A Corpus-Based Study of Noun Modification in English Academic Writing

Abstract

A characteristic feature of English academic writing is the use of complex nominal groups, and the complexity of nominal groups is mainly achieved through nominalization and noun modification. To inform the teaching of academic English to Chinese EFL learners, this study investigated the distributions of different types of noun modifications in English academic writing. Using a corpus-based approach, this study compared head nouns, noun modifiers, and nominal group length across three groups of academic writers: published research article (RA) writers and Chinese EFL master students (CMA) and Chinese EFL doctoral students (CPhD) in linguistics. The findings are as follows: (1) RA writers employed first-person pronouns more frequently to construct authorial stance and engagement; (2) RA writers used more clausal modifiers than Chinese EFL students, which contributed to longer nominal groups; (3) RA writers demonstrated higher proficiency of English through a greater use of higher-stage noun modifiers. These findings suggest that pedagogical efforts should guide Chinese EFL students toward producing more native-like academic texts.

Plain Language Summary

Noun Modification in English Academic Writing

This study examined how different academic writers use and modify nouns in English. We compared research article (RA) writers with Chinese postgraduate students of linguistics to better understand their differences and improve academic writing instruction. The two groups we analyzed were professional linguists who publish research articles and Chinese master’s and doctoral students in linguistics. The main findings are as follows. First, Chinese students used fewer personal pronouns such as “I” and “we” than RA writers. As a result, their writing sounded less personal and sometimes more rigid. Second, RA writers more often used full clauses to modify nouns (e.g., “the theory that the researcher proposed”). This made their noun phrases longer and their writing more detailed. Third, RA writers used a wider and more complex range of noun modifiers, which suggests a higher level of writing proficiency. Overall, the study suggests that the goal of teaching EFL academic writing should not be to promote a fixed and overly formal style. Instead, students should learn to use different structures flexibly so that they can write clear and effective academic English.

Keywords

academic writing EFL nominalization noun modification nominal groups

Introduction

English academic writing has been a major focus of research over the past few decades (e.g., Banks, 2008; Fang et al., 2021; Hyland & Jiang, 2017; Shao et al., 2022). This attention stems not only from its distinctive features compared to other forms of English, but also from its practical benefits: proficiency in academic English serves as a gateway to successful academic publishing and to making effective contributions across both hard and soft sciences. Analyses of academic writing suggest that its most prominent grammatical characteristics include longer sentences, heavier subordination and/or coordination, and a greater use of nominalization and clausal embedding (e.g., Brown & Yule, 1983; Chafe, 1982; Halliday, 2004).

However, the notion that academic writing is characterized by longer sentences and heavier subordination and/or coordination is subject to misunderstanding and challenge. For instance, longer sentences and clausal embedding are dominant in spoken texts, particularly in spontaneous spoken language (Halliday, 1985). Some studies (e.g., Bardovi-Harlig, 1992; Rimmer, 2006) further indicate that academic writers with higher English proficiency do not necessarily produce longer sentences or employ clausal embedding. Rather, the grammatical complexity of English academic writing is argued to reside in the use of complex nominal groups or noun phrases, which enable the packaging of compact, lexically dense and more abstract content (Fang et al., 2006; Lei et al., 2023).

Complex nominal groups are formed through two primary grammatical processes: nominalization and noun modification. Both are characteristic of English academic writing (Banks, 2005; Charles, 2003; Galve, 1998; Halliday, 1998). With the nominalization of verbal groups, participant nominal groups are correspondingly adjectivalized to function as modifiers (He, 2019). See example (1) quoted from Halliday and Matthiessen (1999, p. 343):

(1) a. They shredded the documents before they departed for the airport.

b. Their shredding of the documents preceded their departure for the airport.

The two verbal groups shredded and departed in (1a) are nominalized as shredding and departure respectively in (1b). The personal pronoun they functioning as participants of the two verbal groups in (1a) are correspondingly shifted to the possessive pronoun their functioning as modifiers of the two nominalizations in (1b). The participant nominal group the documents is shifted as a postmodifier. The circumstantial prepositional phrase for the airport is also shifted to function as a postmodifier. These shifts result in longer nominal groups and a reduction in the number of clauses, thus creating a higher lexical density overall.

English as a Foreign Language (EFL) academic writing typically differs from that of proficient English native speakers (ENS). For example, EFL academic writers tend to use fewer participle clauses (Granger, 1997) and fewer verbalizations (He & Yang, 2018) than proficient ENS academic writers. Moreover, the two groups of academic writers demonstrate different uses of first-person pronouns (Martínez, 2005). Studies (e.g., Ansarifar et al., 2018; He & Guo, 2021; Lan & Sun, 2019) also show that the use of noun modifiers may also reflect differences in the proficiency of English academic writing.

The investigation of noun modifier usage in EFL and ENS academic writing holds significant relevance for academic writing pedagogy in the Chinese EFL context. This is essential for preparing Chinese EFL learners to contribute to academia through the medium of English. The hypothesis underlying the research reported in this paper is that proficient ENS academic writers write longer and more complex noun phrases in their academic writing than Chinese EFL academic writers. Specifically, this study sought to address the following research questions:

What differences can be identified in the use of head nouns between proficient ENS academic writers and Chinese EFL writers?

What differences can be identified in the use of noun modifiers between proficient ENS academic writers and Chinese EFL writers?

This study contributes to the existing research on noun phrase complexity in three key respects. First, it extends the analysis to a systematic comparison of noun phrase patterns in Chinese EFL dissertation abstracts with those in published research article abstracts. Second, it provides a fine-grained, corpus-driven profile of the distribution of noun modifiers, revealing how different types of noun modifiers function distinctly across developmental and expert writing. Third, it offers empirical implications for Chinese EFL postgraduate academic writing pedagogy, highlighting specific syntactic features that may be targeted to support Chinese EFL learners’ genre enculturation and communicative effectiveness in international publication contexts.

The Nature of Academic Writing

The Concept of Grammatical Complexity

Research on English academic writing has predominantly focused on specific types of texts, such as research articles (e.g., Gao & He, 2023; Jalilifar et al., 2017; Wu et al., 2020), textbooks (e.g., Biber, 2006; Freddi, 2005; Jalilifar et al., 2014), and book reviews (e.g., Groom, 2005), in order to identify the distinctive features of this register. This is an important task for academic research and has practical benefits, particularly for learners of academic English (e.g., Hewings, 2001; Ouyang et al., 2022; Tao et al., 2025).

One salient feature of academic English is grammatical complexity. Simply put, grammatical complexity refers to the elaborate use of language and the varied deployment of syntactic patterning (Lu & Ai, 2015). This elaboration has often been defined in terms of longer sentences with a dense use of subordinate clauses (Hughes, 1996). However, the use of clausal subordination as a measure of grammatical complexity in English academic writing has been challenged. Research indicates that sentence structures in academic English tend to be simpler, and spontaneous conversation actually contains more subordinate clauses than academic texts (Halliday, 1985). In fact, both academic texts and conversations are complex, but their complexities manifest differently within the grammar. That is, academic texts are typically complex in noun phrases, whereas spoken texts tend to be complex in clause structure (Biber & Gray, 2010).

Halliday (1985) distinguishes between these two types of complexity by using the term “lexical density” to refer to the ratio of content words in a clause, and “grammatical intricacy” to describe the dynamic linking of clauses through coordination or subordination to form clause complexes (cf. Halliday & Matthiessen, 2014). Therefore, rather than using clausal subordination as an indicator of grammatical complexity, complex noun phrases have been identified as a more accurate measure of grammatical complexity in English academic writing (Biber et al., 2011). A study by Biber et al. (2016) on the writing of L1 English university students at different academic levels found that clausal features decreased while phrasal features increased as the students progressed academically. Many other studies (e.g., de Haan, 1989; Larsson et al., 2023; Lorenzo & Rodríguez, 2014) also recognized a greater reliance on complex noun phrases in academic writing.

The evolution of modern English academic writing is characterized by the use of complex noun phrases rather than complex clause structures (Biber & Gray, 2016). Noun phrases serve as the primary resource for increasing lexical density, enabling the relatively economical packaging of a considerable amount of information into a limited number of clauses (Halliday, 1993). Nevertheless, within the field of English for academic purposes, grammatical complexity is commonly operationalized through four key parameters: length, ratio, index, and frequency, with length being the most frequently employed measure (Norris & Ortega, 2009). What this parameter is related to is the mean length of sentence, the mean length of clause and the mean length of T-unit (a main clause along with any subordinate clauses) (Ortega, 2003). Given that academic English is distinguished by high lexical density achieved through nominalization, it is necessary to investigate the mean length of nominal groups as a measure of grammatical complexity. Doing so could provide a clearer understanding of the grammatical nature of written academic English.

Nominalization and Noun Modification

Nominalization is one of the most prominent linguistic features of present-day English academic writing (He & Zhang, 2024; McGrath & Liardét, 2023). This process entails a grammatical shift towards greater abstraction. For instance, a verbal group realizing a process can be nominalized to construe the thing within a participant nominal group, thereby allowing both pre- and post-modification. Consider the verbal group manages in the clause The council manages hazardous waste. The nominalization of verbal group manages results in The council’s hazardous waste management, which can then function as a participant in another clause, such as The council’s hazardous waste management is very effective. Here the original participant nominal group the council has now shifted to become a modifier of the nominalization head noun management, resulting in a longer and lexically denser nominal group.

Premodifiers can be realized as adjectives, participles or nouns, whereas postmodifiers can be realized as finite or non-finite relative clauses, prepositional phrases or appositive noun phrases. Nominalization and the resulting modification patterns of this kind are prevalent in English academic writing but relatively rare in spontaneous spoken English. According to Biber et al. (1999), nearly 60% of all noun phrases in academic writing contain at least one modifier, whereas in conversational English the figure drops to about 15%. Regarding the use of nominalization, He and Yang (2018) found that academic writers within the field of medical sciences use more nominalizations than those within the field of linguistics.

Given that spoken language development precedes written language, Biber et al. (2011) posited that the characteristic features of conversation are acquired earlier and are therefore easier to produce than those of academic writing, which typically require formal instruction. Building on this premise, Biber et al. (2011) hypothesized a 5-stage developmental framework of grammatical features that can be applied to the development of the linguistic competence of the L2 English learners. The framework reflects the progression from conversational competence towards academic writing competence in L2 learners. Structures that are easier and are acquired the earliest are positioned at Stage 1. Such structures are typical of conversational English. Structures that are more difficult and are acquired later are typically characteristic of academic writing and are placed at Stage 5. The framework also displays the relative complexity of noun modifiers. For example, an attributive adjective modifying a head noun (e.g., an interesting story) represents a Stage 1 complexity feature, whereas a nonfinite clause functioning as the complement of a preposition (e.g., the idea of opening the door) exemplifies a Stage 5 complexity feature.

Methodology

Corpus

The corpus used in this study is a specialized, self-compiled collection of academic abstracts from three groups of writers. As the abstract of a typical research article serves as a concise summary which is conventionally structured around key rhetorical moves such as introducing the topic, stating the purpose, describing the method, presenting results, and drawing conclusions (Hyland, 2000), writers are compelled to achieve maximum efficiency, clarity, and economy within a limited word count (Swales & Feak, 2009). Biber and Gray (2011) observed that noun phrases significantly contribute to the compressed style characteristic of academic abstracts. Gray (2013) further noted that abstracts exhibit the densest use of phrasal features, which serve as a strong indicator of noun phrase complexity (Gray, 2013). Moreover, because abstracts consist exclusively of full sentences without non-linguistic insertions (e.g., figures or tables), they provide a reliable basis for analyzing clause constituents.

We collected 30 abstracts from linguistics theses written by Chinese master students (CMA) and 20 abstracts from linguistics dissertations authored by Chinese doctoral students (CPhD), all retrieved from the CNKI Digital Dissertations database. The selection of these two groups of English learners complements previous research that primarily focuses on English learners with lower or intermediate proficiency (e.g., Biber et al., 2020; Durrant & Brenchley, 2023). We concentrated on the discipline of linguistics because, in mainland China, nearly all masters’ theses and PhD dissertations are written in Chinese, with English-related disciplines (such as linguistics and English literature) being notable exceptions. For comparative analysis, 70 abstracts were also collected from recently published research articles (RA) in linguistics. This comparative design follows the methodological precedent established by Xie et al. (2024), who investigated diachronic changes in authorial stance in the discussion sections of Chinese master theses and published research articles. These research articles were sourced from leading international linguistics journals such as Applied Linguistics, Language, Lingua, Linguistic Inquiry, and Linguistics. All these documents are publicly accessible.

The reason for selecting these five journals is that they are widely recognized as leading international journals in the field of linguistics, representing diverse theoretical orientations and sub-disciplinary coverage. By sampling abstracts from these outlets, the study aimed to construct a reliable benchmark of proficient English academic writing produced by expert scholars. In addition, selecting multiple journals rather than relying on a single source helps ensure disciplinary representativeness and reduces the influence of journal-specific stylistic conventions. These journals provide a balanced and comprehensive representation of contemporary research writing practices in linguistics, thereby strengthening the validity of the comparative analysis.

It should be noted that while both the masters’ theses and the doctoral dissertations were produced by Chinese EFL learners, the RAs were not necessarily authored by native English speakers. To ensure the reliability of the collected data, all selected RAs were written by scholars affiliated with prestigious universities in the United States or the United Kingdom. The native English authorship was further verified by checking the names of the authors and the contextual details in the acknowledgement section to ensure that they are from English-speaking countries. These authors are regarded as proficient English academic writers based on their demonstrated advanced proficiency in English, as evidenced by their publication records in those high-impact journals of linguistics.

CPhD abstracts are typically longer than CMA abstracts, which in turn are generally longer than RA abstracts. To ensure comparability across the three sub-corpora, we balanced the total word count by including fewer CPhD and CMA abstracts alongside a comparatively larger number of RA abstracts. The final corpus comprises 120 texts, totaling 35,002 words. The descriptive details of the corpus are presented in Table 1.

Table 1.

Descriptive Statistics for the Corpus.

	No. of texts	No. of words	Average no. of words	Word range
CMA	30	13,629	454	241–674
CPhD	20	11,200	560	381–882
RA	70	10,713	145	86–274

Data Collection

The data coding in this study is grounded in two theoretical frameworks. Specifically, the noun modifiers are classified according to the hypothesized developmental stages of complexity features proposed by Biber et al. (2011), a scheme adopted for the analysis of noun modifiers in our corpus. See Table 2.

Table 2.

Biber et al.’s (2011) Hypothesized Developmental Stages for Noun Modification.

Stage	Noun modifiers	Examples from our corpora
1	No modification	research
2	Attributive adjectives	a systematic way
3	*Participle premodifiers	a surprising prediction
	Finite relative clauses	strategies which these learners used
	Nouns as premodifiers	language production
	Possessive nouns as premodifiers	participant’s lexical density
	Of phrases (concrete/locative meanings)	urban center of the country
	Of phrases (abstract meanings)	instances of lexical ties
	Prepositional phrases other than of (concrete/locative meanings)	immigrants in the United States
4	Nonfinite relative clauses	the methods proposed
	Attributive adjectives, nouns as premodifiers	nonlinguistic symbol systems
	Prepositional phrases other than of (abstract meanings)	the possibility for extraction
5	Preposition + nonfinite complement clause	fluency of reading derived words
	*Preposition followed by wh- phrase	a theory of how extraction from this structure interacts with the theory of locality
	Complement clauses controlled by nouns	evidence that this prediction is false
	Appositive noun phrases	English as a Second Language (ESL)
	Multiple prepositional phrases as postmodifier	the crucial role of prominence in the structuring of discourse on each of the three levels

Not included in Biber et al.’s (2011) developmental stages for complexity features.

To measure the average length of nominal groups, the constituent elements within each nominal group are counted. A useful approach is illustrated by Halliday and Matthiessen’s (2014) description of nominal groups in terms of logical structure and experiential structure. The logical structure comprises a Head along with its pre- and/or post-modifiers, while the experiential structure consists of a Thing accompanied by one or more characterizing elements that precede it—Deictic, Post-Deictic, Numerative, Epithet₁, Epithet₂ and Classifier—and a Qualifier that follows it. The experiential structure of the nominal group those two splendid old electric trains with pantographs (Halliday & Matthiessen, 2014, p. 364) can be illustrated in Figure 1.

Figure 1.

Experiential structure of a nominal group.

The Head of a nominal group can be realized as a common noun, a proper noun, a pronoun or a nominalization. Since noun modification can occur at multiple levels within a nominal group, different levels are used for marking the position of head nouns in a nominal group. For example, aspects in the nominal group aspects of duration and intensity and formants of intervocalic stops in unstressed syllables of male announcers in VOA News in causal speech is a level 1 head noun, duration, intensity and formants are level 2 head nouns, and speech is a level 7 head noun. This study focuses exclusively on phrasal subordination. Therefore, for the three coordinated level 2 head nouns, we only coded the head noun carrying the largest number of modifying elements, or the first head noun when they are carrying the same number of elements. The number of head nouns at different levels reflects the degree of embedding within the nominal group. The relative proportions of head nouns at different levels indicate the structural complexity of the nominal groups in a text or corpus.

The corpus was annotated manually using UAM CorpusTool 3.3. Guided by the analytical frameworks, we designed the coding scheme for the types and levels of head nouns, and that for noun modifiers and their positions within the developmental stages. It is expected that academic writers with higher English proficiency would produce a greater proportion of head nouns at higher levels and fewer head nouns at lower levels compared to those with lower English proficiency. The data collected from our corpus are presented in Tables 3 to 5:

Table 3.

Head Nouns Retrieved from the Corpus.

	CMA	CPhD	RA	Total
Common noun	2,109	1,750	1,573	5,432
Nominalization	1,057	880	798	2,735
Pronoun	188	189	214	591
Proper noun	105	76	111	292
Total	3,459	2,895	2,696	9,050

Table 4.

Noun Modifiers Retrieved from the Corpus.

Stage	Noun modifiers	CMA	CPhD	RA	Total
2	Attributive adjectives	1,233	962	970	3,165
3	*Participle premodifiers	189	182	182	553
	Finite relative clauses	37	23	81	141
	Nouns as premodifiers	495	525	358	1,378
	Possessive nouns as premodifiers	67	112	26	205
	of phrases (concrete/locative meanings)	27	27	13	67
	of phrases (abstract meanings)	309	269	261	839
	Prepositional phrases other than of (concrete/locative meanings)	30	25	18	73
4	Nonfinite relative clauses	68	80	81	229
	adjectives + nouns as premodifiers	206	152	124	482
	Prepositional phrases other than of (abstract meanings)	154	131	141	426
5	Preposition + nonfinite complement clause	7	7	6	20
	*Preposition followed by wh- phrase	1	3	3	7
	Complement clauses controlled by nouns	7	14	19	40
	Appositive noun phrases	84	49	108	241
	Multiple prepositional phrases as postmodifier	175	146	154	475
Total		3,089	2,707	2,545	8,341

Table 5.

Distribution of Functional Elements in Nominal Groups.

	CMA	CPhD	RA	Total
Deictic	1,439	1,132	1,167	3,738
Post-Deictic	211	150	110	471
Numerative	283	170	155	608
Epithet₁	217	191	177	585
Epithet₂	2	0	1	3
Classifier	1,804	1,495	1,375	4,674
Qualifier	1,038	897	964	2,899
Total	4,994	4,035	3,949	12,978

The length of a nominal group is defined as the number of functional elements that specify its head noun. For instance, the nominal group shown in Figure 1 has a length of six, as it comprises one Deictic, one Numerative, two Epithets, one Classifier and one Qualifier. When calculating the average length of nominal groups in a text, only level-1 head nouns were included. Because pronouns and proper nouns are seldom modified, any modifiers attached to them were excluded from the data.

Data Analysis

Pearson’s Chi-squared test was conducted in SPSS 29.0 to examine the distribution of the data across the three groups of texts and to assess whether the observed differences were statistically significant. The formulas for Pearson’s Chi-squared test are as follows:

χ^{2} = \sum \frac{(O_{ij} - E_{ij})^{2}}{E_{ij}}

E_{ij} = \frac{row {total}_{i} \times column {total}_{j}}{n}

where $O_{ij}$ represents the observed frequency in each category, and $E_{ij}$ is the expected frequency for each category. The indices i and j correspond to the rows and columns respectively in the contingency table, while n is the total number of observations. A significant difference between the two variables is indicated if the $χ^{2}$ value exceeds the critical value or if p < .05.

However, despite the statistical significance (p < .05) of the Chi-squared test results, the association between the variables may not hold meaningful practical relevance. This highlights the importance of interpreting the results with an emphasis on effect size rather than relying solely on significance testing. The formula for effect size test is as follows:

{Cramer}^{'} s V = \sqrt{\frac{χ^{2}}{N \times (k - 1)}}

where N is the sample size, while k is the minimal dimension (row or column). Cramer’s V ranges between 0 and 1. Values closer to 1 indicate a stronger association, while values closer to 0 reflect a weaker association. For nonparametric analysis, we used Cohen’s w test.

Cohen' s w = \sqrt{\frac{χ^{2}}{N}}

According to Cohen (1988), a value below 0.1 indicates a negligible or very weak association between variables, which, even if statistically significant, generally holds little practical meaning. Values between 0.1 and 0.3 suggest a weak effect, implying a modest effect. A value ranging from 0.3 to 0.5 reflects a moderate effect, denoting a clear and substantive association that warrants attention in applied contexts. Finally, a value of 0.5 or higher represents a strong effect, signaling a pronounced relationship with meaningful practical or theoretical implications.

We also compared the distributions of the frequencies of head nouns across the three groups of texts at different levels. For this purpose, we first examined the regularity of the distributions of the data and then compared the distributional patterns across the three groups of texts by performing regression analyses based on the power-function model:

y = {ax}^{- b}

where parameters a and b jointly describe the distribution pattern. The coefficient a reflects the starting point of complexity and the magnitude of change. The exponent b determines the rate of decay in complexity as the level deepens. A larger b value signifies a more limited capacity to construct or sustain deeply nested structures, resulting in a short-tailed distribution. The two parameters together quantify the balance between efficiency and capacity in the syntactic strategies of different groups of academic writers.

Distribution of Head Nouns in Noun Phrases

This section analyzes the distribution of different types of head nouns and examines the complexity of noun phrases along two parameters: average length and structural complexity.

Distribution of Head Nouns

Of the four types of head nouns, common nouns account for more than half of the total frequency, while nominalizations represent approximately half the frequency of common nouns. To facilitate comparison, all raw frequencies were converted to normalized frequencies per 10,000 words. We then compared the normalized frequencies of each type of head nouns across the three sub-corpora. See Figure 2:

Figure 2.

Distribution of the four types of head nouns.

Figure 2 shows that the distributions of the four types of head nouns are generally significantly different in the three sub-corpora ( $χ^{2}$ = 23.347, df = 6, p < .05). However, the effect size is small (Cramer’s V = .039), indicating a very weak association among distributions of the four types of head nouns. Common nouns and nominalizations are the most prevalently used head nouns in all the three sub-corpora. However, they occur most frequently in the CPhD corpus and least frequently in the RA corpus. The two groups of data show a similar distribution across the three sub-corpora (χ² = .046, df = 2, p = .977 > .05). The Cramer’s V value of .002 further confirms this highly consistent distribution pattern. In contrast, pronouns and proper nouns tend to occur more in the RA corpus ( $χ^{2}$ = 3.494, df = 2, p = 0.174 > .05), but the effect size is relatively small (Cramer’s V = .063).

In English academic writing, pronouns are generally discouraged. This is evident from the fact that their frequency in spoken conversation is approximately seven times higher than in academic writing and there are also more pronouns in other non-academic written registers than in academic writing (Biber et al., 1999). Given this, one might expect that the RA writers would use fewer pronouns than the less experienced Chinese EFL students. However, as shown in Figure 2, the RA writers actually employed fewer nominalizations but more pronouns than the Chinese EFL students.

Personal pronouns, particularly the first-person pronouns I and we, are a core linguistic resource for writers to engage with readers, construct a stance, and claim authority (Hyland, 2005). The RA writers used pronouns as head nouns significantly more frequently than both the CMA and CPhD writers. This pronounced usage demonstrates their greater proficiency in crafting a distinct authorial voice to guide arguments, establish authority, and interact with readers. The fact that the CMA group used pronouns a little more frequently than the CPhD group may suggest that the CPhD writers pursued a higher degree of formality in academic writing or they might be more inclined to avoid pronouns that could be perceived as subjective. Coupled with the finding that the CPhD group used common nouns and nominalizations the most frequently, this pattern suggests a strategic adjustment phase during academic socialization. The CPhD writers appeared to be deliberately shifting from seemingly subjective pronouns toward more objective nominalized structures, aligning their writing with perceived norms of advanced academic writing.

An examination of pronoun types, however, reveals an interesting trend in RA academic writing. As presented in Table 6, first-person pronouns and other pronouns show distinct distribution patterns, and their normalized frequencies are compared in Figure 3.

Table 6.

Distribution of Pronouns in the Corpus.

	CMA	CPhD	RA	Total
First-person pronouns	5	23	75	103
Other pronouns	183	166	139	488
Total	188	189	214	591

Figure 3.

Distribution of person pronouns in the three sub-corpora.

Figure 3 shows a significant difference in the distribution of first-person pronouns across the three sub-corpora ( $χ^{2}$ = 78.287, df = 2, p < .001), and the effect size is moderate (Cramer’s V = .364). Specifically, the RA writers used significantly more first-person pronouns than the Chinese EFL students, and the CPhDs used relatively more first-person pronouns than the CMAs. Notably, the RA writers used other pronouns the least frequently. Although traditionally associated with informal English, the use of first-person pronouns in academic English has emerged as a recent, albeit complex, trend (Hyland & Jiang, 2017). The patterns observed in our data appear to align with this overall trend.

The frequent use of first-person pronouns (e.g., I argue, we propose) by the RA writers provides direct linguistic evidence of their efforts to construct an authorial identity, explicitly claim originality, and proactively guide readers through academic texts. The fact that the CPhD writers use first-person pronouns more frequently than the CMAs indicates that, during the process of academic socialization and with increasing experience in academic writing, the CPhDs begin to consciously increase their use of this important rhetorical resource, thereby aligning their practice more closely with that of the RA writers. This finding revised the earlier speculation that CPhDs inclined to avoid pronouns, revealing instead that they were strategically adjusting their usage patterns.

The distribution of other personal pronouns shows no significant difference across the three groups, with the RA group using them least frequently. This implies that the RA group’s preference is highly selective—only first-person pronouns are significantly increased, not all personal pronouns. This selective use underscores the precision of their rhetorical strategy. Their increased use of first-person pronouns thus serves specific functions, such as constructing a stance and engaging readers, rather than reflecting a general increase in pronoun use.

Nominal Group Complexity

The complexity of nominal groups can be reflected in their length and structure. Since the RA writers used fewer nominalizations and more personal pronouns, we could hypothesize that the RA writers would produce shorter and structurally less complex nominal groups.

We annotated 12,978 functional elements in nominal groups in the three sub-corpora, including 5,192 level 1 head nouns, finding that the average length of nominal groups is 2.53. See Table 7.

Table 7.

Average Length of Nominal Groups.

	CMA	CPhD	EL	Total
Level 1 common nouns	1,358	1,149	931	3,438
Level 1 nominalizations	740	551	463	1,754
Functional elements	4,994	4,035	3,949	12,978
Mean length	2.38	2.37	2.83	2.53

It can be seen from Table 7 that the mean length of nominal groups used by the RA writers is greater than those by the CMA and CPhD groups, although the Chinese EFL students use both common nouns and nominalizations more frequently than ELs. The Chi-squared test shows that the distribution of the frequencies of head nouns and functional elements in the CMA texts is not significantly different from that in the CPhD texts ( $χ^{2}$ = .005, df = 1, p = .941 > .05). The Cramer’s V value of .001 further confirms this highly consistent distribution pattern. However, the distribution in the RA texts is significantly different from that in both the CMA texts ( $χ^{2}$ = 18.403, df = 1, p < .001) and the CPhD texts ( $χ^{2}$ = 17.341, df = 1, p < .001), but the Cramer’s V is .038 and .040 respectively, indicating association strength is very weak. In addition, common nouns and nominalizations are similarly distributed in the three sub-corpora ( $χ^{2}$ = 3.709, df = 2, p = 0.157 > .05, Cramer’s V = .027). This indicates that nominalizations are not the reason for the longer noun phrases produced by the RA writers; rather, it is noun modification that determines the length of nominal groups.

The frequency of head nouns is extremely high in all the three sub-corpora, but it decreases with the increase of levels. We, therefore, compared the distributions of the frequencies of head nouns across the three groups of texts. Based on the data shown in Table 7, we expected that the distributions of the head nouns at different levels should abide by the power function.

Table 8 and Figure 4 illustrate the distribution patterns, and the results of the non-linear regression analysis are shown in Table 9.

Table 8.

Data Collected for the Levels of Head Nouns.

CMA			CPhD		RA
	Observed	Expected	Observed	Expected	Observed	Expected
1	2,326	4,542.59	1,895	4,270.34	1,623	2,968.68
2	799	456.76	723	374.06	682	426.26
3	242	119.16	212	90.02	261	136.97
4	64	45.93	49	32.77	91	61.21
5	17	21.92	13	14.96	26	32.77
6	6	11.98	3	7.89	10	19.67
7	2	7.19	0	4.59	3	12.77
8	1	4.62	0	2.87	0	8.79
9	1	3.13	0	1.90	0	6.32
10	1	2.20	0	1.31	0	4.71

Figure 4.

Fitting the level distributions of head nouns.

Table 9.

The Power Regression for the Levels of Head Nouns.

	a	b	Adjusted R²
CMA	4,542.585	3.314	0.908
CPhD	4,270.335	3.513	0.878
RA	2,968.678	2.800	0.889

The results indicate that all the three groups of data exhibit a strong power-law decay pattern (adjusted R² > .87), confirming that the distribution of head nouns at different levels follows a systematic quantitative pattern. However, key parameters revealed notable differences among the three groups. The highest value of parameter a in the CMA corpus indicates the highest starting point and a relatively rapid decay. The largest value of parameter b in the CPhD corpus corresponds to the most rapid decline of the curve with the increase of the levels. Both indicate the higher frequency of level 1 head nouns. The lowest a and b values in the RA corpus indicate that the frequency of level 1 head nouns is the smallest and the frequency declines the slowest with the increase of the levels, and that the RA writers maintain the ability to produce relatively higher number of complex structures at higher levels (e.g., Levels 3–6).

The CMA and CPhD groups used more level 1 head nouns than the RA group, but the s used fewer level 2 head nouns than the other two groups of writers. Head nouns at levels 3 to 6 are all used the most frequently by RA writers. This distribution trend indicates that RA writers produced more complex nominal groups than Chinese EFL students and that the CPhDs produced more complex nominal groups than the CMAs. The distributions of the data at the first four levels are significantly different in the three sub-corpora ( $χ^{2}$ = 53.077, df = 6, $p$ < .001), and the Cramer’s V is .077, indicating weak correlation strength. This agrees with our hypothesis that proficient ENS academic writers would write longer and more complex nominal groups in their academic writing than Chinese EFL academic writers. This is also consistent with Berlage (2014) that the length and complexity of nominal groups are positively correlated.

It can be concluded that although the RA writers used a smaller number of nominalizations and a larger number of first-person pronouns in their academic writing, their nominal groups are more complex in terms of length and structure than those produced by the CMA and CPhD groups. These differences reflect a divergence in syntactic strategies. RA writers are more adept at precisely constructing deeply embedded, complex nominal groups when rhetorically necessary to achieve high information density and logical integration, whereas Chinese EFL writers, particularly the CPhDs, may adopt a more conservative strategy characterized by steeper decay and a shorter tail, possibly to manage the structural risk. This finding offers quantitative evidence from the perspective of syntactic distribution for understanding the development of academic language proficiency and the expert balance between efficiency and capacity in academic writing.

When combined with the previous finding that the RA writers used first-person pronouns with the highest frequency, it can be argued that a hallmark of advanced academic writing lies in the ability to skillfully and effectively switch between and balance highly integrated nominal structures and clearly engaging first-person strategies, depending on specific rhetorical purposes (e.g., compressing information vs. constructing stance). This precise orchestration of syntactic-rhetorical resources may be a key factor distinguishing the RA writers from the Chinese EFL learners.

Distribution of Noun Modifiers

In this section, we delved deeper into understanding the complexity of the nominal groups used by the three groups of writers. We examined the distribution of different types of noun modifiers and explored which types contributed to the complexity of the nominal groups used by the RA writers.

Overall Distribution of Noun Modifiers

As shown in Figure 5, which presents the normalized frequencies of the seven types of noun modifiers, attributive adjectives are the most common across all the three sub-corpora, followed by prepositional phrases. Nouns acting as premodifiers rank the third, whereas the remaining four types of modifiers collectively account for only a small proportion.

Figure 5.

Distribution of the seven types of noun modifiers.

The analysis of noun modifiers reveals distinct patterns across groups. The distributions of attributive adjectives in the three sub-corpora are not significantly different ( $χ^{2}$ = 1.409, df = 2, p = .494 > .05). The corresponding effect size, Cohen’s w, is .021, suggesting that any differences are negligible. Clausal modifiers, however, are significantly differently distributed in the three sub-corpora ( $χ^{2}$ = 33.586, df = 2, p = .000 < .001), with moderate effect size (Cohen’s w = 0.312). This suggests that clausal modifiers are used the most frequently by the RA writers and the least frequently by the CMAs. Prepositional phrases are used with similar frequencies in the three sub-corpora ( $χ^{2}$ = 1.547, df = 2, p = 0.461 > .05, Cohen’s w = .031). Both noun and possessive noun premodifiers are used the most frequently in the CPhD sub-corpus and the least frequently in the RA sub-corpus ( $χ^{2}$ = 25.988, df = 2, p = .000 < .001; $χ^{2}$ = 52.035, df = 2, p = .000 < .001), and the association strength is modest (Cohen’s w = .149) and strong (Cohen’s w = 0.548) respectively. Appositive noun phrases are used the most frequently by the RA writers and the least frequently by the CPhDs ( $χ^{2}$ = 24.609, df = 2, p = .000 < .001) with moderate effect size (Cohen’s w = .345).

The distributions of adjective plus noun premodifiers are not significantly different in the three sub-corpora ( $χ^{2}$ = 4.591, df = 2, p = 0101 > .001). However, the effect size reached the threshold for a modest effect (Cohen’s w = .107). This suggests that there may be a weak tendency toward distributional unevenness, though it has not attained statistical significance. The examination of different types of noun modifiers further reveals fundamental differences between the RA writers and the Chinese EFL writers in their repertoire of syntactic resources and their strategic choices in deploying them.

The noun modifiers investigated in this study can be categorized as premodifiers (e.g., attributive adjectives, nouns, possessive nouns and adjectives plus nouns) and postmodifiers (e.g., clauses, prepositional phrases, appositive noun phrases). In all the three sub-corpora, premodifiers collectively occur more frequently than postmodifiers. See Figure 6.

Figure 6.

Distribution of premodifiers and postmodifiers.

The Statistical analysis reveals a distinct pattern. The frequency of premodifiers is the lowest in the RA sub-corpus ( $χ^{2}$ = 16.998, df = 2, p < .001, Cohen’s w = .006). This indicates that the observed distributional differences may not hold substantial practical importance although the statistical test is significant. However, the frequency of postmodifiers is the highest in the RA sub-corpus ( $χ^{2}$ = 11.86, df = 2, p = .003 < .05), but the correlation strength is also very weak (Cohen’s w = .079). In contrast, the two Chinese EFL sub-corpora show no significant difference in the distribution of these two modifier types ( $χ^{2}$ = .142, df = 2, p = 0.706 > .05, Cramer’s V = .006). This distribution pattern indicates a clear preference among the RA writers for postmodifiers in constructing nominal groups. Postmodifiers, particularly relative clauses and prepositional phrases, are the primary means of creating syntactic nesting and hierarchical organization. This directly explains why the RA writers are able to construct nominal groups with deeper embedding. Conversely, while stacking premodifiers can increase informational density, it does not readily facilitate the formation of deeply embedded syntactic structures.

Distribution of Noun Modifiers at Different Developmental Stages

Given that experienced academic writers typically employ more advanced modifiers, it is reasonable to hypothesize that higher-stage noun modifiers would occur the most frequently in the RA sub-corpus, followed by the CPhD sub-corpus, and then the CMA sub-corpus. Figure 7 presents the distributions of these normalized frequencies across stages for comparison with this hypothesis.

Figure 7.

Different stage noun modifiers in the corpora.

The distributions of the normalized frequencies of noun modifiers across the four developmental stages are significantly different in the three sub-corpora ( $χ^{2}$ = 34.698, df = 6, p = < .001), but the correlation strength is weak (Cramer’s V = .05). As expected, the normalized frequencies in the CMA and the RA sub-corpora decreased with the increase of the stages. However, in the CPhD corpus, Stage 3 modifiers constituted a larger proportion than those at Stage 2. Further analysis by stage revealed distinct patterns. While Stage 2 noun modifiers showed no significant difference across the three sub-corpora ( $χ^{2}$ = 1.586, df = 2, p = 0.452 > .05, Cohen’s w = .024), Stage 3 noun modifiers dominate in the CPhD sub-corpus ( $χ^{2}$ = 22.919, df = 2, p = .000 < .01) with weak association strength (Cohen’s w = .091) and they are evenly distributed in the CMA and RA sub-corpora. Stage 4 distributions were similar across groups ( $χ^{2}$ = 0.352, df = 2, p = 0.850 > .05, Cohen’s w = .018), and Stage 5 modifiers were most frequent in the RA sub-corpus ( $χ^{2}$ = 15.375, df = 2, p < .001) with modest effect size (Cohen’s w = .152).

In line with our hypothesis, higher stage modifiers are the most prevalent in the RA sub-corpus, whereas lower stage modifiers tend to occur more in the CPhD sub-corpus. This indicates that the RA writers possessed a richer and more accessible repertoire of complex structures, enabling them to deploy higher stage modifiers as needed to achieve precise logical expression and higher information integration.

Discussion

The findings of this study present a nuanced picture of academic writing proficiency. On the one hand, the RA writers used significantly fewer nominalizations and more first-person pronouns than Chinese EFL students. This pattern, rather than indicating informality, reflects the RA writers’ strategic use of conversational resources to engage readers and construct an authoritative stance. On the other hand, consistent with Biber et al.’s (2011) hypothesized developmental model, the RA writers employed a greater proportion of higher-stage noun modifiers, demonstrating their command of complex syntactic structures. This syntactic sophistication stems primarily from their predominant use of postmodifiers, which enables deeper embedding within nominal groups. In contrast, no significant difference in nominal group complexity was found between the CPhD and CMA groups, suggesting comparable proficiency levels in this domain. This section discussed these key findings.

Across all the three sub-corpora, attributive adjectives, prepositional phrases and nouns as premodifiers are the three most common types of noun modifiers, collectively accounting for over 80% of all noun modifiers. This finding is in agreement with the findings of previous studies (e.g., Ansarifar et al., 2018; Hu & He, 2023; Parkinson & Musgrave, 2014). The overall frequency of prepositional phrases is similar across groups. Within this category, of-phrases predominate, constituting over 45% of all prepositional phrases in each sub-corpus. Biber et al. (1999) attributed the prevalence of of-phrases to the wide functional range of. Semantically, most of-phrases and other simple prepositional phrases carry abstract meanings rather than concrete/locative meanings (over 90% and over 80%, respectively). This proportion exceeds the 60% reported by Biber and Gray (2010) for contemporary academic English, underscoring the strongly abstract nature of noun modification in our corpus.

The greater use of postmodifiers by the RA writers facilitates the production of longer and structurally more complex nominal groups. This structural potential arises because, unlike premodifiers which are exclusively phrasal or lexical in the Hallidayan sense (Halliday, 1961), postmodifiers in English can be either phrasal (e.g., prepositional phrases) or clausal (Biber et al., 2009). This distinction is illustrated in the following example quoted from Biber and Gray (2010, p. 9):

(5) a. the participant perspective →

the perspective that considers the participant’s point of view

b. a systems, theoretical orientation →

an orientation which is theoretical and which focuses on the analysis of systems

The data collected in this study indicate a clear preference for phrasal modifiers over clausal modifiers in English academic writing (see Table 10). To facilitate comparison, we converted the total nominalized frequency of clausal modifiers (356) to that of phrasal modifiers (6,703). See Figure 8.

Table 10.

Phrasal and Clausal Modifiers in the Three Sub-Corpora.

	CMA	CPhD	RA	Total
Phrasal modifiers	2,977	2,590	2,364	7,931
Clausal modifiers	112	117	181	410
Total	3,089	2,707	2,545	8,341

Figure 8.

Comparing the distributions of phrasal and clausal postmodifiers (equal totality).

Phrasal modifiers are similarly distributed across the three sub-corpora ( $χ^{2}$ = 4.263, df = 2, p = 0.119 > .05, Cohen’s w = .025), whereas clausal modifiers dominate in the RA sub-corpus ( $χ^{2}$ = 646.211, df = 2, p < .001) with moderate effect (Cohen’s w = 0.310). Furthermore, the CPhDs used relatively more clausal modifiers than the CMAs ( $χ^{2}$ = 16.315, df = 1, p = .000 < .001) but the association strength is weak (Cramer’s V = .045). This pattern indicates that the RA writers possessed advanced academic syntactic competence, enabling them to achieve high information integration and precise logical argumentation. Concurrently, the greater use of clausal structures by the CPhDs suggests that with the advance of their academic proficiency, the CPhDs significantly increased their use of clausal structures, gradually approaching the level observed in the RAs.

The fundamental divergence in nominal group complexity between the RA writers and the Chinese EFL writers lies not in phrasal modifiers, but specifically in the use of clausal modifiers. Proficiency in employing clausal modifiers, particularly relative clauses, is a key syntactic hallmark of advanced academic writing. This proficiency accounts for the RA writers’ ability to construct more deeply embedded nominal groups and their overall preference for postmodification. The comparative analysis of phrasal versus clausal postmodifiers ultimately pinpoints the linguistic essence underlying the differences in syntactic strategies between the RA writers and the Chinese EFL learners.

A key finding of this study is that the RA writers used first-person pronouns more frequently than Chinese EFL students, with the CPhDs also using them more than the CMAs. This pattern diverges from the traditional emphasis in Chinese EFL pedagogy, where students are often taught to avoid first-person pronouns to uphold the objectivity characteristic of academic language (Hyland, 2006). In light of this teaching norm, the lower frequency among Chinese EFL writers is understandable. Conversely, the prevalent use of first-person pronouns by the RA writers in our corpus suggests that strategic self-mention may be interpreted not as a breach of objectivity, but as a legitimate and effective rhetorical resource in academic writing.

Academic writing is fundamentally an interactive process between writers and readers, in which self-mention serves as a key linguistic resource for constructing authorial stance (Hyland, 2005). Specifically, first-person pronouns are a crucial means of building authorial identity and claiming originality (Hyland, 2001, 2008). Research demonstrates that writers strategically employ such devices to convey evaluative attitudes (Wang & Hu, 2023; Xie, 2020; Xie et al., 2024). This strategic use exhibits disciplinary variation: it is more frequent in soft sciences to express viewpoints, whereas in hard sciences it primarily describes methods and procedures (Hyland, 2004). Furthermore, cross-linguistic and comparative studies indicate that self-mention is more prevalent in English than in Italian academic texts (Molino, 2010), and that expert international writers use it more frequently than novice writers to construct a persuasive authorial identity (Xie et al., 2024).

Previous research suggests that many written registers in English have become less formal (Adel, 2008; Foster, 2005; Leedham, 2015), including academic texts (Alipour & Nooreddinmoosa, 2018; Hyland & Jiang, 2017). For example, Hyland and Jiang (2017) found an increase in informal features in English academic writing, while Alipour and Nooreddinmoosa (2018) found that informal features are more frequent in academic texts written by proficient English writers than those written by Iranian EFL writers. This aligns with our finding that the RA writers used more first-person pronouns and fewer nominalizations than the Chinese EFL writers at the graduate and doctoral levels.

The claim that English academic writing is becoming more informal, however, has been challenged in recent years. Hyland and Jiang (2017) presented a more nuanced view, noting significant disciplinary variation—for instance, a marked increase in first-person pronouns in applied linguistics. Moreover, while first-person pronouns are common in spoken English, Biber and Gray (2010) argued that the historical trajectory of English academic writing is defined not by becoming more speech-like, but by its increasing reliance on complex noun phrases, a feature that fundamentally distinguishes it from conversation. They posited that this structural complexity is a more pervasive register-defining phenomenon than the use of informal features in English academic writing.

The findings of the present study complicate this narrative. We observed that the trend toward informal features (e.g., first-person pronouns) coincided with particular patterns in the use of complex nominal groups. Specifically, despite their higher proficiency, the RA writers produced less technical or less heavily nominalized constructions than Chinese EFL students. This suggests that advanced academic writing may involve a strategic integration of interpersonal, stance-oriented features like self-mention with a particular type of syntactic complexity, rather than a straightforward choice between informality and nominal group complexity.

We are not suggesting that academic writing is converging with spoken English. The pervasive use of complex nominal groups remains a key feature that distinguishes academic writing from conversation. However, our analysis reveals a more nuanced relationship between proficiency and formal complexity: within academic writing, the RA writers tended to use less compressed and therefore less technical nominal groups than Chinese EFL writers. Thus, whereas advanced writing proficiency among the Chinese EFL learners is often associated with increased technicality, the RA writers reveal a strategic divergence: high proficiency can also involve the deliberate choice to reduce formal complexity.

Conclusion

This study investigated the characteristics of nominal groups in English academic writing to explore their relationship with academic writing proficiency, based on a self-built corpus of three groups of linguistics texts. The findings indicate that the RA writers produced longer nominal groups than the Chinese EFL students. This supports our hypothesis that the average length of nominal groups in English academic writing is positively correlated with the writers’ academic writing proficiency. This increased length is primarily due to their more frequent use of clausal postmodifiers rather than nominalizations. While phrasal modifiers which contribute more to lexical density are fundamental to academic writing and their distribution aligns with that of nominalizations, the RA writers’ predominant use of clausal modifiers does not indicate lower proficiency. Instead, along with their higher frequency of first-person pronouns and lower frequency of nominalizations, it suggests a strategic stylistic choice that may reflect a broader trend toward a less formal style in expert academic writing.

This study yields important pedagogical implications for teaching English academic writing to Chinese EFL university students. First, instruction should aim to demystify certain traditional prescriptions, such as the prohibition of first-person pronouns and the consequent overreliance on passive voice. Encouraging appropriate authorial involvement could be presented as a contemporary dimension of effective academic writing. Accordingly, teaching should balance the need for objectivity with training in strategic self-mention, thereby enhancing the expressive power and interpersonal engagement of academic writing.

Second, students should be guided to use modifiers more strategically. Specifically, they are recommended to reduce an over-reliance on certain premodifiers (e.g., nouns, possessive nouns, and adjective-noun combinations) and to develop greater competence with postmodifiers, particularly finite and non-finite relative clauses. This shift would help foster the production of more structurally complex and linguistically precise nominal groups in academic writing, better equipping students to meet the demands of academic expression.

This study has several limitations. First, the corpus is confined to linguistics texts, which constrains the generalizability of the findings to other academic disciplines. Second, our criterion for identifying expert writers—affiliation with prestigious Anglophone universities—conflates language proficiency with institutional, editorial, and disciplinary influences, making it difficult to attribute differences solely to language proficiency. Third, the corpus size is relatively small, and the deliberate imbalance in the number of texts across sub-corpora (a design choice to balance total word count) may have potentially impacted the representativeness and statistical power of some comparisons.

Footnotes

ORCID iD

Qingshun He

Ethical Considerations

This article does not contain any studies with human or animal participants.

Funding

The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work was supported by the Major Program of National Fund of Philosophy and Social Science of China [24&ZD250]

Declaration of Conflicting Interests

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

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

References

Ädel

(2008). Metadiscourse across three varieties of English: American, British and advanced learner English. In Connor

Nagelhout

Rozycki

W. V.

(eds.), Contrastive rhetoric: Reaching to intercultural rhetoric (pp. 45–62). Benjamins. https://doi.org/10.1075/pbns.169.06ade

Alipour

Nooreddinmoosa

(2018). Informality in applied linguistics research articles: Comparing native and non-native writings. Eurasian Journal of Applied Linguistics, 4(2), 349–373. http://dx.doi.org/10.32601/ejal.464196

Ansarifar

Shahriari

Pishghadam

(2018). Phrasal complexity in academic writing: A comparison of abstracts written by graduate students and expert writers in applied linguistics. Journal of English for Academic Purposes, 31, 58–71. https://doi.org/10.1016/j.jeap.2017.12.008

Banks

(2005). On the historical origins of nominalized process in scientific texts. English for Specific Purposes, 24(3), 347–357. https://doi.org/10.1016/j.esp.2004.08.002

Banks

(2008). The development of scientific writing: Linguistic features and historical context. Equinox.

Bardovi-Harlig

(1992). A second look at T-unit analysis: Reconsidering the sentence. TESOL Quarterly, 26(2), 390–395. https://doi.org/10.2307/3587016

Berlage

(2014). Noun phrase complexity in English. Cambridge University Press.

Biber

(2006). University language: A corpus-based study of spoken and written registers. John Benjamins.

Biber

Gray

(2010). Challenging stereotypes about academic writing: Complexity, elaboration, explicitness. Journal of English for Academic Purposes, 9(1), 2–20. https://doi.org/10.1016/j.jeap.2010.01.001

10.

Biber

Gray

(2016). Grammatical complexity in academic writing: Linguistic change in writing. Cambridge University Press.

11.

Biber

Gray

Poonpon

(2011). Should we use characteristics of conversation to measure grammatical complexity in L2 writing development? TESOL Quarterly, 45(1), 5–35. https://doi.org/10.5054/tq.2011.244483

12.

Biber

Grieve

Iberri-shea

(2009). Noun phrase modification. In Rohdenburg

Schlüter

(eds.), One language, two grammars? Differences between British and American English (pp. 182–193). Cambridge University Press.

13.

Biber

Johansson

Leech

Finegan

(1999). Longman Grammar of spoken and written English. Pearson Education.

14.

Biber

Reppen

Staples

Egbert

(2020). Exploring the longitudinal development of grammatical complexity in the disciplinary writing of L2-English university students. International Journal of Learner Corpus Research, 6(1), 38–71. https://doi.org/10.1075/ijlcr.18007.bib

15.

Brown

Yule

(1983). Discourse analysis. Cambridge University Press.

16.

Chafe

(1982). Integration and involvement in speaking, writing, and oral literature. In Tannen

(ed.), Spoken and written language: Exploring orality and literacy (pp. 35–54). Ablex.

17.

Charles

(2003). A corpus-based study of the use of nouns to construct stance in theses from two contrasting disciplines. Journal of English for Academic Purposes, 2(4), 313–326. https://doi.org/10.1016/S1475-1585(03)00048-1

18.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Academic Press.

19.

de Haan

(1989). Postmodifying clauses in the English noun phrase: A corpus-based study. Rodopi.

20.

Durrant

Brenchley

(2023). Development of noun phrase complexity across genres in children’s writing. Applied Linguistics, 44(2), 239–264. https://doi.org/10.1093/applin/amac032

21.

Fang

Gresser

Cao

Zheng

(2021). Nominal complexities in school children’s informational writing. Journal of English for Academic Purposes, 50, 100958. https://doi.org/10.1016/j.jeap.2021.100958

22.

Fang

Schleppergrell

Cox

(2006). Understanding the language demands of schooling: Nouns in academic registers. Journal of Literacy Research, 38(3), 247–273. https://doi.org/10.1207/s15548430jlr3803_1

23.

Foster

(2005). Effective writing skills for public relations (3rd edi.). Kogan Page.

24.

Freddi

(2005). Arguing linguistics: Corpus investigation of one functional variety of academic discourse. Journal of English for Academic Purposes, 4(1), 5–26. https://doi.org/10.1016/j.jeap.2003.09.002

25.

Galve

I. G.

(1998). The textual interplay of grammatical metaphor on the nominalizations occurring in written medical English. Journal of Pragmatics, 30(3), 363–385. https://doi.org/10.1016/S0378-2166(98)00002-2

26.

Gao

(2023). A corpus-based study of dependency distance differences in English academic writing. SAGE Open, 13(3), 1–12. https://doi.org/10.1177/21582440231198408

27.

Granger

(1997). On identifying the syntactic and discourse features of participle clauses in academic English: Native and non-native writers compared. In Aarts

de Mönnink

Wekker

(eds.), Studies in English language and teaching (pp. 185–198). Rodopi. https://doi.org/10.1163/9789004653351_014

28.

Gray

(2013). More than discipline: Uncovering multi-dimensional patterns of variation in academic research articles. Corpora, 8(2), 153–181. https://doi.org/10.3366/cor.2013.0039

29.

Groom

(2005). Pattern and meaning across genres and disciplines: An exploratory study. Journal of English for Academic Purposes, 4(3), 257–277. https://doi.org/10.1016/j.jeap.2005.03.002

30.

Halliday

M. A. K.

(1961). Categories of the theory of grammar. Word, 17(2), 241–292. https://doi.org/10.1080/00437956.1961.11659756

31.

Halliday

M. A. K.

(1985). Spoken and written language. Deakin University Press.

32.

Halliday

M. A. K.

(1993). The construction of knowledge and value in the grammar of scientific discourse: Charles Darwin’sThe Origin of Species. In Halliday

M. A. K.

Martin

J. R.

(Eds.), Writing science: Literacy and discursive power (pp. 86–105). The Falmer Press.

33.

Halliday

M. A. K.

(1998). Things and relations: Regrammaticising experience as technical knowledge. In Martin

J. R.

Veel

(Eds.), Reading science: Critical and functional perspectives on discourses of science (pp. 185–235). Routledge.

34.

Halliday

M. A. K.

(2004). Language and knowledge: The unpacking of text. In Webster

(Ed.), Collected works of M. A. K. Halliday, vol. 5: The language of science (pp. 24–48). Continuum.

35.

Halliday

M. A. K.

Matthiessen

C. M. I. M.

(1999). Construing experience through meaning: A language-based approach to cognition. Continuum.

36.

Halliday

M. A. K.

Matthiessen

C. M. I. M.

(2014). An Introduction to functional grammar (4th ed.). Routledge.

37.

(2019). Types of transfer in ideational metaphor: A corpus-based study. Studia Neophilologica, 91(3), 249–272. https://doi.org/10.1080/00393274.2019.1619093

38.

Guo

(2021). A corpus-based study of adjectivizations in English. Lingua, 260, 103100.

39.

Yang

(2018). A corpus-based study of the correlation between text technicality and ideational metaphor in English. Lingua, 203, 51–65. https://doi.org/10.1016/j.lingua.2017.10.005

40.

Zhang

(2024). A corpus-based study of live grammatical metaphor in English academic writing. Studia Neophilologica, online, 1–20. https://doi.org/10.1080/00393274.2024.2373728

41.

Hewings

(2001). Academic writing in context: Implications and applications. The University of Birmingham Press.

42.

(2023). A corpus-based study of the distributions of adnominals across registers and disciplines. Journal of Quantitative Linguistics, 30(2), 183–203. https://doi.org/10.1080/09296174.2023.2209487

43.

Hughes

(1996). English in speech and writing: Investigating language and literature. Routledge.

44.

Hyland

(2000). Disciplinary discourses: Social interactions in academic writing. Longman.

45.

Hyland

(2001). Humble servants of the discipline? Self-mention in research articles. English for Specific Purposes, 20(3), 207–226. https://doi.org/10.1016/S0889-4906(00)00012-0

46.

Hyland

(2004). Disciplinary interactions: Metadiscourse in L2 postgraduate writing. Journal of Second Language Writing, 13(2), 133–151. https://doi.org/10.1016/j.jslw.2004.02.001

47.

Hyland

(2005). Stance and engagement: A model of interaction in academic discourse. Discourse Studies, 7(2), 173–192. https://doi.org/10.1177/1461445605050365

48.

Hyland

(2006). English for academic purpose: An advanced resource book. Routledge.

49.

Hyland

(2008). Persuasion, interaction and the construction of knowledge: Representing self and others in research writing. International Journal of English Studies, 8(2), 1–23. http://revistas.um.es/ijes/article/view/49151

50.

Hyland

Jiang

F. K.

(2017). Is academic writing becoming more informal? English for Specific Purposes, 45, 40–51. https://doi.org/10.1016/j.esp.2016.09.001

51.

Jalilifar

Alipour

Parsa

(2014). Comparative study of nominalization in applied linguistics and biology books. Journal of Research in Applied Linguistics, 5, 24–43.

52.

Jalilifar

Saleh

Don

(2017). Exploring nominalization in the introduction and method sections of applied linguistics research articles: A qualitative approach. Romanian Journal of English Studies, 14, 64–80. https://doi.org/10.1515/rjes-2017-0009

53.

Lan

Sun

(2019). A corpus-based investigation of noun phrase complexity in the L2 writings of a first-year composition course. Journal of English for Academic Purposes, 38, 14–24. https://doi.org/10.1016/j.jeap.2018.12.001

54.

Larsson

Sardinha

Gray

Biber

(2023). Exploring early L2 writing development through the lens of grammatical complexity. Applied Corpus Linguistics, 3(3), 100077. https://doi.org/10.1016/j.acorp.2023.100077

55.

Leedham

(2015). Chinese students’ Writing in English: Implications from a corpus-driven study. Routledge.

56.

Lei

Wen

Yang

(2023). A large-scale longitudinal study of syntactic complexity development in EFL writing: A mixed-effects model approach. Journal of Second Language Writing, 59, 100962. https://doi.org/10.1016/j.jslw.2022.100962

57.

Lorenzo

Rodríguez

(2014). Onset and expansion of L2 cognitive academic language proficiency in bilingual settings: CALP in CLIL. System, 47, 64–72. https://doi.org/10.1016/j.system.2014.09.016

58.

(2015). Syntactic complexity in college-level English writing: Differences among writers with diverse L1 backgrounds. Journal of Second Language Writing, 29, 16–27. https://doi.org/10.1016/j.jslw.2015.06.003

59.

Martínez

I. A.

(2005). Native and non-native writers’ use of first-person pronouns in the different sections of biology research articles in English. Journal of Second Language Writing, 14(3), 174–190. https://doi.org/10.1016/j.jslw.2005.06.001

60.

Molino

(2010). Personal and impersonal authorial references: A contrastive study of English and Italian linguistics research articles. Journal of English for Academic Purposes, 9(2), 86–101. https://doi.org/10.1016/j.jeap.2010.02.007

61.

Norris

J. M.

Ortega

(2009). Towards an organic approach to investigating CAF in instructed SLA: the case of complexity. Applied Linguistics, 30(4), 555–578. https://doi.org/10.1093/applin/amp044

62.

Ortega

(2003). Syntactic complexity measures and their relationship to L2 proficiency: A research synthesis of college level L2 writing. Applied Linguistics, 24(4), 492–518. https://doi.org/10.1093/applin/24.4.492

63.

Ouyang

Jiang

Liu

(2022). Dependency distance measures in assessing L2 writing proficiency. Assessing Writing, 51, 100603. https://doi.org/10.1016/j.asw.2021.100603

64.

Parkinson

Musgrave

(2014). Development of noun phrase complexity in the writing of English for academic purposes students. Journal of English for Academic Purposes, 14, 48–59. https://doi.org/10.1016/j.jeap.2013.12.001

65.

Rimmer

(2006). Measuring grammatical complexity: The Gordian knot. Language Testing, 23(4), 497–519. https://doi.org/10.1191/0265532206lt339oa

66.

Shao

Zhang

Zhong

(2022). Phrasal complexity in English argumentative writing: Variations across Chinese STEM versus English majors’ production and EFL textbook essays. Journal of English for Academic Purposes, 55, 101070. https://doi.org/10.1016/j.jeap.2021.101070

67.

Swales

J. M.

Feak

C. B.

(2009). Abstracts and the writing of abstracts. The University of Michigan Press.

68.

Tao

(2025). Longitudinal development of L2 learners’ linguistic complexity in reading-to-write argumentative tasks. Journal of English for Academic Purposes, 75, 101502. https://doi.org/10.1016/j.jeap.2025.101502

69.

Wang

(2023). Changes in the ways authors refer to themselves: A diachronic study of self-mention in English research articles. Humanities & Social Sciences Communications, 10, 646. https://doi.org/10.1057/s41599-023-02099-7

70.

Mauranen

Lei

(2020). Syntactic complexity in English as a lingua franca academic writing. Journal of English for Academic Purposes, 43, 100798. https://doi.org/10.1016/j.jeap.2019.100798

71.

Xie

(2020). A review of research on authorial evaluation in English academic writing: A methodological perspective. Journal of English for Academic Purposes, 47, 100892. https://doi.org/10.1016/j.jeap.2020.100892

72.

Xie

Bui

(2024). A diachronic study of authorial stance in the discussion of Chinese MA theses and published research articles. Journal of English for Academic Purposes, 67, 101320. https://doi.org/10.1016/j.jeap.2023.101320