A contrastive study of grammatical metaphors in abstracts of Chinese MA theses and expert academic writing

4.1 Research questions

The present research aims to conduct a corpus-based study of ideational GM use in abstracts of MA theses and expert academic writing which consists of quantitative and qualitative analyses. The research questions of this study are presented as follows: (1) What are the frequency and usage characteristics of ideational GMs in the abstracts of MA theses and expert RAs? (2) What are the similarities and differences between abstracts of MA theses and expert RAs in ideational GM use?

4.2 Data collection

The present research selected from the library database 30 MA theses abstracts written by English majors in a top Chinese university from 2017 to 2018, and 30 abstracts of the research articles published from 2019 to 2020 in the journal of Applied Linguistics.

4.3 Instruments

The research instruments consist mainly of AntConc and SPSS. AntConc is a multiplatform instrument for conducting corpus research and the use of it helps to process data accurately, thus improving the scientificity and reliability of this study. By referring to the previous studies on formulaic language (Li and Jiao 2013; Zhou and Liu 2017), this study used AntConc to research some typical formulaic language to help make annotations. In the process of retrieval, the changes of morphemes were considered and the wildcard was used. Data analyses were conducted with SPSS 20.0 such as the paired samples t-test, the independent t-test, and the Pearson correlation test.

4.4 Procedure and data analysis

The study is a combination of quantitative and qualitative analysis based on the self-built corpus. The procedure is shown as follows.

5.1 Data analysis of ideational GMs in the abstracts

In this part, GM frequencies in MA theses and expert RAs, distribution of 13 types of GM, as well as the correlations between different GM types, are presented by charts and graphs through the use of SPSS.

5.1.1 GM frequencies in MA theses and expert RAs

Table 2 shows that the frequency of GMs in expert RAs is higher than that of MA theses, which means that experts use various GMs more frequently than student researchers.

Table 2:

The overall GM frequency.

	Total numbers of GMs	Total numbers of clauses	Frequencies of GMs
Expert RAs	1,557	447	3.48
MA theses	2,290	802	2.86

A paired samples t-test was conducted. Results suggest that the two groups (expert RAs vs. MA theses) differ significantly in terms of the overall GM use (r = 0.998, t (12) = 2.39, p = 0.03). Expert RAs (M = 0.27, SD = 0.46) have a higher mean GM frequency than MA theses (M = 0.22, SD = 0.39).

To further analyze the usage of every type of GM, the independent t-test was used to examine whether there are differences between expert RAs and MA theses in the usage of every single GM type and the result is shown in Table 3. To represent the characteristics of distribution visually, the comparative usage conditions of every type of GMs are shown in Figure 1. Paper type1 represents the expert group and Paper type2 stands for the student group.

Table 3:

Descriptive statistics and independent t-test results.

Variable: GM types	Paper types	M ± SD	t (df)	P (2-tailed)	CI (d)
T1	1	0.22 ± 0.37	1.62 (58)	0.11	−0.03, 0.25
	2	0.11 ± 0.09
T2	1	1.88 ± 3.84	1.31 (58)	0.20	−0.49, 2.32
	2	0.96 ± 0.25
T3	1	0.05 ± 0.07	0.25 (58)	0.80	−0.03, 0.04
	2	0.05 ± 0.06
T4	1	0.00 ± 0.00	−2.10 (58)	0.04	−0.01, 0.00
	2	0.00 ± 0.01
T5	1	0.35 ± 0.24	2.48 (58)	0.02	0.02,0.21
	2	0.23 ± 0.10
T6	1	0.20 ± 0.18	1.96 (58)	0.05	0.00, 0.15
	2	0.12 ± 0.10
T7	1	0.00 ± 0.00	−2.82 (58)	0.01	−0.02, 0.00
	2	0.01 ± 0.02
T8	1	0.04 ± 0.06	−0.75 (58)	0.46	−0.03, 0.02
	2	0.04 ± 0.04
T9	1	0.06 ± 0.06	2.5 (58)	0.02	0.01, 0.06
	2	0.03 ± 0.04
T10	1	0.07 ± 0.10	−0.36 (58)	0.72	−0.05, 0.04
	2	0.08 ± 0.06
T11	1	0.02 ± 0.04	1.66 (58)	0.10	0.00, 0.03
	2	0.01 ± 0.02
T12	1	0.00 ± 0.02	1.44 (58)	0.16	0.00, 0.01
	2	0.00 ± 0.00
T13	1	1.52 ± 0.66	2.18 (58)	0.03	0.02, 0.54
	2	1.23 ± 0.27

Figure 1:

Distribution of 13 GMs in both kinds of articles.

As can be seen from the descriptive statistics section of Table 2 and Figure 1, for most GM types, the frequency in expert RAs is higher than that in MA theses (e.g. T1, T2, T5, T6, T13, etc.), which indicates the inadequacy of students’ use of various GMs. Moreover, it is shown that most standard deviations of the expert RA group are higher than that of the MA theses group, indicating that the experts have more individual differences in GM choices.

The independent t-test results show that there are significant differences in GM use between the two author groups in T4, T5, T6, T7, T9, and T13 (p [2-tailed] < 0.05).

5.1.2 Distribution of 13 types of GMs in MA theses and expert RAs

Annular charts are used to present the percentage of 13 types of GMs in each abstract group to show the usage structures.

From Figures 2 and 3, it can be seen that the usage structures of the two groups are quite similar, with T13, T2, T5, T6, and T1 occupying the top five most frequently used GMs and sharing an alike percentage with its counterpart in the other group.

Figure 2:

The GM usage structure of MA theses.

Figure 3:

The GM usage structure of expert RAs.

Among all the GM types, T13 and T2 are the most dominant ones, with a total percentage of over 70%. T5 and T6, as means of adjectivization, also cover a relatively big proportion at 12 and 14% respectively.

5.2 Similarities between abstracts of MA theses and RAs in ideational GM use

According to the data analyses above, some similarities between the two groups can be found.

In both groups, approximately all of the 13 types of ideational GM are used in the abstracts, which gives evidence to the universal existence of ideational GMs in academic writing and shows that advanced English learners in China have a basic awareness of GM use. Also, expert RAs and MA theses share similar GM usage structures. Both the types of the most frequently used GMs and their corresponding percentages in MA theses are comparable to those in expert RAs. T13 and T2 are the most frequently used types and they are both concerned with noun transference: T2 transfers process to entity by rewording the processes as nouns which function as Thing in the nominal group rather than Process or Attribute (Halliday 2000 [1994]: 352); T13 transfers entity into expansion forms. Thus, T2 and T13 can constitute a whole procedure of packing a process into an entity which is then flexibly used to modify other components to greatly improve the information density of the academic papers. The similar dominant position of T2 and T13 in the two groups may indicate that advanced English learners in China have basically mastered such GM use to compact information in academic discourse by effectively simplifying the narrative level and structure and express the logical relationship between things more accurately and concisely. For instance, in the sentence “[…] to elicit multiple demonstrations [T2] and confirmations [T2] of understanding [T2] [T13]”, the action of “demonstrate,” “confirm,” and “understand” are transferred into an entity and discuss them concisely or use them as a modifier to better convey meanings. Also, the expression, “migrants’ [T13] settlement [T2]”, is more concise and flexible than the expression “migrants settled down here and this […]”.

Moreover, the combined use of T2 and T13 can improve the degree of objectivity and coherence. Such use can materialize processes, phenomena and characteristics, and turn them into describable objects, thereby enhancing the objectivity of the statement. In terms of coherence, the combination of T2 and T13 demonstrates logical relations among words, clauses, and sentences and plays an important role in the coherence and fluency of texts, because it condenses the information of the former expressions into nouns and uses the nouns as modifiers in the latter discourse, thus highlighting the logical relationship between the sentences. Also, students’ use of T5 and T6, as means of adjectivization, share a similar proportion with the experts’ papers, which may suggest students’ relatively mature ability to use adjectivization GMs.

As for the GM types, T1, T2, T3, T8, T10, T11, and T12, no significant differences emerge, which indicates student authors’ basic command of these GM usages. Furthermore, T1, T2, and T3 are all about nominalization, which suggests that learners already have the consciousness and ability to use nominalization GMs. However, it is found that certain T10 expressions belong to set phrases memorized by students. For example, the use of “at the same time,” “besides this,” and “in the meantime” instead of the congruent forms of “while” and “and”. Such memorized GMs may lack flexibility and conciseness, and cannot simplify the grammatical structure, thus deviating from the original intention of concise language expression of academic discourse.

5.3 Differences between abstracts of MA theses and RAs in indeational GM use

Based on the data analyses, some differences between the two groups have been identified.

5.4 Discussion

As shown in the above sections, there are differences between the abstracts of the two groups. In the following part, several possible factors are given to explain these differences.

Due to the limited learning and understanding of GM, students may not be able to use various GMs freely and appropriately, which leads to a narrow range of GM use and many repetitions. The study finds that the realization ways of GM in students’ theses lack diversity when compared to those in expert RAs. For example, a large proportion of T2 in MA theses is achieved by adding common suffixes, such as “-ing” and “-ation”, to a verb, or by using common T2 expressions, such as “practice” and “design”, which are just part of the many ways to realize T2. For instance, in the following sentences, the student writers frequently and monotonously used the common suffixes to achieve T2:

Excerpt (1): […] student’s [T13] vocabulary [T13] learning [T2], reading [T2] and listening [T2] [T13] practice [T2] than on their writing [T2], translation [T2] [T13] and oral [T5] practice [T2].

Excerpt (2): […] task [T13] presentation [T2] on students’ [T13] critical thinking [T2] performance [T2] with a view to [T10] providing [T2] some suggestions [T2] concerning [8] the cultivation [T2] of [T13] critical thinking [T2] skills via English [T13] writings [T2].

It is also found that, when discussing the thesis topics, students tend to repeatedly use the same expressions and phrases through the whole abstracts while such repetitions were annotated once according to the counting principle, which leads to lower GM frequency. This suggests students’ limited GM resources to draw upon when constructing academic discourse. The missing GM use such as T5, T6 and T9 in some MA theses also indicates that students may not have fully comprehended such GM devices so they fail to apply them properly in the abstracts.

Also, the underdevelopment of the cognitive ability required by those GM transferences may result in the deficiency of learners’ GM use and can explain why in MA theses those significant correlations between different GM types found in expert RAs are missing and why the frequency of a set of GM types is significantly lower than that in professional papers. As mentioned above, in expert RA group, there are significant correlations between T1 (quality→entity) and T2 (process→entity), as well as T8 (circumstance→process) and T9 (relator→process). The correlated types share the same elemental semantic category, entity for T1–T2 and process for T8–T9. T1 and T2 both concern nominalization, while T8 and T9 concern verbalization. Therefore, experts may possess more mature GM thinking systems so they can employ various means to achieve specific metaphorical purposes.

Moreover, according to the independent t-test, there are significant differences in GM use between the two groups in T4, T5, T6, T7, T9, and T13. It is worth mentioning that T5, T6, and T7 all belong to the same category, adjectivization, and T13 is concerned with transferring nouns to various forms and using nouns as a modifier, which shares similarities with the nature of adjectives. Thus, this gap in quasi-adjectivization may indicate a certain deficiency of the language thinking ability by which one transfers various concepts into adjectives or phrases so that they can be used to describe the character or quality of people or things. Such ability is indispensable when giving criticism and analyses in academic papers.

Learners’ underdeveloped cognitive ability may result from the lack of systematic GM training or their young age. For instance, nominalization is related to the age of users, and as Halliday (1998) points out, generally speaking, people begin to be exposed to nominalization after entering the middle school, so nominalization is a phenomenon of adult discourse. Therefore, learners may not be proficient at the relevant GM types.

On the other hand, because of the limited cognitive ability related to GM, students may regard GM usage more as an external requirement of academic writing rather than an internal and spontaneous expression device to organize one’s academic language. Therefore, part of students may use GMs for GMs’ sake instead of treating GM as a method to achieve the ultimate goal of presenting the logical connections and reasoning more clearly and plainly, which is the case in experts’ RAs. This can explain why student writers tend to overuse certain GM types and their discourse is relatively lengthy compared with experts’ writings. Meanwhile, this possibly accounts for the higher standard deviation of the expert RA group. Since professional scholars make use of various GMs as needed with flexible adjustments to various research topics, they tend to have more individual differences in GM choices.

Another possible factor may be the genre difference. According to Halliday (1996), generally speaking, the more technical the text is, the more frequently nominalization occurs; conversely, the more nominalization is used, the more technical the text becomes. Compared with abstracts of expert RAs, which introduce the frontier research results of the academia, MA theses are less complex and professional. Therefore more GMs may be needed to realize the complicated genre features of expert RAs.

Furthermore, first language effects may contribute to the differences in GM use. It is worth mentioning that English and Chinese are different in the conceptualization of adjectives. Chinese adjectives contain semantic structures of dynamic time processes, which can be extended into verb usage more easily, while English adjectives lack dynamic time processes and are more likely to be transferred into an adverb or noun usage (Huo and Liu 2009). Therefore, the deficiency of students’ use of certain GM types, such as T1 (quality→entity), may be attributed to the effects of their first language, Chinese.

6.1 Major findings

The major findings of this research involve the similarities, differences, and interpretation.

In both groups, almost all of the 13 types of ideational GMs are used indicating the universal existence of ideational GMs in academic writing and Chinese advanced English learners’ basic awareness of GMs. Expert RAs and MA theses also share similar GM usage structures with T13, T2, T5, T6 and T1 occupying the top five most frequently used GM types. Meanwhile, T2 and T13 make up a completed procedure of packing a process into an entity to be used to modify other sentence components to significantly increase the information density. For T1, T2, T3, T8, T10, T11, and T12, there are no significant differences in the two groups, showing students’ basic command of these GMs and especially their consciousness and ability to use nominalization GMs, although sometimes such use comes from memorized phrases and expressions which lack flexibility and conciseness.

Some differences have been identified. GMs are used more frequently in expert RAs and there is a statistically significant difference in GM use between the groups. The frequencies of many GM types (e.g. T1, T2, T5, T6, T13, etc.) are higher in expert RAs than in MA theses and significant differences exist particularly in T4 (relator→ entity), T5 (process→ quality), T6 (circumstance→ quality), T7 (relator→ quality), T9 (relator→ process), and T13 (entity→ expansion). The very significant correlations between T1 (quality→ entity) and T2 (process→ entity), as well as T8 (circumstance→ process) and T9 (relator→ process), are missing in MA theses group. There are also more individual differences in expert RAs’ GM use.

The students’ inability to apply GMs freely and appropriately may come from their limited knowledge of GM. Their cognitive ability, which is indispensable for GM transferences, may be inadequate due to a lack of systematic GM training or their young age, thus leading to (1) the use of different GM types as split approaches rather than a whole where elements are correlated with one another, and (2) the treatment of GM as an external requirement of academic writing rather than an internal and spontaneous expression device to organize one’s academic language. Consequently, students may neglect the ultimate goal of presenting research logically and clearly and blindly pursue GM despite the consequence of lengthy discourse. In contrast, experts tend to use GMs flexibly by adjusting the GM strategy as needed in different contexts thus having more individual differences in GM choices. Furthermore, the use of certain GM types is related to the technical levels of articles and therefore, probably due to the intricacy of the genre of professional RAs, more GMs are applied. The insufficiency of Chinese students’ GM use may also be attributed to the L1 effects which concern different ways of language conceptualization.

6.2 Implications

As shown above, Chinese students’ academic writing needs to be improved. However, there has been little literature on the research and teaching of writing and publication in international journals (Cai 2019) in college English education. Thus, there is a need for more training in academic English writing. In the process of teaching, academic English teachers should explain and emphasize different types of GMs, so as to improve students’ GM awareness and help them increase GM output. By analyzing GM examples in academic discourse, teachers can direct learners’ attention to the application of various GM types and the expressive effects achieved, laying the foundation for students’ making full use of different GM types to effectively construct professionalism and authority in academic writing. Meanwhile, the theoretical foundation of GM should be explained so that students can fully comprehend how GMs work to increase the information density and to make academic discourse more concise and professional, instead of simply memorizing phrases and blindly overusing GM expressions. Teachers may also help students develop their cognitive ability required by GM transferences to enhance their GM ability fundamentally since the significant deficiency of several GM types in students’ theses tends to involve specific cognitive processes in common. Especially, teachers can give a detailed explanation of the differences in conceptualization between English and Chinese to help students form English GM thinking. Furthermore, student researchers should be advised to read a large number of expert academic papers in English in the relevant academic field to get acquainted with GMs and process GM information proficiently, which is the basis of free and appropriate GM use.

6.3 Limitations

Despite the findings and implications, there are still limitations worth noticing. The limited sample size, 60 abstracts in total, affects the research universality; moreover, this study involves only one discipline, linguistics, while for different disciplines, the generic structures, rhetoric devices, meta-discourse use, and language expressions of abstracts may be different. Besides, GM identification and annotation were made by one researcher, which may affect the internal reliability.

6.4 Suggestions for future study

Future studies can increase the sample size to build bigger corpora containing a larger number of abstracts from different authoritative journals and student theses. Likewise, further research can extend the research scope to articles from a wider range of disciplines and may explore disciplinary differences in terms of the contrast between expert RAs and novice writings. Apart from the writers’ professional level, researchers can examine other variables such as gender and age, to find out more about GM use. Multiple identification and annotation are also advised, where researchers can first unify the annotation standards, then carry out trial labeling, discuss the differences after comparison until reaching a consensus, and subsequently start independent labeling. After the independent labeling, researchers may compare and discuss the results, and analyze the doubts together to obtain consistent results. In this way, the reliability of the research can be greatly improved.