Gender Effects in Student Technical and Scientific Writing: A Corpus-Based Study

Researcher’s note (June 2026). Stefanie Wulff and I wrote this paper because the question of whether gender shapes technical writing kept being answered by anecdote rather than data, and we wanted to ask it the way corpus linguistics lets us—by measuring what 87 student writers actually did with adverbs and passive voice across a 757,533-word corpus. The core finding has stayed with me: females and males largely used the same style markers but to fulfill different rhetorical functions, which means the interesting story is in how a feature is used, not just how often. That lens—looking past surface frequency to rhetorical function in real student writing—now anchors my work on AI in writing and assessment, where we again have to separate what a measure counts from what a writer is doing.

Abstract

Background: This study adopted a corpus-linguistics approach to investigate the gender effects in students’ technical and scientific writing. Specifically, we analyzed whether gender influenced how males and females used adverbs (e.g., very, really, and definitely) and passive voice (e.g., the article was published in the journal). The overuse of both adverbs and passive voice has been associated with poor writing clarity and concision. Literature review: Previous research works on gender effects in language have been mixed. Since these are all the essential elements of effective technical communication, teachers need to know what gender effects might exist. Research questions are as follows: 1. Does gender influence the student writers’ use of adverbs? 2. Does gender influence the student writers’ use of passive voice? Methodology: The sample included 87 writers (46 females and 41 males) who contributed to a 757,533-word corpus. Researchers analyzed 12,111 instances of adverbs and 4,732 instances of passive voice within a variety of technical texts. Results/discussion: Female writers used significantly more adverbs as well as more additive/restrictive, degree, and stance adverbs than expected. Male writers used more linking and manner adverbs than expected. Female writers also used significantly more passives, particularly passive verbs associated with reporting findings and interpretation. In contrast, male writers associated with passive verbs used to describe methods and analyses. Overall, the results suggested that females and males used the same style markers to fulfill different rhetorical functions.

Index Terms—Adverbs, corpus linguistics, passive voice, technical communication, technical writing.

Introduction

There is wide interest in how gender influences language choices and, thus, in how males and females might appeal differently to their audience. Empirical research across multiple registers suggests that males typically use language that is more direct, succinct, and personal, and females typically use language that is more indirect, elaborate, and affective. (See [1] for an overview.) Only a few studies have examined the gender differences within technical and professional communication. For example, Tebeaux observed that the language choices made by her technical writing students appeared to be influenced more by experience than biological sex. The writing by her students with prior work experience had an androgynous quality, whereas the writing produced by students without this experience reflected the gender differences identified in previous research [2, pp. 26–28]. However, Tebeaux acknowledged that directness and dynamism, as well as tact and creativity, are essential elements to all technical writing, and that if gender-based writing differences exist, teachers need to understand how these differences influence their students’ writing and the effectiveness of that writing [2, p. 26].

In this article, we use a corpus-linguistics approach to investigate how female and male students use adverbs and passive voice in their technical writing. Adverbs and passives contribute to effective communication, but their misuse, according to many textbooks, could impact its clarity and concision. Both are interesting features to study because they fulfill multiple purposes. Adverbs can function as emotive or organizational devices, while passives can create cohesion and objectivity or deliver negative information. Therefore, male and female writers might use the same features for different purposes.

Our sample included 87 student writers, who contributed to a 757,533-word corpus of technical and scientific writing. The results from this study offer pedagogical insights for teaching technical writing, and address the dearth of gender scholarship in the field. Scholars have argued for a more focused study of the ways inclusivity has emerged in technical communication [3]. In addition, the field’s existing gender research is not often replicable, aggregable, or data supported [4], [5]. We acknowledge that gender is only one of a number of variables that could be studied; however, exploring different social variables in isolation contributes important data that can facilitate a more focused, future inquiry. Our study simultaneously contributes data in the presence and application of two language features. In a recent editorial, Graham identified clarity and concision as two “cardinal virtues” of good technical writing, as well as two areas in serious need of data-driven inquiry [6, p. 14].

Practitioner Takeaway

• A small number of previous studies have reported no gender effects in student writing, but linguistics research has suggested that males associate more with direct and dynamic communication, and females with tactful and creative communication.
• This study analyzed a 757,533-word corpus to determine whether gender influences how male and female students use adverbs and passive voice.
• We found that female writers used significantly more adverbs as well as more additive/restrictive, degree, and stance adverbs than expected. Male writers used more linking and manner adverbs than expected.
• Female writers also used significantly more passives, particularly in reporting findings and interpreting. Male writers used passive verbs to describe methods and analyses.

Literature Review

This section begins with a summary of the previous investigations of gender effects across communication contexts. Subsequent results are mixed, a fact that has been attributed to methodological and environmental limitations. In fact, business and technical communication research often finds no significant differences between how females and males write. Gender effects have been said to decrease when writers produce texts that report, explain, persuade, or instruct [7], and we address how writers’ prior learning experiences might influence these developments. Collectively though, these types of studies report only how frequently females and males use specific language features and offer little analysis of how these language features are applied. With respect to this study, females and males may use passives with similar frequencies [8], [9], but we do not know whether gender influences when writers choose to use passives. Furthermore, females may use more intensive adverbs [1], [10], but we do not know how they use adverbs to link ideas. Therefore, we conclude this section with a discussion of these two style features, which serve legitimate purposes in communicating technical information, but their misuse or overuse could signal poor concision and clarity.

Language differences between females and males have been explored in a variety of contexts, including nonverbal, group, dyadic, oral, and written communication, and in a variety of features, including sentence length, emotional language, and tentativeness. Results have been mixed. For every study that reports no differences between females and males (e.g., [8], [9], [11]–[13]), there is a study that reports significant differences (e.g., [1], [14]–[18]). These differences (or lack thereof) have also been summarized in meta-analyses (e.g., [1], [19]).

Newman et al. [20] attributed these contradictory results to the absence of a commonly accepted metric for studying language. Previous studies have also included small samples (i.e., 50 or fewer participants per cell) and a limited amount of data. Data were often subjected to hand-coding, which limited the number of features that could be explored. Researchers often selected language associated with gender stereotypes (e.g., hedges) rather than less obvious features (e.g., pronouns). Nonetheless, these sample limitations have not deterred researchers from making general claims about gender effects, perhaps contributing to a greater divide concerning the existence of any meaningful differences in language by gender [20, pp. 212, 214–215].

However, Tebeaux argued that the reports of gender effects are often unrelated to business and technical communication because job-related contexts differ from the contexts typically studied [2]. In her experience, students with prior “people-intensive work experiences” had internalized how to respond to typical workplace communication situations. These experiences produced androgyny in their writing that responded to readers’ needs in terms of quality and appropriateness. In contrast, writing from students who lacked prior experience often exhibited the gender effects reported in previous research [2, pp. 26–28].

Tebeaux’s observations have some empirical support. It has been hypothesized that male and females practice similar linguistic behaviors in the workplace but then adjust these behaviors in other contexts [21]. Data-driven studies have also identified little to no significant differences in how males and females apply over 20 different conventions to their business communication [8], [9], [12].

Collectively though, the results from most gender effects studies—regardless of the context studied—are limited in the information they provide for writing teachers. Little discussion is devoted to why males and females might write differently, as well as whether there are any differences in how males and females apply the same language features.

Why males and females might write differently is a complex, multidisciplinary topic that encompasses biological, cultural, and social factors. Within the scope of this study, we believe it is important to better understand how prior writing learning has influenced the students who enter college-level courses in business and technical communication. Collecting exhaustive writing history from every student would be an arduous task and would rely heavily on self-report; however, the general findings about how adolescent girls and boys learn to write offers a baseline for understanding the potential differences in language and, therefore, a means for tailoring instruction.

Educational assessment is a research area that offers more consistent and replicable data regarding gender differences in writing. Standardized test results consistently show gender differences across multiple school subjects. Writing remains an area where adolescent boys significantly underperform compared to girls [22]–[24]. Hypothetical standardized test designs that consider different combinations of constructs (e.g., math, science, reading, and writing) suggest that a writing requirement negatively impacts the number of males who graduate high school [25]. On standardized tests for higher or graduate education, adult women tend to score better on both direct and indirect measures of writing skills. In fact, the addition of writing assessments in these tests often balances the gender ratio of high performers, in which there are, otherwise, more men scoring in the top percentile [25, p. 286].

But evidence also suggests that these test designs and subsequent results underestimate the true performance of males (particularly at an early age) and, in turn, actually privilege females. Adolescent boys are often considered analytical writers or writers of plot-driven narratives, while girls are associated with the emotive descriptions common to the approved “literary canon” promoted in most English classrooms [26, p. 97]. Millard described boys as “differently literate,” and the texts they choose to read do not reflect the figurative and descriptive language that they are required to use in their school writing [27]. In turn, evidence suggests that teachers evaluate narratives higher when they perceive the writer as female [28], [29] and that the gender stereotypes associated with writing are often more significant than the actual differences found in writing [30].

Mixed-method approaches have confirmed little difference in the overall quality of writing by adolescent boys and girls but have shown significant differences in the language devices that they choose to achieve this quality. In particular, boys have been observed to follow the writing practices associated with more successful writers [26]. They are categorized as “rapid switchers,” showing a greater tendency to balance between writing and pausing to think about writing than that shown by girls [p. 102]. Similarly, boys have been observed to choose more of the conventions associated with “good” writing [31]. For example, and relevant to the current study, boys used more manner adverbials to create cohesion between paragraphs, while girls preferred proper nouns as their linking device [p. 469].

The specific language patterns that create quality and appropriate technical writing are another complex, multidisciplinary topic. In a recent editorial, Graham wrote that good technical writing is nearly universally understood to be clear, concise, situated within a genre, and directed toward a specific audience. However, research on both clarity and concision is incomplete or anecdotal and, as a result, is “the stuff of technical communication lore” [6, p. 13]. Graham suggested that our definitions of these terms are limited, noting that writing clarity was often determined more by the absence than the presence of certain language features.

Technical communication research includes several empirical illustrations of Graham’s argument. Wolfe reported that style recommendations in 12 popular technical communication textbooks often run counter to the standards used by scientists and engineers [32]. Boettger and Wulff found that technical writing students’ decision to use this as a free-standing pronoun rather than attend it with a noun phrase fulfills specific rhetorical purposes and does not, as many textbooks state, signal poor writing [33]. The study also identified gender as one of the variables that interacts with students’ language choices, returning to the idea that the field needs more research on how students use clarity and concision markers—and how social variables like gender influence these students’ language choices—before we can determine which markers to promote. Addressing this gap is what motivated us to investigate students’ use of adverbs and passive voice in their technical writing.

Adverbs

Adverbs were considered one of “highly decorative or opaque” features used to describe the (now outdated) windowpane theory of language [34]. In technical communication textbooks, adverbs are often used to demonstrate poor concision, which weakens writers’ credibility by exaggerating or overstating claims. For example, the “clutter words” listed in one textbook contain mostly adverbs [35]. The list is accompanied with the following example to support that clutter words unnecessarily stretch the message:

Actually, one aspect of a business situation that could definitely make me quite happy would be to have a somewhat adventurous partner who really shared my extreme attractions to risk. [p. 214, bold in original]

Six of the eight “clutter words” emphasized in this sentence are adverbs.

Although illustrative, the extremeness of this example might not provide writers with a relatable reference point. In fact, actually and really are common amplifiers in conversation [36], but the textbook never acknowledges how adverbs vary across registers and, therefore, might be used differently in conversation than in technical writing. Another textbook cautioned writers to “avoid fillers” and listed mostly adverbs as examples, including basically, essentially, and rather [37]. These authors broadly connect “fillers” to conversation, writing that “such words are common in oral communication, when we need to think fast, but they are meaningless in writing” [37, p. 235].

Technical writing handbooks cover adverbs with more specificity. For example, one noted that intensifiers serve legitimate purposes and necessary functions [38]. The authors recommend the elimination of intensifiers that do not “make an obvious contribution and replace them with specific details” [38, pp. 273–274]. This advice is expanded with a discussion of really and actually, which are common in speech but should be avoided in formal and professional writing [38, p. 468].

Corpus-linguistics research has demonstrated that adverb types vary across registers. Intensifiers are common to both conversation and informational writing, but really is more common to the former register, whereas highly and extremely are more common in the latter. Very is one of the only adverbs common to both conversation and information writing [36]. Similarly, many linking and manner adverbs (e.g., thus, hence, finally, and approximately) are more common to academic writing than to conversation [36]. These distinctions could be particularly relevant to this study, as females have been found to use significantly more intensive adverbs compared to males. (See overview in [1].)

Passive Voice

Passive voice is the most researched clarity marker in technical communication. In her analysis, Wolfe concluded that most technical communication textbooks focus more on warning students against using passives than discussing their legitimate uses [32]. For example, one textbook advised to “use the active voice unless you have a good reason to use passive voice,” highlighting the ethical considerations of using passives to hide the actors’ identity [39, pp. 239–240]. Wolfe correlated this type of advice to the humanities-based training of the textbooks’ authors, which understandably favored the humanities-based guidelines for using active and passive voice. Although technical communication might be a human-centered field, and, therefore, active voice is more frequent and expected, many other disciplines are more object-centered.

On average, passives account for 25% of all finite verbs in academic writing [40]. However, published writing in the STEM fields often includes higher shares of passives. One study found that 67% of the transitive verbs in professional engineering documents were constructed in passive voice [41]. More important, these writers appear to use passives for legitimate purposes. Conrad found that published writers in ecology and civil engineering use passives to focus readers on objects rather than the people who interacted with the objects, as well as to create cohesion between sentences and to stay on topic [42], [43]. Alternate applications of passives create a sense of objectivity in expository prose and relay bad news in a way that does not implicate specific people [40], [44].

In fairness, some technical communication textbooks have updated their discussion of passives. One author acknowledges that readers’ previous writing experiences likely emphasized the avoidance of passive voice. This author argues, ”… we should expose [this] writing boogeyman as a fraud” [45, p. 220]. In regard to possible gender effects, the two studies that we identified in business communication found no significant differences between female’s and male’s use of this language marker [8], [9].

We designed this study around the following research questions:

RQ1. Does gender influence student writers’ use of adverbs?

RQ2. Does gender influence student writers’ use of passive voice?

The phrasing of these questions allowed us to move beyond simply addressing how frequently a particular language feature appeared and, instead, enabled us to conduct a finer-grained examination of how females and males use the same features. The results better inform future gender effects research and suggest strategies for improving teaching.

Methods

This study adopts a corpus-linguistics-based approach to investigate how male and female students use adverbs and passive voice in their technical writing. Corpus linguistics is a modern applied linguistics approach that uses computer-assisted techniques to facilitate large-scale empirical analyses. Results reveal significant language patterns and provide the interpretation of reasons for typical and unusual patterns [46]. Corpus linguistics has been referenced as a useful (and complementary) tool for empirically oriented writing researchers and has been applied to a number of technical text types [33], [43], [47].

Text Corpus

The texts in this study are a subset of the Technical Writing Project (TWP), a corpus of student technical writing [48]. The current inhouse version of the TWP includes more than 6,000 student-written technical texts, including abstracts, white papers, briefing notes, and proposals. The research from the TWP is currently funded by the National Science Foundation (#1708360/#1708362), and all data were collected with Institutional Review Board approval.

The corpus compiled for this study included 1,096 texts, 757,533 words, and 25,097 unique words. The corpus represented a variety of technical text types: 25% were broadly classified as policies and procedures (N = 275), 22% were reports (N = 245), 17% were abstracts (N = 184), 13% were correspondence (N = 145), 11% were white papers and proposals (N = 125), and 11% were job materials (N = 122). Data were collected from 87 students who were enrolled across three different technical communication courses: 33 students enrolled across two sections of a sophomore-level technical writing course, 43 students enrolled across two sections of a senior-level scientific writing course, and 11 students enrolled in a graduate-level professional writing course. Overall, the sample included a relatively balanced gender distribution (53% females and 47% males). In addition, 83% of the students were self-identified as native English speakers and 17% as non-native English speakers. Furthermore, 57% of the students self-identified their ethnicity as White or Caucasian (N = 50), 17% as Asian/Pacific Islander (N = 15), 10% as Black or African American (N = 9), 10% as Hispanic or Latino (N = 9), and 5% as Other (N = 4). Finally, 65.5% of the student writers (N = 57) were enrolled in a STEM major, and 28.7% (N = 30) were in the humanities or business.

The participating students also granted access to their assignment grades, adding a quality measure to our analysis. Overall, the students averaged a grade of 84.48% on these assignments (SD = 7.53). There were no statistically significant grade differences between females (M = 84.86%, SD = 7.38) compared to males (M = 83.28%, SD = 7.94).

Adverb Analysis

To facilitate exhaustive retrieval of the adverbs, we worked with a part-of-speech-tagged version of our data. We searched for all of the words tagged as adverbs in AntConc [49] and copied the resulting concordance into a spreadsheet. The original concordance comprised 24,039 hits, of which 12,111 remained in the sample after manual identification of true hits (few instances were wrongly tagged as adverbs) and exclusion of adverbs denoting time and space, which are not examined here. The 12,111 confirmed hits contained 498 different adverbs. All 12,111 hits were then classified by applying the categories distinguished in the Longman Grammar of English [50]: additive/restrictive, degree, linking, manner, and stance.

• Additive/restrictive adverbs indicate that an item is being added or an action is being restricted [see Examples 1. a–b].
• Degree adverbs suggest the intensity of a characteristic and classified as amplifiers or diminishers [see Examples 2. a–b].
• Linking adverbs make connections between ideas and are classified as expressing apposition, contrast/concession, enumeration/addition, result/inference, summation, or transition [see Examples 3. a–f].
• Manner adverbs express information about how an action is performed (4).
• Stance adverbs express opinion, evaluation, and circumstances and are classified by attitude, epistemic stance, or style [see Examples 5. a–c].

1. a. This will also prevent additional unforeseen costs. (additive)
   b. Such a lesson can only be learned in the workplace … (restrictive)
2. a. Cognitive responses are completely deteriorated … (degree: amplifier)
   b. Curvatures and base size are slightly bigger than expected … (degree: diminisher)
3. a. Specifically, I plan to work … (linking: apposition)
   b. However, by 1975, only … (linking: contrast/concession)
   c. Besides, the courses teach us … (linking: enumeration/addition)
   d. Hence, the existing techniques … (linking: result/inference)
   e. Finally, authors conclude that … (linking: summation)
   f. Similarly, if you find yourself … (linking: transition)
4. HR will formally introduce a policy … (manner)
5. a. The numbers provided in Appendix 1 are curiously low (stance: attitude)
   b. typically … these jobs requires an associate’s degree … (stance: epistemic)
   c. Since this is basically the result of … (stance: style)

These five categories were then organized into subcategories for further analysis. Two people independently classified the data: a graduate student in linguistics, who is a monolingual speaker of American English, and the second author of this paper, who is a bilingual speaker and speaks English as a second language. After an initial interrater agreement of 97%, the remaining instances were discussed until 100% agreement was yielded.

We then ran a distinctive collexeme analysis (DCA) of the data [51]. DCA is a corpus-linguistics-based analysis originally designed to identify words that are distinctly associated with two or more alternating syntactic constructions. It has since been applied in various contexts, including word-sense disambiguation; characterizing differences between language varieties, such as British versus American English; and identifying differences of language use across different time periods (e.g., [52] and [53]). Here, we applied DCA to identify adverbs that were distinctively associated with either female or male writers.

We ran the DCA by using Gries’ script coll.analysis3.2a, written for the programming and statistics environment R [54]. The script takes as its input a list of all the adverbs and information about the gender of the author who produced the adverb; it returns a ranked list of all the adverbs in descending order of their association with female and male writers, as expressed by a Fisher–Yates exact (FYE) test. Since FYE tests return a p-value that can be hard to interpret, the script returns the p-values as negative logged values to the base of 10; converted values higher than 1.3 correspond to a p-value of 5% or lower; values higher than 2 correspond to a p-value of 1% or lower; values higher than 3 correspond to a p-value of 0.1% or lower.

Passives Analysis

To retrieve all instances of passives, we searched for any form of the verb BE (am, be, is, are, were, has been, have been, had been, having been) in AntConc and copied the resulting concordance into a spreadsheet. We manually inspected the 15,437 concordance lines to identify 4,732 true hits of passives.

TABLE I Distribution of Adverb Classes Across Female and Male Writers, Standardized Residuals (z), and Totals

Adverb Class	nfemale (z)	nmale (z)	Totals
additive/restrictive	1,235 (+1.42)	886 (−1.6)	2,121
degree	1,093 (+0.86)	811 (−0.97)	1,904
linking	1,607 (−1.04)	1,342 (+1.17)	2,949
manner	2,133 (−1.3)	1,790 (+1.46)	3,923
stance	705 (+1)	509 (−1.13)	1,214
Totals	6,773	5,338	12,111

To determine whether female and male writers exhibited any verb-specific uses of the passive, we ran a DCA. The input for the script was a list of all the instances of passives together with the information about the gender of the author of each instance. The DCA returned a list of all the verbs in descending order of their association with either female or male writers, as expressed by an again log-transformed FYE test.

Results

This section organizes the results derived by the two research questions.

Results: Adverbs

Our sample included 12,111 adverbs and 498 different adverbs. The most frequent adverbs were only (N = 793, or 6.5% of the sample), well (N = 717, 5.92%), however (N = 701, 5.79%), then (N = 574, 4.74%), and very (N = 471, 3.89%). When examined by classification, 32% of the adverbs were manner, 24% were linking, 18% were additive/restrictive, 16% were degree, and 10% were stance (see Examples 1.–5.). Female writers used more adverbs in their technical writing compared to males (N = 6773 or 56% of the sample compared to N = 5338 or 44%). Similarly, females used a wider range of adverbs (N = 375) compared to males (N = 347).

The results from a chi-square test confirmed that females used adverbs of all classes more often compared to males (see Table I, F = 14.83; df = 4; p < 0.05). However, the standardized residuals for each cell (or the z-values provided in parentheses) suggest a more nuanced understanding of adverb use. Female writers used additive/restrictive, degree, and stance adverbs more often than hypothesized or expected (and conversely, male writers used them less than expected). Male writers used linking and manner adverbs more often than expected.

We further analyzed the adverbs according to their five subclasses. The DCA identified 43 adverbs distinctively associated with female writers and 64 adverbs distinctively associated with male writers. Fig. 1 provides the top 20 most distinctive adverbs for females (in yellow) and males (in green) sorted by the converted FYE values, with the highest values for females on the far left and highest values for males on the far right.

[Figure 1: Adverbs most highly associated with female (yellow) and male (green) writers. — image to be added.]

The distribution of adverbs by classification reiterates the results in Table I. Twelve of the top adverbs distinctive to female writers were additive/restrictive (just, solely), degree (further, minimally, too, fully, widely, enough), or stance (maybe, really, simply, about). Examples 6. a–c illustrate the applications of these three adverb classifications by female writers within various text types.

6. a. Because DSOUTH’s computer systems are meant for business purposes only, any social media usage during work hours should be related to promoting the business. Additionally, it is recommended that … [additive/restrictive, mechanical and energy engineering major in her senior year, policy memo].
   b. The methods are not extremely clear, so another scientist could not double check the experiment. There were very vague directions given to follow up with the experiment [degree, biology major in her senior year, critical review].
   c. While prescription drugs can help slow the overall deterioration of the brain, it is nearly impossible to regain all of the brain cells lost after dementia begins to progress. Art therapy provides a simplistic way to show how the patient is really feeling … [stance, biology major in her senior year, white paper].

In comparison, 17 of the top adverbs distinctively associated with male writers were classified as linking (thus, hence, firstly, moreover, secondly, finally) or manner (environmentally, socially, proactively, inherently, dynamically, reasonably, syntactically, economically, relatively, professionally, semantically). Examples 7. and 8. illustrate the applications of these two adverb classifications by male writers within various text types.

7. The once loosely packed chromatin is now densely packed into their respective sides of the cell in preparation for the cellular division [manner, biology major in his senior year, process description].

8. Two iPSCs were fused and then implanted into a host using in vitro fertilization. Twenty seven live mice pups were produced from tetraploid complementation; more specifically, 3 cells lines produced 27 live pups when the iPSCs were harvested from day 14. There was also no notable difference between iPSC and embryonic stem cells (ESC) for success rates of the tetraploid complementation. Therefore, the choice between iPSCs and ESCs should be … [linking, mathematics major in his senior year, process description].

The results also suggested a pattern within the specific adverbs that were distinctive to the student writers. Really, which was distinctive to female writers, is also one of the most common intensifiers in conversation [36]. The following really examples [see Examples 9. a–c] represent the female writers with different academic backgrounds and within a variety of text types:

9. a. You are the best person for your current job, so I really want to keep you in our company [stance, accounting major in her junior year, correspondence].
   b. The animals were released and were not really monitored afterward [stance, English major in her senior year, white paper].
   c. Other than that, the methods and materials [section] was really well written [stance, biology major in her junior year, critical review].

[Figure 2: Passives most highly associated with female (yellow) and male (green) writers. — image to be added.]

Female writers were also associated with the amplifier too [see Examples 10. a–b] and the downtoner maybe [see Examples 11. a–b], both of which are also common in conversation.

10. a… . If there is not enough insulin in your body, the sugar levels will pile up in your bloodstream, causing you to have too much sugar [degree, biology major in her junior year, process explanation].
    b. I am dedicated, hardworking, I strive to be a role model, and I do not take everything too seriously [degree, criminal justice major in her junior year, LinkedIn profile summary].

11. a. Maybe with the information and influence of the teachings of their school, a child could convince his/her parents that vaccinations are not going to cause adverse health effects [stance, biology major in the second year of her MS program, white paper].
    b. Let me know what is going on, maybe we can move things around or make some changes to better fit your schedule [stance, criminal justice major in her junior year, correspondence].

In contrast, male writers were distinctively associated with very, an intensifier common in both informational writing and conversation, as well as approximately, a common downtoner in academic writing. Male writers were also associated with four common transition adverbs. Thus and hence are frequently used to express a result, and finally and first(ly) are frequently used to list, add, or conclude [36]. In the conclusions section, we speculate on why females might associate with adverbs common to conversation, whereas males appear to use the adverbs common to informational writing. The following examples of transition adverbs represent the male writers with different academic backgrounds and within a variety of text types [see Examples 12. a–c]:

12. a. the budget specified for copying is hundreds of dollars over, thus forcing a new system in place [linking, mechanical and energy engineering major in his sophomore year, correspondence].
    b. Hence, nominal GDP and CE would not be appropriate for trending revenue [linking, math major in his senior year, white paper].
    c. Finally, it shows the fast pathfinder method, which uses a common computer science algorithm to compute the network [linking, computer science and engineering major in his senior year, critical review].

Results: Passives

Our sample included 4,732 true hits of passive voice construction. The corpus comprised 757,533 words; on average, a passive occurred about every 160 words. Females used passives significantly more often than those by males (N_female = 2569; N_male = 2163; p < 0.001). The most frequent passive voice verbs included BE used (N = 446, or 9.43% of the sample), BE done (N = 159, 3.36%), BE made (N = 108, 2.28%), and BE found (N = 100, 2.11%).

The DCA identified 60 verbs distinctively associated with female writers and 54 verbs distinctively associated with male writers. Fig. 2 provides the top 20 most distinctive verbs for females (in yellow) and males (in green) sorted by the converted FYE values, with the highest values for females on the far left and the highest values for males on the far right.

Female writers used passives more often than males and used these passives for different rhetorical purposes. Females were distinctly associated with three of the most common passive voice verbs found in research writing: BE found, BE believed to be, and BE shown [36]. These passive verbs relate to reporting findings as well as interpreting their meaning or connections to other research. In addition, BE found and BE shown are passive verbs that commonly occur in sentences where the agent is not essential, as is also illustrated in Examples 13. a–c.

13. a. Yet, it is actually an unknown that alcohol mixed with energy drinks actually produces a subjective feeling that the individual is less intoxicated and is able to perform regular motor functions properly (Kponee et. al., 2014). This gives the students the idea that they can participate in risky behaviors because they believe they are “sober enough” to have control. In addition, it has been found that students who often mix their drinks are likely to drink more alcohol than those who do not mix (Kponee et. al., 2014) [biology major in her sophomore year, briefing note].
    b. As discussed above, gap junctions, specifically those with connexin43, are believed to be major players in cardiac function [biology major in her senior year, white paper].
    c. Predictive factors of NAS include maternal opiate dose, maternal maintenance agent, secondary exposure to additional substances, gestational age and pharmacogenetics. These factors have been shown to influence the severity and treatment of NAS in infants [biology major in her senior year, white paper].

In contrast, male writers were distinctly associated with two passive verbs commonly used for describing methods and analyses: BE used and BE calculated [see Examples 14. a–b].

14. a. For example, a family could decide that they will not use their dishwasher until it is completely full. This may seem miniscule, but it actually could save 15 to 20 gallons of water per day. The same method could be used to do laundry as well (Dupont, 2013) [construction management master’s student in his first year, white paper].
    b. A company’s revenue is the amount of money the company receives in exchange for its goods and services. Revenue can be calculated by multiplying the amount of goods or services sold and the price the goods and services were sold [math major in his senior year, briefing note].

Interestingly, male writers did not distinctively associate with any passive verbs commonly used for reporting findings, and female writers did not distinctively associate with any passive verbs commonly used for describing methods and analyses (although both used alternate verbs to fulfill these purposes).

Conclusion

Our results indicated that female students used significantly more adverbs and passive voice in their technical writing compared to male students. The adverb results were consistent with at least seven other empirical studies that represented a wide range of communication contexts [1]. However, the passive voice results contradicted the findings from previous studies in business communication [8], [9]. We also identified several instances where these writers used adverbs and passive voice for different purposes.

Female writers used more additive/restrictive, degree, and stance adverbs than expected. Collectively, these adverb types are used to demonstrate the intensity of a characteristic or to make an evaluation. In general, females associate with intensive adverbs across communication contexts, particularly oral communication [55]–[57]. Our female writers are also distinctively associated with three adverbs common to conversation (really, too, and maybe; see Examples 9. a–c, 10. a–b, and 11. a–b, respectively). In contrast, our male writers used more linking and manner adverbs than expected. In fact, 85% percent of males’ most distinctive adverbs were used to connect information or express how an action was performed. There is evidence that adolescent boys prefer manner adverbials to create cohesion between paragraphs [31], so our finding suggests that males retain this preference in adulthood. Finally, male writers did not distinctively associate with adverbs common to conversation, though very is frequent to both written and speech communication [36].

Our writers also used passives differently. Females distinguished with passive verbs were commonly used to report findings, interpret meaning, and make connections to other research (BE found, BE believed, and BE shown; see Examples 13. a–c). Conversely, males distinguished with passive verbs were commonly used to describe methods and analyses (BE used, BE calculated; see Examples 14. a–b). Interestingly, females did not distinguish with any of the passive verbs commonly used to fulfill these functions, nor did males with the passive verbs commonly used to report, interpret, and connect information.

We briefly return to how adolescents typically learn to write as a way to contextualize our results and motivate future research. As discussed in the literature review, the writing instruction in primary and secondary schools favors descriptive narratives that are written in styles more reflective of the approved “literary canon” [26]. Girls appear to easily conform to this approach and earn higher grades and test scores as a result. Boys, on the other hand, prefer to read action-oriented, plot-driven narratives, but the research debates whether they reject the literary styles encouraged by their teachers [26], [58]. If these learning experiences are, in fact, typical, then they might explain why our writers applied adverbs and passive voice differently.

Females used these devices to emphasize and evaluate information, applications that reflect the generalizable claims that females use more indirect, elaborate, and affective language. The female writers also struggled with their adverb choices, associating with three adverbs that were common to conversation but not informational writing. Elaborate and affective language is common to conversation and expository writing, and our finding could be an indicator of how developing female writers transition into writing for the workplace. Female writers’ struggles with adverb choice also provide an indicator of why the gender gap may narrow when writers produce texts that report, explain, persuade, or instruct [7].

In turn, males used these same devices to describe action, methods, and analyses, applications that also reflect the generalizable claims that males use more direct, succinct, and personal language. Male writers’ exposure to and preference for informational text types (as an evidence of their adolescent behavior) could account for the number of distinctive adverbs and passive verbs that are also common to academic and informational writing. Future research needs to investigate the frames of reference that developing writers bring into the technical writing classroom as well as alternate language features that illustrate how developing writers transition to job-related contexts.

Earlier, we suggested that students’ prior writing experiences could establish a baseline for understanding the potential differences in language and, therefore, provide a means for tailoring instruction. Our results suggest that the presence or absence of adverbs or passives did not influence the overall quality and appropriateness of the technical writing. Student writers earned an average of 84.48% on their assignments, and no significant grade differences were found by gender. Gender may not always affect the overall writing quality, but it can inform about the devices that writers choose to achieve this quality [26]. Our results showed, for example, that males preferred to use adverbs to connect information, while females used passives for similar purposes. Technical communication teachers need to understand these patterns and promote language choices in their instruction. Making students aware of how variables like gender could influence their writing makes them more aware of the options that they have in achieving their rhetorical purpose.

Our results also suggest that females and males could struggle with constructing different sections or rhetorical moves in a technical text. If we applied our results to an IMR&D-organized report (introduction, methods, results, and discussion), for example, females might inherently be more comfortable with connecting and evaluating information in introductions and discussions, whereas males might be more comfortable with reporting methods and results. Explaining these possibilities could inform students’ collaborative experiences, improve their comprehension in reading research, and identify potential writing strengths and weaknesses in ways that they have not likely considered.

However, additional research is needed to investigate these claims. Like many quantitative analyses, we investigated a single variable to gain the understanding of its effects. Our results build from previous findings and offer new ways of considering writing instruction, but future research needs to explore whether other variables influence our students’ language choices and how gender might interact with these variables. For example, future research should examine the impact of text type and genre more closely.

Similarly, discipline likely plays a role: academic scholars in rhetoric and composition typically use although and whereas to show contrast and concessions, but scholars in engineering prefer however [59]. Investigations of other language features have also reported that discipline influences writing decisions (e.g., [60]–[63]). Furthermore, gender has been shown to interact with environmental and social variables. Wolfe and Powell found that females and males made an equal number of complaints but made them for different reasons [64]. They concluded that it was the function of these complaints rather than the speaker’s biological sex that informed perception and, thus, contributed to the stereotype that women complain more than men. In another study, Wolfe and Powell reported that gender interacted with ethnicity, finding that European–American women were more likely to focus complaints on themselves, while African–American women were more likely to focus complaints on situational or environmental factors [65].

We conclude by returning to the issue of clarity and concision in technical writing. We selected two language features that fulfill specific rhetorical purposes but whose application could also signal poor communication. The examples provided throughout this article reflect the typical uses of adverbs and passives in the corpus, and we argue that these uses reflect how these same features are used in published writing. However, we concede that the definition of quality technical writing is subjective and that a deliberate language choice made by a writer may appear uninspired to the reader. For example, critics often argue that the frequent use of adverbs like really and very weaken a writer’s credibility. Examples 6. b–c illustrate how some student writers used these same adverbs, but we leave it to the readers to determine whether these examples exaggerate or overstate claims.

Similarly, most literature cautions against using an excessive number of adverbs, but the recommended ratio of adverbs to other parts of speech remains elusive and undefined. Our study included 12,111 confirmed hits and 498 adverb types, which comprised only 1.59% of the 757,533-word corpus. We provide this information to help readers quantify their own definition of excessive, and offer a metric that other researchers and teachers can compare against their corpora. Furthermore, we identified 4,732 true hits of passive voice construction, translating to about one use per 160 words. This finding furthers our belief that students appear to understand when to use features like passives. It is particularly noteworthy here because these applications and the contextual awareness contradict most of the advice provided in technical communication textbooks. Our results provide the evidence of how technical writing students actually use adverbs and passive voice, a fundamental step in moving beyond the vague style advice offered by textbooks and in offering students the ways to understand and vary their language choices.

Acknowledgment

The authors thank N. Lester for helping them tag the adverb data using spaCy (v.1.8.0), an open-source software library for advanced Natural Language Processing in Python and Cython (https://spacy.io/api/tagger).

References

1. A. Mulac, J. J. Bradac, and P. Gibbons, “Empirical support for the gender-as-culture hypothesis,” Human Commun. Res., vol. 27, no. 1, pp. 121–152, 2001.
2. E. Tebeaux, “Toward an understanding of gender differences in written business communications: A suggested perspective for future research,” J. Bus. Tech. Commun., vol. 4, no. 1, pp. 25–43, 1990.
3. N. N. Jones, K. R. Moore, and R. Walton, “Disrupting the past to disrupt the future: An antenarrative of technical communication,” Tech. Commun. Quart., vol. 25, no. 4, pp. 211–229, 2016.
4. R. K. Boettger and C. Lam, “An overview of experimental and quasi-experimental research in technical communication (1992-2011),” IEEE Trans. Prof. Commun., vol. 56, no. 4, pp. 272–293, Dec. 2013.
5. C. Lam and R. K. Boettger, “An overview of research methods in technical communication journals (2012–2016),” in Proc. IEEE Int. Prof. Commun. Conf., 2017, pp. 1–4.
6. S. S. Graham, “Data and lore in technical communication research: Guest editorial,” Commun. Design Quart. Rev., vol. 5, no. 1, pp. 8–25, 2017.
7. J. M. Parr, “Interrogating a mediated corpus of school student writing data for research and practice,” J. Writing Res., vol. 2, no. 2, pp. 129–150, 2010.
8. L. R. Smeltzer and J. D. Werbel, “Gender differences in managerial communication: Fact or folk-linguistics?,” J. Bus. Tech. Commun., vol. 23, no. 2, pp. 41–50, 1986.
9. K. S. Sterkel, “The relationship between gender and writing style in business communications,” J. Bus. Tech. Commun., vol. 25, no. 4, pp. 17–38, 1988.
10. K. A. Hussey, A. N. Katz, and S. A. Leith, “Gendered language in interactive discourse,” J. Psycholinguistic Res., vol. 44, no. 4, pp. 417–433, 2015.
11. B. L. Dubois and I. Crouch, “The question of tag questions in women’s speech: They don’t really use more of them, do they?,” Lang. Soc., vol. 4, no. 3, pp. 289–294, 1975.
12. G. L. Plumlee, A. Wright, and D. N. Wright, “He wrote, she wrote: Gender similarities and differences in written business communication,” J. Bus. Diversity, vol. 16, no. 1, pp. 78–85, 2016.
13. R. Thomson and T. Murachver, “Predicting gender from electronic discourse,” Brit. J. Soc. Psychol., vol. 40, no. 2, pp. 193–208, 2001.
14. S. Argamon, M. Koppel, J. Fine, and A. R. Shimoni, “Gender, genre, and writing style in formal written texts,” Text-Hague Then Amsterdam Then Berlin-, vol. 23, no. 3, pp. 321–346, 2003.
15. D. Biber, S. Conrad, and R. Reppen, Corpus Linguistics: Investigating Language Structure and Use. Cambridge, UK: Cambridge Univ. Press, 1998.
16. M. R. Mehl and J. W. Pennebaker, “The sounds of social life: A psychometric analysis of students’ daily social environments and natural conversations,” J. Personality Soc. Psychol., vol. 84, no. 4, pp. 857–870, 2003.
17. A. Mulac and T. L. Lundell, “Effects of gender-linked language differences in adults’ written discourse: Multivariate tests of language effects,” Lang. Commun., vol. 14, no. 3, pp. 299–309, 1994.
18. A. Mulac, D. R. Seibold, and J. L. Farris, “Female and male managers’ and professionals’ criticism giving: Differences in language use and effects,” J. Lang. Soc. Psychol., vol. 19, no. 4, pp. 389–415, 2000.
19. J. S. Hyde, “The gender similarities hypothesis,” Amer. Psychol., vol. 60, no. 6, pp. 581–592, 2005.
20. M. L. Newman, C. J. Groom, L. D. Handelman, and J. W. Pennebaker, “Gender differences in language use: An analysis of 14,000 text samples,” Discourse Process., vol. 45, no. 3, pp. 211–236, 2008.
21. M. R. Key, Male/female Language: With a Comprehensive Bibliography. Lanham, MD, USA: Scarecrow Press, 1996.
22. L. V. Hedges and A. Nowell, “Sex differences in mental test scores, variability, and numbers of high-scoring individuals,” Science, vol. 269, no. 5220, pp. 41–45, 1995.
23. J. R. Pargulski and M. R. Reynolds, “Sex differences in achievement: Distributions matter,” Personality Individual Differences, vol. 104, pp. 272–278, 2017.
24. J. Y. Fearrington, P. D. Parker, P. Kidder-Ashley, S. G. Gagnon, S. McCane-Bowling, and C. A. Sorrell, “Gender differences in written expression curriculum-based measurement in third through eighth grade students,” Psychol. Sch., vol. 51, no. 1, pp. 85–96, 2014.
25. W. W. Willingham and N. S. Cole, Gender and Fair Assessment. Abingdon, UK: Routledge, 2013.
26. S. Jones, “Composing in the style of Mozart: An exploration of the “struggling boy writer” comparing the composing processes and strategies of boys and girls,” Engl. Teach., vol. 6, no. 1, pp. 96–112, 2007.
27. E. Millard, Differently Literate: Boys, Girls and the Schooling of Literacy. Abingdon, UK: Routledge, 2002.
28. S. Peterson, “Evaluation and teachers’ perceptions of gender in sixth-grade student writing,” Res. Teach. Eng., vol. 33, no. 2, pp. 181–208, 1998.
29. S. Peterson and K. Kennedy, “Sixth-grade teachers’ written comments on student writing: Genre and gender influences,” Written Commun., vol. 23, no. 1, pp. 36–62, 2006.
30. B. Francis, B. Read, L. Melling, and J. Robson, “University lecturers’ perceptions of gender and undergraduate writing,” Brit. J. Sociol. Educ., vol. 24, no. 3, pp. 357–373, 2003.
31. S. Jones and D. Myhill, “Discourses of difference? Examining gender differences in linguistic characteristics of writing,” Can. J. Educ., vol. 30, pp. 456–482, 2007.
32. J. Wolfe, “How technical communication textbooks fail engineering students,” Tech. Commun. Quart., vol. 18, no. 4, pp. 351–375, 2009.
33. R. K. Boettger and S. Wulff, “The naked truth about the naked this: Investigating grammatical prescriptivism in technical communication,” Tech. Commun. Quart., vol. 23, no. 2, pp. 115–140, 2014.
34. C. R. Miller, “A humanistic rationale for technical writing,” Coll. Engl., vol. 40, no. 6, pp. 610–617, 1979.
35. J. M. Lannon and L. J. Gurak, Technical Communication, 14th ed. New York, NY, USA: Pearson, 2016.
36. S. Conrad, D. Biber, K. Daly, and S. Packer, Real Grammar: A Corpus-based Approach to English. New York, NY, USA: Pearson/Longman, 2009.
37. M. Markel and S. A. Selber, Technical Communication, 12th ed. Boston, MA, USA: Bedford/St. Martin’s, 2018.
38. G. J. Alred, C. T. Brusaw, and W. E. Oliu, Handbook of Technical Writing, 10th ed. Boston, MA, USA: Bedford/St. Martin’s, 2012.
39. P. V. Anderson, Technical Communication: A Reader-centered Approach, 8th ed. Boston, MA, USA: Wadsworth 2013.
40. D. Biber, S. Conrad, and G. N. Leech, Longman Student Grammar of Spoken and Written English. London, UK: Longman 2002.
41. D. Ding, “Object-centered—How engineering writing embodies objects: A study of four engineering documents,” Tech. Commun., vol. 48, no. 3, pp. 297–308, 2001.
42. S. Conrad, “Investigating academic texts with corpus-based techniques: An example from biology,” Linguist. Educ., vol. 8, no. 3, pp. 299–326, 1996.
43. S. Conrad, “The use of passives and impersonal style in civil engineering writing,” J. Bus. Tech. Commun., vol. 32, no. 1, pp. 38–76, 2018.
44. C. D. Rude and A. Eaton, Technical Editing, 5th ed. (The Allyn & Bacon Series in Technical Communication). Boston, MA, USA: Longman, 2011.
45. R. Johnson-Sheehan, Technical Communication Today, 3rd ed. New York, NY, USA: Longman, 2010.
46. S. Conrad, “Corpus linguistic approaches for discourse analysis,” Annu. Rev. Appl. Linguist., vol. 22, pp. 75–95, 2002.
47. T. Orr, “Introduction to the special issue: Insights from corpus linguistics for professional communication,” IEEE Trans. Prof. Commun., vol. 49, no. 3, pp. 213–216, Sep. 2006.
48. R. K. Boettger and S. Wulff. 2018. The Technical Writing Project. [Online]. Available: http://www.technicalwritingproject.com/
49. L. Anthony, “Developing AntConc for a new generation of corpus linguists,” in Proc. Corpus Linguist. Conf., 2013, pp. 14–16.
50. D. Biber, S. Johansson, G. Leech, S. Conrad, E. Finegan, and R. Quirk, Longman Grammar of Spoken and Written English. Cambridge, MA, USA: MIT Press, 1999.
51. S. Th. Gries and A. Stefanowitsch, “Extending collostructional analysis: a corpus–based perspective on ‘alternations,’” Int. J. Corpus Linguist., vol. 9, no. 1, pp. 97–129, 2004.
52. S. Wulff, A. Stefanowitsch, and S. Th. Gries, “Brutal Brits and persuasive Americans,” in Aspects of Meaning Construction. Amsterdam, The Netherlands: John Benjamins, 2007, pp. 265–281.
53. M. Hilpert, “Distinctive collexeme analysis and diachrony,” Corpus Linguist. Linguist. Theory, vol. 2, no. 2, pp. 243–256, 2006.
54. S. Th. Gries, “Coll.analysis 3.2. A program for R for Windows 2.x,” 2007.
55. F. Crosby and L. Nyquist, “The female register: An empirical study of Lakoff’s hypotheses,” Lang. Soc., vol. 6, no. 3, pp. 313–322, 1977.
56. A. Mulac and T. L. Lundell, “Linguistic contributors to the gender-linked language effect,” J. Lang. Social Psychol., vol. 5, no. 2, pp. 81–101, 1986.
57. L. H. Turner, K. Dindia, and J. C. Pearson, “An investigation of female/male verbal behaviors in same-sex and mixed-sex conversations,” Commun. Rep., vol. 8, no. 2, pp. 86–96, 1995.
58. S. Cavazos-Kottke, “Tuned out but turned on: Boys’(dis) engaged reading in and out of school,” J. Adolescent Adult Literacy, vol. 49, no. 3, pp. 180–184, 2005.
59. L. Wilder and J. Wolfe, “Sharing the tacit rhetorical knowledge of the literacy scholar: The effects of making disciplinary conventions explicit in undergraduate writing about literature courses,” Res. Teach. Engl., vol. 44, no. 2, pp. 170–209, 2009.
60. A. Rozumko, “Adverbial markers of epistemic modality across disciplinary discourses: A contrastive study of research articles in six academic disciplines,” Studia Anglica Posnaniensia, vol. 52, no. 1, pp. 73–101, 2017.
61. K. Hyland, Disciplinary Discourses: Social Interactions in Academic Writing. Ann Arbor, MI, USA: Univ. Michigan Press, 2004.
62. C. Samson, ” … Is different from … : A corpus-based study of evaluative adjectives in economics discourse,” IEEE Trans. Prof. Commun., vol. 49, no. 3, pp. 236–245, Sep. 2006.
63. M. Peacock, “Linking adverbials in research articles across eight disciplines,” Ibérica, vol. 20, no. 20, pp. 9–34, 2010.
64. J. Wolfe and E. Powell, “Gender and expressions of dissatisfaction: A study of complaining in mixed-gendered student work groups,” Women Lang., vol. 29, no. 2, pp. 13–20, 2006.
65. J. Wolfe and E. Powell, “Biases in interpersonal communication: How engineering students perceive gender typical speech acts in teamwork,” J. Eng. Educ., vol. 98, no. 1, pp. 5–16, 2009.