware quickly.	0.66***	18.13
	CULA5	I have the ability to recognize a hardware limitation easily.	0.76***	25.94
	CULA6	I can learn to use the latest hardware without difficulty.	0.78***	28.91
	CULA7	I am able to learn to use peripherals without problems.	0.80***	31.72
	CULA8	I can troubleshoot hardware problems successfully with little effort.	0.76***	26.69
Computer User Learning Aptitude—Internet	CULA9	I am able to learn how to surf the Internet using new browsers quickly.	0.80***	30.15
	CULA10	I have the ability to understand how to join a chat room.	0.77***	26.69
	CULA11	I am capable of learning to chat online in different types of chat rooms without a lot of effort.	0.75***	24.13
	CULA12	I can easily learn how to upload files using different methods.	0.76***	25.06
	CULA13	Downloading software using different types of file transferring is not a problem for me to learn.	0.72***	22.09
	CULA14	I am able to learn how to find information on the internet easily.	0.69***	19.63
	CULA15	I have the ability to understand how to browse a site quickly.	0.69***	19.78
Perceived Usefulness	PU1	Using the Internet for research enhances my productivity.	0.81***	38.19
	PU2	Using the Internet enhances my effectiveness completing research for my work.	0.94***	82.31
	PU3	Using the Internet for research improves my performance.	0.91***	69.27
Perceived Ease of Use	PEOU1	Learning to operate the Internet is easy for me.	0.90***	65.50
	PEOU2	I find it easy to get the Internet to do what I want it to do.	0.82***	39.52
	PEOU3	It is easy for me to become skillful at using the Internet.	0.88***	59.37
	PEOU4	I find the Internet easy to use.	0.91***	75.38
Behavioral Intention	BI1	I plan to use the Internet for research in the future.	0.97***	196.08
	BI2	I intend to continue using Internet for research in the future.	0.98***	231.14
	BI3	I expect my use of the Internet for research to continue in the future.	0.91***	85.30
Computer User Learning Aptitude (Second Order)	Computer User Aptitude—Software/Hardware		0.61***	11.92
Computer User Learning Aptitude (Second Order)	Computer User Aptitude—Internet		0.90***	19.12
	Computer User Aptitude—Programming		0.33***	5.09

Notes. Fit indices (overall): χ² (df) = 1635.62 (848); χ²/df= 1.93; CFI (comparative fit index) = 0.91; TLI (Tucker–Lewis index) = 0.90; RMSEA (root mean square error of approximation) = 0.05.

***

p < .001.

Attitude toward RT was measured using a combination of items from a previously developed instrument known as the Choice Dilemma Questionnaire (CDQ; Kogan & Wallach, 1964) and other indicators. The CDQ instrument includes 12 situations that describe a choice between two alternative courses of action. Due to the length of the questionnaire, three representative situations were included in a pilot study of which one was dropped due to poor psychometric properties. In addition, five nonsituational items were added. An example of the nonsituational items is, “I would describe myself as a risk-taker” (RT4).

Indicators for tolerance of ambiguity are based on a tolerance of ambiguity instrument developed specifically for second language learning (Ely, 1989) and derived from the commonly used general scales of Norton (1975). The measures were adapted for the information technology context. TA has been viewed as having up to eight dimensions (Herman, Stevens, Bird, Mendenhall, & Oddou, 2010). Based on parsimony and relevance, we opted to operationalize it via three dimensions; two dimensions identified from the Norton (1975) scale—(a) Philosophy and (b) Problem-solving—and one from Ely (1989)—(c) Understanding. An example of an item operationalizing the philosophy aspect is as follows: “Nothing gets accomplished in this world unless you stick to some basic explicit rules” (TA6). Similarly, “It bothers me that sometimes the teacher uses vague terms about the computer, even though I understand the general idea” (TA1) is an observed indicator for the concept of understanding. Finally, “A problem has little attraction for me if I don’t think it has a solution” (TA8) is one of the manifest items for the problem-solving dimension.

We operationalized CULA by creating an initial set of items based on the underlying conceptualization that aptitude is primarily a reflection of the learning ability of an individual with regard to computerese, that is, any computer terminology, language, or syntax. The items were reviewed by experts in information systems and research methodology for appropriateness and comprehensiveness. The scale included items reflecting the learning ability of an individual with respect to multiple dimensions of computer-related tasks and activities such as software, hardware, programming, and use of the Internet. This is in contrast to the ability of an individual to perform a specific task or activity. For example, there is an important distinction between the following statements: “I can fly an airplane” and “I am capable of learning how to fly an airplane.” Also, “I can speak Spanish” versus “I am capable of learning how to speak Spanish.” Likewise, “I can upload files using different methods” versus “I can learn to upload files using different methods.” In all three sets of statements, the first statement is a measure of self-efficacy, whereas the second is a measure of aptitude or learning capacity.

A specific effort was made to include wording emphasizing learning ability and to cover the various nuances of computer technology. Examples of items include: “I can learn new programming languages without a lot of effort” (CULA2) focusing on programming; “I have the ability to learn new software easily” (CULA3) and “I can learn new hardware quickly” (CULA4), both focusing on general software/hardware; and “I am capable of learning to chat online in different types of chat rooms without a lot of effort” (CULA11) focusing on the Internet.

We argue that each of the three dimensions of computer technology—programming, general software/hardware, and the Internet—are important to sufficiently represent the domain of CULA. Programming has a very complex syntax and grammar with its own languages, making programming a desirable CULA dimension to understand. Previous research suggests that learning to use software seems to be analogous to any other academic learning and not necessarily explained specifically by programming aptitude (Kagan & Pietron, 1987); therefore, learning ability for general software and hardware use appears to be a separate dimension. With over 2.4 billion users worldwide (Internet World Stats, 2013), the Internet is perhaps the most widely used technology in the world. Evidence indicates that there is an Internet-specific language with “distinctive written feature, primarily acronyms, abbreviations and respellings” (Squires, 2010, p. 457) used in cyberspace, supporting the conceptualization of Internet language learning ability as a separate dimension of CULA.

PU and PEOU constructs were used as developed and tested by Davis et al. (1989) and validated in a multitude of studies (e.g., Taylor & Todd, 1995). Given the nature of the survey subject, we opted to focus on the use of the Internet as the context of the empirical inquiry. Thus, survey items regarding PU, PEOU, and BI are in reference to the Internet. The Internet was considered appropriate for two main reasons: (a) it is voluntary in the school curriculum and students have access to the technology throughout the campus in open computer labs and at home and (b) it is representative of technology that would be used by the general public, not merely technical specialists. While operationalizing TAM constructs via a “programming language” context might appear to be optimal for illustrative purposes (since programming languages are analogous to natural languages and have a more structured syntactic framework than graphical user interface–based technology), we believe this would elicit an overly biased sample through self-selection due to the specific technical orientation programming languages demand.

Data Collection

The proposed research model was operationalized through a field study using a survey methodology for data collection. Data were collected from students of various business majors enrolled in introductory Management Information Systems (MIS) courses at a large southern state university. The surveys were administered during class time and no incentives were offered to potential respondents. Participation was voluntary and almost 100% of the attending students participated in the study. The ratio of males to females was almost equal, with 52.5% male and 47.5% female. The racial mix was 19% Black, 51% White, 17% Hispanic, 4% Asian/Pacific Islander, 9% Other. An iterative process was used for the survey instrument development beginning with a pilot sample of 64 responses. The results of the pilot survey analysis were used to refine the survey instrument. A total of 350 surveys were collected toward confirmatory data analyses, which we describe below.

Confirmatory Factor Analysis

The measurement model comprising all the items was subjected to confirmatory factor analysis (CFA) using Mplus 6 and maximum likelihood estimation. The overall fit of the hypothesized model is assessed by examining the chi-square to degrees of freedom (χ²/df), comparative fit index (CFI), Tucker–Lewis fit index (TLI), and root mean square error of approximation (RMSEA). Convergent validity is evaluated by examining the individual items’ standardized coefficients (factor loadings) and their respective t values. Discriminant validity is established by comparing the average variance extracted (AVE) with the squared correlations between constructs (Fornell & Larcker, 1981; Table 2). Evidence for reliability is obtained by examining the respective composite reliability (CR) measure and the average variance extracted (Fornell & Larcker, 1981).

Table 2.

Correlation Matrix and Reliability.

Constructs	1	2	3	4	5	6	7	8	9	10	11	12
1. Computer Anxiety—Fear	0.76^a (0.52^b)
2. Computer Anxiety—Confidence	0.46^c *** (0.21)^d	0.77^a (0.53^b)
3. Risk Taking	0.06 (0.00)	−0.01 (0.00)	0.75^a (0.45^b)
4. Tolerance Ambiguity—Understanding	−0.51*** (0.26)	−0.24*** (0.06)	−0.17** (0.03)	0.85^a (0.54^b)
5. Tolerance Ambiguity—Philosophy	−0.20** (0.04)	0.10* (0.01)	−0.14** (0.02)	0.33*** (0.11)	0.68^a (0.52^b)
6. Tolerance Ambiguity—Problem-Solving	−0.45*** (0.20)	−0.17** (0.03)	0.20*** (0.04)	0.46*** (0.21)	0.50*** (0.25)	0.55^a (0.38^b)
7. Computer User Learning Aptitude—Software/Hardware	0.19*** (0.04)	0.16** (0.03)	0.08^† (0.01)	−0.16** (0.03)	−0.02 (0.00)	−0.10* (0.01)	0.85^a (0.75^b)
8. Computer User Learning Aptitude—Internet	0.36*** (0.13)	0.31*** (0.09)	0.14** (0.02)	−0.30*** (0.09)	−0.05 (0.00)	−0.19*** (0.04)	0.69*** (0.47)	0.87^a (0.54^b)
9. Computer User Learning Aptitude—Programming	0.52*** (0.27)	0.45*** (0.20)	0.21*** (0.04)	−0.43*** (0.19)	−0.07 (0.00)	−0.28*** (0.08)	0.30*** (0.09)	0.55*** (0.30)	0.89^a (0.55^b)
10. Perceived Usefulness	0.23*** (0.05)	0.26*** (0.07)	0.11*** (0.01)	−0.06 (0.00)	0.03 (0.00)	−0.07 (0.01)	0.16** (0.03)	0.30*** (0.09)	0.44*** (0.19)	0.92^a (0.79^b)
11. Perceived Ease of Use	0.30*** (0.09)	0.32*** (0.10)	0.10* (0.00)	−0.23*** (0.05)	−0.03 (0.00)	−0.09^† (0.01)	0.23*** (0.05)	0.42*** (0.18)	0.62*** (0.38)	0.72*** (0.51)	0.93^a (0.77^b)
12. Behavioral Intention	0.20*** (0.04)	0.17** (0.03)	0.06 (0.00)	−0.03 (0.00)	0.04 (0.00)	−0.05 (0.00)	0.15** (0.02)	0.27*** (0.07)	0.40*** (0.16)	0.71*** (0.50)	0.76*** (0.57)	0.97^a (0.91^b)

Notes. Fit indices (overall): χ² (df) = 1635.62 (848); χ²/df = 1.93; CFI (comparative fit index) = 0.91; TLI (Tucker–Lewis index) = 0.90; SRMR (standardized root mean square residual) = 0.08. On the diagonal: a = composite reliability; b = average variance extracted. Off the diagonal: c = correlation; d = Squared correlation in parentheses.

†

p < .1. *p < .05. **p < .01. ***p < .001.

Based on our theoretical arguments, it was deemed necessary to test for the existence of a higher order factor structure for CULA within the context of the overall measurement model. We deploy a hierarchical approach recommended by Koufteros, Babbar, and Kaighobadi (2009). Using this approach, we compare four different models and select the model that is conceptually supported and fits the data well. In the first model (Model 1), we specify that all 15 indicators for CULA load onto a single first-order factor, whereas the second model (Model 2) specifies three uncorrelated first-order factors. The primary difference between Model 2 and Model 3 is that all the first-order factors are correlated in Model 3. The last model (Model 4) specifies three first-order factors and one second-order factor. After establishing an appropriate measurement model, we analyze the structural model to test our hypotheses.

Results

The assessment of the overall measurement model includes testing whether a second-order factor structure describes computer user learning aptitude (see Table 3). All latent variables are specified at the first level of abstraction, whereas CULA is examined via four alternative models. Model 1 specifies a first-order model with one factor related to all manifest variables. This model produced the worst fit among all four models tested. Although Model 2, which specifies three uncorrelated first-order factors, had a better fit than Model 1, it still generated poor fit indices as compared with Models 3 and 4. Model 3 included three correlated first-order factors and produced a better model fit than Model 4, which specified three first-order factors and one second-order factor. This is often the case, as a second-order model can never have better fit indices than its corresponding first-order correlated model (Koufteros et al., 2009). However, a second-order model that is comparable in terms of its fit indices with the correlated first-order model can serve as an attractive option if it can be conceptually supported. The fit of Model 4 proved to be similar to the fit generated by Model 3 and can be conceptually supported. Thus, Model 4 appears to be a suitable alternative and was used for as the model for evaluating the hypotheses.

Table 3.

Comparative Model Fit.

Goodness of fit indices for alternative models of factor structure	Model 1: One first-order factor	Model 2: Three first-order factors uncorrelated	Model 3: Three first-order factors correlated	Model 4: Three first-order factors, one second-order factor
χ² (df)	1401.02 (501)	1149.36 (489)	938.92 (486)	977.28 (498)
χ²/df	2.80	2.35	1.93	1.96
Comparative fit index	0.82	0.87	0.91	0.91
Tucker–Lewis index	0.81	0.85	0.90	0.90
Standardized root mean square residual	0.08	0.14	0.06	0.07

The measurement model presents the following fit indices: χ² (df) = 1635.62 (848), χ²/df = 1.93, CFI = 0.91, TLI = 0.90, RMSEA = 0.05. These fit indices are suggestive of a good fitting model based on the criteria that χ²/df < 2, CFI and TLI > 0.90, and RMSEA ≤ 0.08 (Hu & Bentler, 1999). Furthermore, all the items loaded significantly onto their respective factors based on their t values (see Table 1). Most of the factor loadings were above 0.50 and a great majority of those were above 0.60. Table 2 provides information about the AVE, CR, and squared factor correlations. Evidence for discriminant validity can be obtained by comparing the AVE with the squared factor correlations. The highest squared correlation was observed between behavioral intention and PEOU at 0.57 and the respective AVEs are 0.91 and 0.77. The AVEs for the respective constructs are higher than the squared correlation between the constructs rendering support for discriminant validity (Fornell & Larcker, 1981). Evidence for reliability is obtained by probing the CR and AVE values. In our case, the CR value for every construct except for tolerance ambiguity-philosophy (0.68) and problem-solving (0.55) are above 0.70, and the AVE values for every construct is above or close to 0.50 with the exception of tolerance of ambiguity-problem-solving (0.38), which is operationalized by only two indicators and thus the relatively lower values reported for CR and AVE.

The results for the structural model are presented in Table 4. The overall fit indices of the structural model are acceptable under the guidelines proposed by Hu and Bentler (1999): χ² (df) = 1668.59 (867), χ²/df = 1.92, CFI = 0.91, TLI = 0.90, RMSEA = 0.05. Hypothesis 1 suggests that anxiety impacts CULA and involved two types of anxiety. We first test the relationship between anxiety (fear) and CULA. The standardized path coefficient is indicative of a significant relationship (γ = 0.30, t = 3.35, p < .001). Given that anxiety (fear) was reverse coded this suggests that fear is negatively related to CULA. We also find evidence that relates anxiety (confidence) with CULA positively (γ = 0.28, t = 3.72, p < .001).

Table 4.

Structural Model Coefficients.

Structural path (From → To)	Hypotheses	Coefficient (t value)
Computer anxiety—Fear → Computer User Learning Aptitude	H1	0.30^a *** (3.35)
Computer Anxiety—Confidence → Computer User Learning Aptitude	H1	0.28*** (3.72)
Tolerance Ambiguity—Understanding → Computer User Learning Aptitude	H2	−0.19** (−2.42)
Tolerance Ambiguity—Philosophy → Computer User Learning Aptitude	H2	0.05 (0.58)
Tolerance Ambiguity—Problem-Solving → Computer User Learning Aptitude	H2	0.01 (0.10)
Risk Taking → Computer User Learning Aptitude	H3	0.17** (2.78)
Computer User Learning Aptitude → Perceived Usefulness	H4a	0.47*** (8.91)
Computer User Learning Aptitude → Perceived Ease of Use	H4b	0.65*** (15.81)
Perceived Usefulness → Behavioral Intention	H5a	0.35*** (6.30)
Perceived Ease of Use → Behavioral Intention	H5b	0.50*** (9.32)

Notes. Fit indices (overall): χ² (df) = 1668.59 (867); χ²/df = 1.92; CFI (comparative fit index) = 0.91; TLI (Tucker–Lewis index) = 0.90; RMSEA (root mean square error of approximation) = 0.05.

Completely standardized coefficient, One-tailed significance level.

p < .01. ***p < .001.

Tolerance for ambiguity was operationalized via three factors and therefore Hypothesis 2 is divided into three parts. First, we find evidence that relates tolerance for ambiguity (understanding) to CULA (γ = −0.19, t = −2.42, p < .01). The negative coefficient signifies that lack of understanding has adverse effects on CULA. The second relationship specifies that tolerance for ambiguity (philosophy) is related to CULA, but there is no statistical evidence to support it (γ = 0.05, t = 0.58, p > .05). Finally, there is no evidence to suggest that tolerance for ambiguity (problem-solving) affects CULA (γ = 0.01, t = 0.09, p > .05). Hypothesis 2 is partially supported.

Hypothesis 3 posits that RT positively impacts CULA. The hypothesis is supported (γ = 0.17, t = 2.78, p < .01) and demonstrates that higher propensity for RT can have fruitful effects on CULA. Hypotheses 4a and 4b suggest that CULA positively affects PU and PEOU. There is sufficient evidence to support both hypotheses (β = 0.47, t = 8.91; p < .001 and β = 0.65, t = 15.80, p < .001, respectively). The instrumental role of CULA cannot be ignored.

Finally, Hypotheses 5a and 5b suggest that PU and PEOU do positively affect behavioral intention. We find evidence to support both Hypotheses 5a and 5b (β = 0.35, t = 6.30, p < .001; and β = 0.50, t = 9.32, p < .001, respectively). It appears that the effects of PEOU on behavioral intention are more pronounced.

Conclusion

In this study, we have proposed and tested a new construct that we labeled as computer user learning aptitude. We relied on the literature of language learning to explicate the process by which individuals perceive the usefulness and ease-of-use of technology as well as the user’s behavioral intention to use a technology. There is empirical evidence to suggest that CULA can be operationalized as a second-order construct and it is modeled to include three first-order nuances regarding learning aptitude as it pertains to (a) programming, (b) general hardware/software, and (c) the Internet.

Resting on theories regarding second language adoption, we specified that three individual differences (i.e., anxiety, tolerance for ambiguity, and propensity to take risk) can explain variability in CULA. We find that CULA correlates the highest with CA, whether it is CA demonstrating fear or CA demonstrating confidence. Thus, it appears very important to know what range of CA a user is experiencing in order to ascertain whether it will inhibit or facilitate CULA. Only one tolerance of ambiguity factor (i.e., general understanding) is related to CULA. The relationship appears to be negative, which suggests that lack of TA for understanding a technology will have an adverse impact on CULA, whereas a lack of TA regarding the general philosophy of approach or problem solving will not significantly impact CULA. Therefore, it appears that closely monitoring a user, as he or she learns about a technology and adjusting the delivery of specific technology usage information as needed for understanding would be more important than knowing what the user’s general approach is to rules and ambiguous situations. Finally, the propensity for RT does have a positive relationship to CULA indicating that risk takers are by nature better able to learn and use technology.

With regard to the core or direct explanatory variables of behavioral intention, CULA appears to have a strong relationship with PEOU and PU, although the effect size and respective significance of the relationship with PEOU is rather large. The findings suggest that if PU and PEOU are important factors in using a technology, then individual learning aptitude plays a significant role. As anticipated, both PU and PEOU were found to have a positive relationship with behavior intention.

This study has several implications for research and practice. For research purposes, honing on the concept of individual differences and cognition within the technology behavioral models, such as TAM, can shed light in to the “black box” of EVs, which include a collection of antecedents to PU and PEOU. Additionally, introducing concepts that impact success in another cognitive process, that is, second language learning, provides an avenue for understanding success in technology acceptance. For practice, understanding the path to technology acceptance by users and the impact of their learning ability is critically important in many contexts, such as computer application training, online shopping site design, and general computer application design and development. This study highlights CA, tolerance of ambiguity, and RT as important antecedents to a newly conceptualized construct—CULA—all of which have significant relationships leading up to a user’s perception of usefulness, ease of use, and ultimately behavioral intention. If a practitioner is looking to have effective and timely acceptance and use of a technology they need to take the CULA, TA, CA, and RT user differences into account.

As always, we should note limitations to the study. One such limitation is that the target technology for the PU and PEOU constructs is a voluntary, general purpose technology (Internet); therefore, the significant relationship between CULA and PEOU may be deemed dependent on the particular target technology. Future studies should look at other categories of technology targets. For example, Java, 3D printing, or Excel could be used as target technologies to validate the nomological network developed here. The current model was cross-validated using an alternative technology (i.e., the use of PowerPoint), and the results suggest that the model is rather robust; the fit indices and path coefficients do not differ substantially from the results produced using the Internet as the context.

The study is also limited because the sample population was composed of college-level students. However, we assert that the use of students for this study is valid since the target technology is the Internet, a technology that is available to users in the academic, as well as in the professional context. Future research could replicate this model with other groups to verify this assertion.

The ultimate dependent variable deployed in this study (i.e., behavioral intention) is used as a proxy for SU. We deem this dependent variable to be acceptable for two reasons. First, the focus of this study is on developing a new external variable, computer user learning ability, and its antecedents. Second, the relationships between PEOU, PU, BI, and SU have been researched and theoretically justified in numerous prior studies in the Management Information Systems extant literature (e.g., see a review in Lee, Kozar, & Larsen, 2003). BI continues to be an acceptable proxy for SU when the focus of the study is on antecedent construct development (Polites & Karahanna, 2012).

While we believe that CULA is vital in explaining variation in PU and PEOU, moderator variables may also play a significant role. For instance, the relationship between CULA and PU may be subject to the utility or potential a user identifies with a specific technology. Similarly, the relationship between CULA and PEOU may be susceptible to the complexity of specific technology where the relationship is more salient for more complex adaptations to technology. Future research should incorporate moderators that can help explain relationships more rigorously.

The posited model fits the data relatively well. However, other model specifications can perhaps describe the data more adequately. Toward this end, we tested an alternative model where the exogenous variables were also specified as direct antecedents to PU and PEOU. In other words, we specified a partial mediation model. We subsequently compared it with the full mediation model we advanced. The χ² difference of 17.08 (12 df) between the two models is nonstatistically significant (p > .15) and favors the more parsimonious model we hypothesized. Furthermore, the χ²/df and Akaiki information criterion (AIC) indices also favor the specified model. In addition, an examination of the 12 direct paths between the exogenous variables and PU and PEOU reveals that only one path is statistically significant. The addition of 12 paths does not appear to be warranted.

Current studies continue to expand on the understanding of technology acceptance with research focusing on contemporary themes such as virtual communities (Lin, 2009), mobile devices (Chang, 2010), and online shopping (Chiu et al., 2009). Regardless of the context or target technology, users must learn to use the new application and its associated language. Thus, their ability to learn is key to the process of technology acceptance. The development of an acceptable measurement instrument for the proposed construct is only the first step in the process of understanding the impact of learning aptitude differences.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Agarwal

Karahanna

(2000). Time flies when you’re having fun: Cognitive absorption and beliefs about information technology usage. MIS Quarterly, 24, 665-694.

Agarwal

Prasad

(1999). Are individual differences germane to the acceptance of new information technologies? Decision Sciences, 30, 361-391.

Anderson

C. L.

Agarwal

(2010). Practicing safe computing: A multimethod empirical examination of home computer user security behavioral intentions. MIS Quarterly, 34, 613-643.

Bandura

(1986). Social foundations of thought and action. Englewood Cliffs, NJ: Prentice Hall.

Bommer

Jalajas

(1999). The threat of organizational downsizing on the innovative propensity of R & D professionals. R&D Management, 29(1), 27-34.

Carroll

J. B.

(1981). Twenty-five years of research on foreign language aptitude. In Diller

K. C.

(Ed.), Individual differences and universals in language learning aptitude (pp. 83-118). Rowley, MA: Newbury House.

Chang

C. K

(2010). Acceptability of an asynchronous learning forum on mobile devices. Behaviour & Information Technology, 29, 23-33.

Chiu

C. M.

Lin

H. Y.

Hsu

M. H.

(2009). Understanding customers’ loyalty intentions towards online shopping: An integration of technology acceptance and fairness theory. Behaviour & Information Technology, 28, 347-360.

Compeau

Higgins

(1995). Computer self-efficacy: Development of a measure and initial test. MIS Quarterly, 19, 189-211.

10.

Computerese. (1994). Webster’s encyclopedic unabridged dictionary of the English language. New York, NY: Gramercy Books.

11.

Cronbach

Snow

(1977). Aptitudes and instructional methods. New York, NY: Wiley.

12.

Davis

, (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly, 13, 319-340.

13.

Davis

Bagozzi

Warshaw

(1989). User acceptance of computer technology: A comparison of two theoretical models. Management Science, 35, 982-1003.

14.

Deng

Doll

Troung

(2004). Computer self-efficacy in an ongoing use context. Behaviour & Information Technology, 23, 395-412.

15.

Desai

(2001). Computer anxiety and performance: An application of a change model in a pedagogical setting. Journal of Instructional Psychology, 28(3), 70-78.

16.

Dewaele

J. M.

Wei

(2013). Is multilingualism linked to a higher tolerance of ambiguity? Language and Cognition, 1, 231-240.

17.

Ehrman

Oxford

(1995). Cognition plus: Correlates of language learning success. Modern Language Journal, 79(1), 67-89.

18.

Ely

(1989). Tolerance of ambiguity and use of second language strategies. Foreign Language Annals, 22, 437-445.

19.

Evans

G. E.

Simkin

M. G.

(1989). What best predicts computer proficiency? Communications of the ACM, 32, 1322-1327.

20.

Fakun

(2009). How to mitigate the significant negative influence of computer anxiety on ease of use perceptions. Behaviour & Information Technology, 28, 223-238.

21.

Fornell

Larcker

(1981, February). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18, 39-50.

22.

Gardner

(1983). Frames of mind: The theory of multiple intelligences. New York, NY: Basic Books.

23.

Gardner

(2011). Theories of intelligence. In Bray

M. A.

Kehle

T. J.

(Eds.), The Oxford handbook of school psychology (pp. 79-100). New York, NY: Oxford University Press.

24.

Gefen

Karahanna

Straub

(2003). Trust and TAM in online shopping: An integrated model. MIS Quarterly, 27, 51-90.

25.

Ghosh

Ray

(1992). Risk attitude, ambiguity intolerance and decision making: An exploratory investigation. Decision Sciences, 23, 431-444.

26.

Heinssen

Glass

Knight

(1987). Assessing computer anxiety: Development of the computer anxiety rating scale. Computers in Human Behavior, 3(1), 49-59.

27.

Herman

Stevens

J. L.

Bird

M. J.

Mendenhall

Oddou

(2010). The tolerance for ambiguity scale: Towards a more refined measure for international management research. International Journal of Intercultural Relations, 34(1), 58-65.

28.

Horwitz

(2000). It ain’t over ‘til it’s over: On foreign language anxiety, first language deficits, and the confounding of variables. Modern Language Journal, 84, 256-259.

29.

Horwitz

Cope

(1986). Foreign language classroom anxiety. Modern Language Journal, 70, 125-132.

30.

Howell

Vincent

Gay

(1967). Testing aptitude for computer programming. Psychological Reports, 20, 1251-1256.

31.

L. T.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6, 1-55.

32.

Ingardia

(2003). Selecting A/E software: It’s not just how well it works. Design Firm Management & Administration Report, 3.

33.

Internet World Stats. (2013). Internet world stats: Usage and population statistics. Retrieved from www.Internetworldstats.com/stats.htm

34.

Johnson

(1983). Educator’s lexicon of computerese. The Arithmetic Teacher, 30(6), 46-49.

35.

Johnson

Marakas

(2000). Research report: The role of behavioral modeling in computer skills acquisition—Toward refinement of the model. Information Systems Research, 11, 402-417.

36.

Johnston

A. C.

Warkentin

(2010). Fear appeals and information security behaviors: An empirical study. MIS Quarterly, 34, 549-566.

37.

Kagan

Pietron

L. R.

(1987). Cognitive level and achievement in computer literacy. Journal of Psychology, 121, 317-327.

38.

Kantorowitz

Sudarsky

(1989). The adaptable user interface. Communications of the ACM, 32, 1352-1358.

39.

Kim

H. W.

Kankanhalli

(2009). Investigating user resistance to information systems implementation: A status quo bias perspective. MIS Quarterly, 33, 567-582.

40.

Kogan

Wallach

(1964). Risk-taking: A study in cognition and personality. New York, NY: Holt, Rinehart & Winston.

41.

Konvalina

Woleman

Stephens

(1983). Math proficiency: A key to success for computer science students. Communications of the ACM, 26, 377-382.

42.

Koufaris

(2002). Applying the technology acceptance model and flow theory to online consumer behavior. Information Systems Research, 13, 205-225.

43.

Koufteros

Babbar

Kaighobadi

(2009). A paradigm for examining second-order factor models employing structural equation modeling. International Journal of Production Economics, 120, 633-652.

44.

Lazar

Jones

Shneiderman

(2006). Workplace user frustration with computers: An exploratory investigation of the causes and severity. Behaviour & Information Technology, 25, 239-251.

45.

Lee

Kozar

K. A.

Larsen

K. R.

(2003). The technology acceptance model: Past, present, and future. Communications of the Association for Information Systems, 12, 752-780.

46.

Lin

H. F.

(2009). Examination of cognitive absorption influencing the intention to use a virtual community. Behaviour & Information Technology, 28, 421-431.

47.

Marcoulides

(1989). Measuring computer anxiety: The computer anxiety scales. Educational & Psychological Measurement, 49, 733-739.

48.

Martinez-Torres

M. R.

Toral Marin

S. L.

Barrero Garcia

Gallardo

Arias Oliva

Torres

(2008). A technological acceptance of e-learning tools used in practical and laboratory teaching, according to the European higher education area. Behaviour & Information Technology, 27, 495-505.

49.

Mikulincer

Kedem

Paz

(1990). Anxiety and categorization: I. The structure and boundaries of mental categories. Personality and Individual Differences, 11, 805-814.

50.

Miller

M. D.

Rainer

R. K.

Jr. (1995). Assessing and improving the dimensionality of the Computer Anxiety Rating Scale. Educational and Psychological Measurement, 55, 652-657.

51.

Norton

(1975). Measurement of ambiguity tolerance. Journal of Personality Assessment, 39, 607-619.

52.

Polites

G. L.

Karanhanna

(2012). Shackled to the status quo: The inhibiting effects of incumbent system habit, switching costs, and inertia on new system acceptance. MIS Quarterly, 36, 21-55.

53.

Raschen

J. F.

(1928). The dual aspect of language. Modern Language Journal, 8, 169-177.

54.

Roszkowski

Devlin

Snelbecker

Aiken

Jacobsohn

(1988). Validity and temporal stability issues regarding two measures of computer aptitudes and attitudes. Educational and Psychological Measurement, 48, 1029-1035.

55.

Schneiderman

Hochheiser

(2001). Universal usability as a stimulus to advanced interface design. Behaviour & Information Technology, 20, 367-376.

56.

Shih

(2006). The effect of computer self-efficacy on enterprise resource planning usage. Behaviour & Information Technology, 25, 407-411.

57.

Snow

(1992). Aptitude theory: Yesterday, today and tomorrow. Educational Psychologist, 27, 5-32.

58.

Squires

(2010). Enregistering internet language. Language in Society, 39, 457-492.

59.

Szajna

(1994). An investigation of the predictive validity of computer anxiety and computer aptitude. Educational and Psychological Measurement, 54, 926-1035.

60.

Szajna

Mackay

(1995). Predictors of learning performance in a computer-user training environment: A path-analytic study. International Journal of Human-Computer Interaction, 7, 167-185.

61.

Taylor

Todd

(1995). Understanding information technology usage: A test of competing models. Information Systems Research, 6, 144-176.

62.

Tobias

(1979). Anxiety research in educational psychology. Journal of Educational Psychology, 71, 573-582.

63.

Towell

Hawkins

(1994). Approaches to second language acquisition. Clevedon, Avon, England: Philadelphia Multimedia Matters.

64.

Van Schaik

Radford

Hogg

(2010). Modelling the acceptance of internet sites with domestic-violence information. Behaviour & Information Technology, 29, 615-620.

65.

Venkatesh

Davis

(1996). A model of the antecedents of perceived ease of use: Development and test. Decision Sciences, 27, 451-481.

66.

Venkatesh

Davis

(2000). A theoretical extension of the technology acceptance model: Four longitudinal field studies. Management Science, 46, 186-204.

67.

Venkatesh

Morris

M. G.

Davis

G. B.

Davis

F. D.

(2003). User acceptance of information technology: Toward a unified view. MIS Quarterly, 27, 425-478.

68.

Venkatesh

Speier

Morris

(2002). User acceptance enablers in individual decision making about technology: Toward an Integrated Model. Decision Sciences, 33, 297-316.

69.

Wallendorf

(2001). Literally literacy. Journal of Consumer Research, 27, 505-511.

70.

I. L.

K. W.

(2005). A hybrid technology acceptance approach for exploring e-CRM adoption in organizations. Behaviour & Information Technology, 24, 303-316.

71.

Zemanek

(1966). Semiotics and programming languages. Communications of the ACM, 9, 139-143.