CHAPTER 2. LITERATURE REVIEW
2.4 Nominal Modification and Phrasal Complexity in Academic/Scientific Discourse
2.4.1 Integrating General and Specific Syntactic Complexity Measures
As described in section 2.2, T-unit measures are based on an approach favoring clause lengthening through subordinate clauses as well as any other nonclausal structure and generate a holistic score. However, this holistic score does not reveal if the T-unit length is influenced by dependent clauses or other non-clausal structures (Norris & Ortega, 2009; Scott & Balthazar, 2010). For this reason, reporting holistic scores poses a challenge to determine the linguistic bases of the complexification. For example, some studies have observed an increase in the mean length of T-unit (MLTU), but a decrease in subordination in the higher levels of writing (e.g., Byrnes, Maxim, & Norris, 2010; Nippold et al., 2005). This means that the T-unit is possibly
lengthened by phrases or subclausal constituents rather than by finite dependent clauses at the higher levels of language proficiency. Nonetheless, the holistic measurement of MLTU may not help researchers find out which linguistic features are leading to the increase in the MLTU values (Scott & Balthazar, 2010).
Building on this relatively limited functionality of holistic measurement, Biber and his colleagues have argued for a linguistically motivated register approach enabling the investigation of complexity through individual lexico-grammatical features as opposed to the mainstream general complexity measures which yield holistic scores “regardless of the specific grammatical devices used to create the longer structure” (Biber et al., 2014, p. 10). The corpus-based
approach has maintained that phrasal features and non-finite clauses deserve to be included in complexity measurement as being among the integral components of academic writing. Their main argument is that grammatical complexity is a multifaceted construct; accordingly, alternative measures are needed to address the complexity manifested by phrasal features and non-finite clauses (e.g., adjectives and noun as nominal premodifiers, PPs as nominal
postmodifiers). Their approach is grounded in the examination of specific linguistic features in the texts as opposed to the holistic scores produced by mainstream complexity measures.
Mainstream SLA-based measures examine the complexity at the sentence, clause or T- unit level, advocating that “subordination finds its representation and bears its meaning at the sentence level” (Yang, 2013, p. 189). The result of the T-unit analysis is generally a ratio or average number that represent the whole grammatical complexity of the text. Even though sentence length is attributed as the primary determinant of syntactic complexity in most indices, it cannot account for the whole phenomenon of linguistic complexity. In contrast to longstanding measures of syntactic complexity favoring clause length via subordination, a corpus-based
approach treats complexity by examining individual/discrete linguistic features mostly at the subclausal level and reports rates of occurrences per linguistic feature. Despite their reductionist perspective to complexity, dependent clauses are still claimed to be key indicators of syntactic growth and continue to be used in studies. While traditional global complexity measures have a limited scope of complexity, ignoring the subclausal constituents of syntax, they have been used for a long time consistently in first and second language complexity research as one of the most objective and reliable means of assessing syntactic complexity (Frear & Bitchener, 2015; Jiang, 2013). However, due to the controversies in the definition of T-unit and the mixed results of the studies employing T-unit measures, there has been a growing call for the inclusion of phrase- level complexity measures along with clausal features (Biber et al., 2011; Biber et al., 2013; Norris & Ortega, 2009; Ravid & Berman, 2010; Staples et al., 2016). In response to these calls, only a handful of studies have incorporated phrasal complexity features in their analyses of syntactic complexity (e.g., Lu, 2011; Mazgutova & Kormos, 2015; Staples et al., 2016; Vyatkina et al., 2015; Yang et al., 2015).
In a recent study on specific phrase level measurement of complexity, Vyatkina et al. (2015) explored how the size and range of syntactic modifiers changed over two years of instructed language study. Their study was based on 185 learner texts coming from 12
participants in the Falko L2 German corpus. They analyzed attributive adjectives, predicative and adverbial adjectives, PPs, adverbial clauses, and relative clauses. The results of their study suggested that there is not a uniform pattern for modification as syntactic development. The size and range of the modifiers mentioned above did not change much over the observed period. The findings of the analysis showed a slight increase in attributive adjectives. However, the authors could not verify if each individual learner showed an increase. One interesting finding of the
modifier analysis was that predicative adjectives (e.g., This is wonderful) showed a noticeable decrease over time. This can be meaningful as predicative adjectives occur mostly in clauses and do not directly modify a noun.
Mazgutova and Kormos (2015) conducted a similar study that investigated lexical and syntactic characteristics of students’ academic writing change. The study was carried out with pre-sessional EAP (English for Academic Purposes) students over an intensive four-week course. The authors reported that they merged global measures of syntactic complexity (e.g., mean length of T-unit, the mean number of dependent clauses per T-unit) with specific indices (e.g., the ratios of conditional clauses, relative clauses, prepositional phrases, and infinitive clauses as noun postmodifiers) in their study. Their findings revealed that lower proficiency students showed an increase in their use of complex noun phrases. However, syntactic change at the clause level did not differ much between the higher and lower proficiency groups with one exception. In that, the authors observed a decrease in the use of infinitive clauses among the higher proficiency group. Mazgutova and Kormos, based on this reduction in clausal complexity and the increase in the frequency of words found in the academic word list, pointed out that the students tended to use more nominalizations. However, pre-and post-modification of nouns was not observed to increase with the higher proficiency learners.
Yang et al. (2015) similarly counted nonfinite elements as clauses and added noun phrase complexity as a separate complexity measure to investigate the phrasal complexity of the texts. They found that mean length of sentence and mean length of T-unit worked well as global measures to predict writing scores. Lu (2011) assessed measures to evaluate particular structures via complex nominals per clause and T-units as well as nonfinite verb phrases. His complete list
of measures encompassed major dimensions of syntactic complexity such as length of production, subordination, coordination, and phrasal elaboration.
In summary, the nominal style of academic writing has been the focus of attention in the increasing number of studies reviewed here. However, recent investigations of nominal
characteristics of writing across different proficiency levels have produced conflicting results in terms of the use of these resources in different proficiency levels (e.g., Mazgutova & Kormos, 2015; Vyatkina et al., 2015). Hence, more studies are needed to ascertain the results of
conflicting studies. In addition, since disciplines vary substantially in their investigations and conventions, deployment of nominal groups might differ across disciplines. Thus, the next section briefly focuses on nominal group in terms of disciplinary differences.