The second research question is: Are there any statistically significant differences in the mean written syntactic complexity scores across disciplines and between registers as measured by twelve individual phrasal and clausal syntactic complexity features? To answer this question, normalized mean frequencies for the discrete linguistic features were used to perform a two-way between-groups ANOVA. Main effects of discipline and register and their interaction were reported.
Main effect of discipline on the dependent variables
As the factorial design of the study suggests, discipline with three levels and register with two levels are the independent variables in this study. Dependent variables (linguistic features) were examined to see whether they differ across the disciplines and between the registers. The overall findings from the factorial ANOVA procedure revealed that discipline as an independent variable resulted in more variance in accounting for the differences of linguistic features. That is, discipline was a significant factor in determining the variance of almost all nominal modification features. More specifically, across the 12 nominal modification features, only two features (adjectives as prenominal modifiers and nonfinite relative clauses) did not distinguish across the disciplines. As depicted in figures displaying profile plots of the disciplines, 7 out of 12
linguistic features followed a similar path of variation across the disciplines. These linguistic features were finite adverbials, of as a preposition, WH and that relative clauses, finite and nonfinite complement clauses, and nominalizations. In this path shared by these linguistic features, applied linguistics appeared as the discipline that contained the greatest use of these linguistic features whereas agronomy had the least use. Industrial and manufacturing engineering stood in between the two.
On the other hand, 4 out of 12 features showed a different path of variation across the disciplines. In that, nouns as part of speech, adjectives and nouns as nominal premodifiers, and nonfinite relative clauses appeared more frequently in agronomy, followed by industrial and manufacturing systems engineering and applied linguistics respectively. To summarize, it can be concluded that discipline resulted in statistically significant variation for a great majority of linguistic features (10/12) and two distinct paths of disciplinary variation emerged depending on the linguistic features as described above. In the first path, applied linguistics included far greater
use of the mentioned linguistic features, and in the second part agronomy was the leading discipline in terms of the use of those linguistic features. Industrial and manufacturing systems engineering stayed in the middle in both of the disciplinary variation paths.
Main effect of register on the dependent variables
While discipline emerged as a relatively distinguishing factor, register did not account for the variance in the nominal modification features as much as discipline did. In fact, the results showed that only three linguistic features showed variance based on the register. Finite adverbial clauses, nonfinite complement clauses, and nouns as part of speech were found to differ
significantly between published research articles and disciplines. This finding is not surprising considering that both registers share quite common characteristics. As seen in the post hoc comparison table and profile plots for register, master’s theses made greater use of finite adverbial clauses and of nonfinite complement clauses than did published research articles.
These findings could lend support to the previous literature that found that dependent clauses tend to decrease at advanced stages of writing (Nippold, 2007; Scott, 2004). If published research articles are considered as a more advanced sample of academic writing than master’s theses, considering the notable professional experience of the research articles authors, it is not surprising to observe a decrease in finite adverbial clauses in research articles. As noted in situational characteristics of the corpus registers in Chapter 3, research articles and master’s theses can be distinguished in terms of the participants. That is, while master’s theses are written by a single author, a great majority of research articles are written by more than a single author. In fact, it is hard to find a single-author research article in some disciplines (e.g., agronomy). Through collaboration, the authors in research articles can combine their expertise on the same article. Moreover, most writers of research articles are professionals who hold terminal degrees
(e.g., Ph.D.) with higher numbers of years of training in their fields than writers of master’s theses, who are relatively newcomers to the academy. Thus research articles could be considered as a product reflecting greater expertise compared to master’s theses. Nouns, however, showed an inverse trend in that nouns appeared more frequently in published research articles (M=339) than in master’s theses (M=329). These numbers are comparable to the findings from previous research that reported that nouns are increasing tremendously in modern day academic writing (e.g., Biber & Gray, 2013).
Despite the statistically significant difference, both registers carried a large number of nouns. This noticeable reliance on nouns in the registers such as published journal articles can be attributed to the fact that nouns are key devices in constructing informational density as they are loaded with heavy content information. As published journal articles are restricted by space, it is plausible to utilize nouns maximally to achieve maximum content under severe space limitations. The fairly limited page length of research articles as demonstrated in register analysis in Chapter 3 could be a contributing factor for the prevalence of nouns in research articles. This finding provides backing for Halliday’s (1998, 2004, 2009) claim that as nouns increase in discourse, there is a greater number of lexical items per clause, leading to higher lexical density, which is defined as the ratio of content words to grammatical words.
Interaction of discipline and register on the dependent variables
The results of factorial ANOVA suggested that the interplay of discipline and register did not result in much impact on nominal modification features. Only three linguistic features had a statistically significant difference according to the combined effect of discipline and register. One of the striking findings related to the interaction of discipline and register emerged in adjectives as prenominal modifiers. The disciplines showed an opposite path in relation to
registers. While master’s theses and published research articles contained fairly similar numbers of adjectives as nominal premodifiers in agronomy and in industrial and manufacturing systems engineering, the difference in applied linguistics was considerably more substantial.
Nominalization is another linguistic feature that displayed an intricate relationship of discipline and register. As shown in Figure 9 in Chapter 4, industrial and manufacturing systems engineering had the highest variability between research articles (M=91.22) and master’s theses (M=74.04) with regard to the use of the nominalizations. Conversely, the variation between the registers in agronomy and applied linguistics was quite small. Given the increased use of nominalizations in engineering research articles, the finding may correspond to the earlier research that described the language of science and academia as nominal rather than verbal (e.g., Banks, 2005; Biber & Gray, 2010; Byrnes, 2009; Fang et al., 2006; Flowerdew, 2003; Halliday, 1998, 2004; Halliday & Martin, 1993).
This finding pertaining to the prevalence of nominalizations in engineering research articles could account empirically for the increased use of nominalized processes in science discourse, considering the fact that the language of science transforms actions into nouns and generates simple clauses but long and complex nominal groups (Halliday, 2009). Despite this supporting evidence for the spread of nominalizations in engineering, agronomy was particularly interesting due to its contradictory findings. In fact, both published research articles (M=75.00) and master’s theses (M=75.94) in agronomy had a rather smaller rate of nominalization
occurrence than did engineering. To some extent, the decreased nominalization in agronomy could be explained by the earlier findings that found that nominalizations are more common in social science and multi-disciplinary science articles (Biber and Gray, 2016). What may be happening is that the greater use of nominalizations in engineering over agronomy could partly
be explained by the inter-disciplinary nature of industrial and manufacturing systems engineering as mentioned at the beginning of Chapter 5. Further research would help clarity these issues.
5.3 Research Question 3: Summary and Discussion of the Major Findings