Introduction

Interpersonal metaphor of modality is an important construct in the architecture of grammatical metaphor in Systemic Functional Linguistics or SFL. It is generally employed for the concealment of sources of information conveyed by modal expressions so as to present a proposition objectively (Tian, 2017) and ultimately to highlight the authority of discourses in question (Schleppegrell, 2004). This objective and authoritative presentation is particularly preferred in academic discourses (Halliday, 1998; Liardét, 2018; He & Yang, 2018). Halliday (1985, 1994), Halliday and Matthiessen (2004, 2014), and Taverniers (2003) argued that modal meanings could be realized either congruently by expressions of modality such as modal verbs (e.g., could, might) and modal adverbs (e.g., possibly, necessarily) within the proposition, or metaphorically by projecting clauses (e.g., it is likely, it is necessary) outside the proposition. The metaphorical realization by means of projecting clauses in SFL is hence termed as interpersonal metaphor of modality. Consider the metaphorical expression of modality exemplified in (1), wherein the projecting clause it is necessary is employed outside the proposition one is able to engage with a fictional world despite its deviations from the actual world to denote the writer’s subjective judgement of the state of affairs. In so doing, the sense of subjectivity is concealed and the objectivity of the academic discourse is highlighted.

  1. (1)

    For regular engagement experiences to occur, it is necessary that one is able to engage with a fictional world despite its deviations from the actual world. (COCA_academic_2019)

Previous studies have investigated the identification (e.g., Taverniers, 2008; Yang, 2019), categorization (e.g., Liardét, 2014, 2018; Yang, 2019), and distribution (e.g., Liardét, 2018; He, 2020; Zhou & Xia, 2024) of interpersonal metaphor of modality. There are also studies exploring its sensitivity across disciplines (e.g., MacDonald, 1994; He & Yang, 2018). However, very few studies have been conducted to examine preferences of interpersonal metaphor of modality across academic disciplines. Although He (2020) recently conducted a research to explore the distribution of interpersonal metaphor of modality in different genres and phases (or time periods) of the Corpus of Historical American English or COHA, he considered instances of metaphor of modality generally. In other words, he compared the normalized frequencies of specific subtypes of interpersonal metaphor of modality in different genres and concluded that interpersonal metaphor of modality is genre-sensitive and preferentially occurred in the academic genre. Underlying He’s (2020) findings, it is still unknown whether the specific instance it is necessary in explicit objective orientation realizing metaphor of modality is significantly preferred in academic discourses or not. In addition, normalized frequencies do not tell us precisely what specific instances of a certain subtype of metaphor of modality are preferred (Diachronic preferences of explicit subjective orientation). Although MacDonald (1994) argued that nominalization as grammatical metaphor is particularly preferred in hard sciences, it is also unclear what the situation of interpersonal metaphor of modality would be like because nominalization as grammatical metaphor is only a subtype of ideational grammatical metaphor. Furthermore, both studies did not address the diachronic shifts of interpersonal metaphor of modality across academic disciplines. Accordingly, this paper intends to examine specific instances of interpersonal metaphor of modality in different phases and academic disciplines in the Corpus of Contemporary American English or COCA by using such corpus linguistic approaches as diachronic collexeme analysis (Zhou, 2023a) and synchronic collexeme analysis (Schönefeld, 2013; Zhou, 2023a, 2023b) which are basically expanded from collexeme analysis (Stefanowitsch & Gries, 2003). Collexeme analysis is by and large implemented to investigate the attraction and/or repulsion of a certain lexical item to a grammatical construction (Approaches) that normalized frequencies fail to account for. In fulfilling our purposes, the following research questions are therefore put forward.

  1. 1.

    What are the diachronic preferences of interpersonal metaphor of modality in academic disciplines?

  2. 2.

    What are the discipline-specific preferences of interpersonal metaphor of modality?

Theoretical frameworks

A sketch on grammatical metaphor

Human language is regarded as a social semiotic system that consists of three intertwined strata, i.e., semantics, lexicogrammar, and phonology/graphology, and represents two types of meanings, i.e., ideational meaning and interpersonal meaning (Halliday & Matthiessen, 2004, 2014). Both types of meanings in the semantic stratum could be realized by grammatical items in the lexico-grammatical stratum. If the realization is embodied by a mismatch between the two strata, then grammatical metaphor occurs (cf. Halliday & Matthiessen, 2004, 2014; Hao, 2020; He, 2020; Zhou, 2023a, 2023c). The classical grammatical metaphor incorporates ideational metaphor and interpersonal metaphor. The mismatch of the former is represented by the fact that the typical construal of a sequence in the semantic stratum by a clause complex in the lexico-grammatical stratum is reconstrued by a clause or a group, and a figure typically realized by a clause is reconstrued by a groupFootnote 1. The mismatch of the latter is represented by the fact that the congruent realization of a modal meaning by a modal verb or a modal adverb is metaphorically realized by a projecting clause outside the proposition. Consider examples in (2) and (3).

  1. (2)

    a. The driver drove the bus too rapidly down the hill, so the brakes failed.

    b. The driver’s overrapid downhill driving of the bus resulted in brake failure.

    (Halliday, 1998, p. 191)

  2. (3)

    a. Although the molecular determinants of dynein force production are not well understood, it is possible that these mutations affect the ability of dynein to remain bound to microtubules under conditions of high load. (COCA_academic_2019)

    b. Although the molecular determinants of dynein force production are not well understood, these mutations possibly affect the ability of dynein to remain bound to microtubules under conditions of high load.

(2b) is the ideational metaphor of (2a) in that the mismatch is represented by the fact that the sequence the driver drove the bus too rapidly down the hill, so the brakes failed in (2a), which is realized by a clause complex, is reconstrued by the simple clause the driver’s overrapid downhill driving of the bus resulted in brake failure in (2b). Similarly, (3a) is the interpersonal metaphor of (3b) in that the modal meaning in (3b) which is realized by the modal adverb possibly is metaphorically realized by the projecting clause it is possible outside the proposition these mutations affect the ability of dynein to remain bound to microtubules under conditions of high load in (3a).

Interpersonal metaphor creates the potential of expressing the interpersonal meanings and makes propositions non-negotiable (Yang, 2019). It incorporates interpersonal metaphor of mood and interpersonal metaphor of modality (which will be enunciated in Interpersonal metaphor of modality). According to Yang (2019, p. 189), the congruent pairings of the four speech functions (i.e., offer, command, statement, and question) and the three mood types (i.e., interrogative, declarative, and imperative) are offer-interrogative/declarative, command-imperative, statement-declarative, and question-interrogative. Interpersonal metaphor of mood is created if speech functions and mood types are mismatched. For instance, if a command is not realized by an imperative but by an interrogative, then interpersonal metaphor of mood occurs. Consider the following examples.

  1. (4)

    a. Pass me the salt.

    b. Could you pass me the salt? (Yang, 2019, p. 207)

(4a) is an imperative clause which performs the function of command, whereas (4b) is an interrogative that performs the same function. Therefore, (4a) is the congruent expression and (4b) is the metaphorical expression of mood.

Interpersonal metaphor of modality

Modality is represented by degrees of modal meanings which produce four types of modality: probability (e.g., it is possible, likely), usuality (e.g., it is usual, customarily), obligation (e.g., it is obligatory, necessarily), and inclination (e.g., will) (Halliday & Matthiessen, 2014). Each of the four types could be represented by means of different orientations. Specifically, modal meanings could be expressed either subjectively or objectively, and they could also be expressed either explicitly or implicitly. Subjective orientation refers to the speaker’s own judgement on the validity of a proposition (e.g., I think, must, etc.); objective orientation refers to the way of objectifying the speaker’s evaluation (e.g., it’s likely, possibly, etc.); explicit orientation is the direct stating of the source of information in a separate clause (e.g., I think, it’s likely, etc.); implicit orientation leaves the source of information indirect (e.g., must, possibly, etc.) (Halliday & Matthiessen, 2014, p. 181). The relationship between the four types of modality and different orientations is tabulated in Table 1.

Table 1 Exemplification of types of modality and orientations (adapted from Halliday & Matthiessen, 2014, p. 693).
Full size table

Another variable of modality is modal value. A modal element could be of high, median, or low modal value. Considering the probability in explicit objective orientation (cf. Table 1), probability with low value could be exemplified by possible and likely, probability with median value by probable, and probability with high value by certain and sure. A full picture of modal values and types of modality is instantiated by modal adjectives in question and presented in Table 2.

Table 2 Instantiation of types of modality and modal values with modal adjectives in explicit objective orientationFootnote

According to Halliday and Matthiessen (2014), no instances of inclination in explicit objective orientation are identified (see also Table 1), and thus they are not considered in this paper.

.
Full size table

As to means of orientations, scholars in SFL argue that only explicit subjective and explicit objective orientations are experienced a mismatch of modal meanings in the semantic stratum and modal expressions in the lexico-grammatical stratum, and the two means of orientations are therefore regarded as realizations of interpersonal metaphor of modality (cf. Taverniers, 2003; Halliday & Matthiessen, 2014; He, 2020). However, explicit subjective and explicit objective orientations perspicuously preclude the linguistic phenomenon of modal nominalizations such as possibility and necessity which are indistinguishably categorized by Halliday (1998) into nominalizations that create ideational metaphor. As a matter of fact, modal nominalizations are also in accordance with the defining features of interpersonal metaphor of modality in that modal meanings in the semantic stratum, which are congruently realized by modal adverbs such as possibly and necessarily in the lexico-grammatical stratum, are mismatched by modal nominalizations such as possibility and necessity. With respect to the orientations, modal nominalizations realize modal meanings within the proposition and the sense of subjectivity is also concealed, and therefore they realize interpersonal metaphor of modality by means of implicit objective orientation. Accordingly, the three orientations that realize interpersonal metaphor of modality are to be enunciated in the following three subsections.

Explicit subjective orientation

Explicit subjective orientation realizing interpersonal metaphor of modality is generally expressed by a mental projecting clause in the sequence of a first personal pronoun and a mental process verb like I think, we believe, etc. This grammatical structure is regarded as “an English modal particle” or “discourse marker” by Aijmer (1997, p. 1), and as “a parenthetical construction” by Fetzer (2014, p. 68) with its preferred position in the left periphery and less preferred medial position or in the right periphery (Kaltenböck, 2009a, 2009b). It is termed as explicit subjective orientation in that this grammatical structure presents the source of information directly, i.e., explicitness, and the speaker’s judgement of the proposition, i.e., subjectivity. Instances of explicit subjective orientation that are considered in this research briefly accord with those considered by Zhou and Gao (2021) who investigated their pathway of transference from grammatical metaphor to post grammatical metaphor, i.e., I/we think, I/we believe, I/we guess, I/we assume, and I/we suppose. The only difference is that we also examined such instances as I’m sure and we are sure which demonstrate the speaker’s explicit subjectivity to the proposition at issue. Considering the examples in (5a-b), I think in (5a), I believe in (5b), and I am sure in (5c) explicitly present the writers’ subjective judgements on the propositions in question by means of projecting clauses.

  1. (5)

    a. I think those who wish to promote feminist theology in mainland China share this idea. (COCA_academic_2019)

    b. This struggle between two ideologies is, I believe, the most important problem confronting education today. (COCA_academic_1995)

    c. I am sure your Excellency realizes dangerous potentialities of continued disturbances in Lahore. (COCA_academic_2018).

Explicit objective orientation

Explicit objective orientation that realizes interpersonal metaphor of modality is structured by a projecting clause in the sequence of it, any form of the copula be, a modal adjective, and a to/that clause. The same grammatical pattern is termed as “it-extraposition” or “it-construction” by Kaltenböck (2005) and Zhou and Gao (2021) respectively, because what they highlight is the extraposed element in the construction, i.e., the to/that clause. Explicit objective orientation is employed in this research in that explicitness, as shown in the explicit subjective orientation, is realized by means of a projecting clause, and the writers’ judgement on the proposition is concealed by using the impersonal pronoun “it”. This is exemplified by (6), wherein the explicit objective orientation realizing interpersonal metaphor of modality is represented by the projecting clause it is obligatory.

  1. (6)

    Because of all of what I have mentioned it seems that it is obligatory to pay respect to the tablets of the loudspeaker, that is, to the cylinders. (COCA_academic_2018)

Explicit objective orientation could be supplemented by variations of the grammatical pattern. In other words, a projecting clause could either have a lexical item modifying the modal adjective or have a for-headed prepositional phrase interpolated in it or both. Variations as such are exemplified in (7a-c). The adverb perfectly in (7a) is employed to modify the modal adjective possible in the projecting clause it is possible; the for-headed prepositional phrase for teachers in (7b) is interpolated in the projecting clause it was possible; and both politically and for a government in (7c) are used to modify the modal adjective possible and interpolate the projecting clause it is possible respectively.

  1. (7)

    a. It is perfectly possible that the resulting welfare gains to those whose welfare depends on volume would exceed the resulting losses to entrepreneurs and labor from the decrease in the economy’s efficiency. (COCA_academic_1998)

    b. It was possible for teachers to report all of the grades they taught. (COCA_academic_2013)

    c. Are you saying that it is politically possible for a government to allow a one-hundred-percent foreign penetration rate in a market, not just a cultural market? (COCA_academic_1997)

Implicit objective orientation

Implicit objective orientation is rarely taken as a means of realizing interpersonal metaphor of modality (cf. Halliday, 1985, 1994; Halliday & Matthiessen, 2004, 2014; Yang, 2019) in that modal expressions that denote implicit objective orientation are solely delimited within modal adjuncts and left modal nominalizations (e.g., possibility) to be discussed in nominalizations that create ideational metaphor (e.g., Hao, 2020). Although He (2020) argued that implicit objective orientation could be regarded as interpersonal metaphor by means of objectification from implicit subjective orientation (e.g., will, may), implicit objective orientation realizing interpersonal metaphor of modality from He’s sense is different from what this research highlights. What we are discussing in the second paragraph of Interpersonal metaphor of modality is modal nominalizations which realize interpersonal metaphor of modality, expressing implicit objective orientation (e.g., possibility, necessity). Specific instances of implicit objective orientation that are considered in academic disciplines in this research are by and large nominalizations derived from modal adjectives such as possible and necessary. By doing so, these modal nominalizations expressing implicit objective orientation include nominalized probability such as certainty, sureness, probability, possibility, and likelihood, and nominalized obligation such as obligation, essentiality, vitality, necessity, importance, properness, and appropriateness. Owing to the lack of attested instances of nominalized usuality in the corpus, it is not considered while discussing implicit objective orientation that realizes interpersonal metaphor of modality. Nominalized probability and obligation that denote implicit objective orientation are exemplified in (8a) and (8b) respectively.

  1. (8)

    a. Considerable political momentum appears to be building toward addressing the possibility of global climate change. (COCA_academic_1991)

    b. Specifications of necessity at the level of primary rules can be found in many areas of international law. (COCA_academic_2017)

Methodology

Corpus and classification of academic disciplines

Academic disciplines that are considered in this research are sourced from the subgenres of academics in COCA. COCA is an online, genre-balanced corpus, ranging from 1990 to 2019 and containing more than one billion words of textsFootnote 3 (words that are considered in the research are exactly 120988361, because our aim is to investigate academic disciplines in the academic genre of COCA). Reasons of choosing academic genre of COCA as our data-collecting source are chiefly two-fold. One is that data of academic genre in COCA span from the year of 1990 to that of 2019, which facilitates the diachronic exploration of a certain linguistic phenomenon such as interpersonal metaphor of modality in question. The other is that academic disciplines included in academic genre of COCA could serve our purpose very well in that academic genre covers a wide range of academic disciplines such as history, education, medicine, etc.Footnote 4 Owing to our purpose of exploring academic disciplines in general, i.e., natural sciences, social sciences, and human sciences, we further categorized these specific disciplines into the three basic categories drawing on Becher and Trowler’s (2001) and Yuan et al. (2010) standards of categorizing disciplinesFootnote 5. Detailed information about the academic disciplines covered in academic genre of COCA is presented in Table 3.

Table 3 Information of academic disciplines in academic genre of COCA.
Full size table

Data collection

For sufficiently retrieving cases of interpersonal metaphor of modality in question, search queries or SQs 1, 2 and 3 are hence constructed. SQ1a, which is employed to retrieve hits of explicit subjective orientation, reads as a grammatical pattern in the sequence of a punctuation mark, a personal pronoun and a mental process verb, and SQ1b, supplementing SQ1a, means a grammatical pattern in the sequence of a punctuation mark, a personal pronoun, any form of copula be, and sure. Instances exemplified in (5a-b) could be successfully retrieved by SQ1a and that in (5c) by SQ1b.

SQ1a: [y*] [p*] think|assume|believe|guess|suppose

SQ1b: [y*] [p*] [vb*] sure

SQ2a-2d are generally constructed to retrieve hits of explicit objective orientation. SQ2a means that the grammatical construction is in the sequence of it, any form of copula be, an adjective, and a to or that clause. It could successfully retrieve such hits as exemplified in (6). After scrutinizing hundreds of the concordances, we confirmed the modal adjectives contained in this construction by using Van linden’s (2012) and Halliday and Matthiessen’s (2014) principles of identifying modal adjectives: possible, important, necessary, probable, essential, common, certain, appropriate, sure, proper, usual, vital, obligatory, customary, and likely. This SQ is supplemented by SQs 2b-2d for the purpose of attesting such hits as exemplified in (7a-c) respectively.

SQ2a: it [vb*] [j*] to|that

SQ2b: it [vb*] * [j*] to|that

SQ2c: it [vb*] [j*] for [pp*] to|that

SQ2d: it [vb*] * [j*] for [pp*] to|that

Implicit objective orientation realizing interpersonal metaphor of modality is chiefly confirmed by referring to modal adjectives attested in explicit objective orientation. We first converted these modal adjectives into their corresponding modal nominalizations and then constructed SQs respectively as exemplified in SQ3, which reads as possibility in a nominal form including both singular and plural forms. This instantiated SQ could attest such cases as exemplified in (8a).

SQ3: [possibility].[nn*]

Approaches

Variability-based neighboring clustering analysis or VNC and (diachronic and synchronic) collexeme analysis are employed in this research. VNC and hierarchical clustering algorithms or HCA are subcategories of cluster analysis. The similarity of the two cluster analyses is that both agglomerate the variables that are correlated with each other into a single cluster; the difference between the two is that HCA does not consider the sequence of these variables while VNC does (Zhou, 2023a). Precisely, VNC is to agglomerate the neighboring/adjacent similar time periods by means of the Pearson correlation analysis without altering their temporal order (Gries & Hilpert, 2008). The whole analysis of VNC is performed by the R script provided by Professor Martin Hilpert (Gries & Hilpert, 2008). Collexeme analysis is employed to examine which lexical items are significantly attracted or repelled to (strongly associated or disassociated with) a grammatical pattern, i.e., the collostructional strength (Stefanowitsch & Gries, 2003). This approach is further expanded in this research to investigate the significant attraction or repulsion to a grammatical pattern in different time phases, i.e., diachronic collexeme analysis (Zhou, 2023a), and in different academic disciplines, i.e., synchronic collexeme analysis (Schönefeld, 2013; Zhou, 2023a, 2023b). Owing to the fundamentality of collexeme analysis proposed by Stefanowitsch & Gries (2003), its operationalization is expounded in the following paragraph.

Feasible operation of a collexeme analysis requires the availability of four numbers (cells a, b, c, and d) in a contingency table shown in Table 4, which is instantiated by the modal adjective possible in the explicit objective orientation realizing interpersonal metaphor of modality. As a matter of fact, a, a + b, a + c, and a + b + c + d in Table 4 could be directly obtained from the corpus, and the others by mathematical calculations. Computation of the collostructional strength of possible in explicit objective orientation is performed by inputting the .txt file to the function of coll.analysis in R language provided by Gries (2014). The statistical significance of association is computed by means of the log likelihood ratio or G2 which is a method of identifying collostructional strength, and a result of 3 shows a statistical significance at the level of 0.05 and 10 at the level of 0.001.

Table 4 Exemplification of the collexeme analysis by the modal adjective possible in the explicit objective orientation.
Full size table

Diachronic preferences of interpersonal metaphor of modality in academic disciplines

Confirmation of time phases in COCA

Preceding the diachronic investigation of preferences of interpersonal metaphor of modality in academic disciplines, a VNC analysis is implemented to confirm the most appropriate time periods in COCA for the purpose of scientifically uncovering the diachronic preferences of interpersonal metaphor of modality in academic disciplines. The unfeasibility of the evenly clustered six phases provided by COCA (each phase consisting five years, ranging from 1990 to 2019) lies in that it does not tell us whether adjacent time phases are significantly different from each other. By implementing VNC which uses the correlation test, those adjacent phases that are significantly correlated are clustered together; in so doing, it will facilitate our investigation of diachronic changes during the years provided by COCA.

Results of VNC analyses are presented in the form of dendrograms in Figs. 1–3. It should be noted that the year in these figures refers to a range of years, e.g., 1995 ranges from 1995 to 1999. Figure 1 suggests three clusters of occurrences of explicit subjective orientation in COCA.Footnote 6 Specifically, the first cluster ranges from the year of 1990 to that of 2009, the second from 2010 to 2014, and the third from 2015 to 2019. Figure 2 also suggests agglomerating occurrences of explicit objective orientation into three clusters; precisely, the first cluster contains fifteen years from 1990 to 2004, the second cluster from 2005 to 2014, and the third from 2015 to 2019. In accordance with the previous two dendrograms, three clusters are read off from Fig. 3, which clusters the years from 1990 to 2009 together, years from 2010 to 2014, and years from 2015 to 2019 respectively.

Fig. 1
figure 1

Dendrogram of VNC analysis of explicit subjective orientation.

Full size image
Fig. 2
figure 2

Dendrogram of VNC analysis of explicit objective orientation.

Full size image
Fig. 3
figure 3

Dendrogram of VNC analysis of implicit objective orientation.

Full size image

Diachronic preferences of interpersonal metaphor of modality

Diachronic preferences of explicit subjective orientation

By implementing a diachronic collexeme analysis, Table 5 presents the results showing preferences for explicit subjective orientation within the academic genre of the COCA. It shows that only mental process verbs think (coll.s = 140.36), believe (137.48), and guess (71.81) in mental projecting clauses are significantly attracted to the explicit subjective orientation in phase 1, and this significant attraction is transferred either into insignificant level except believe (49.46) in phase 2 or into a level of being non-attestable in phase 3. This diachronic change of preferences of explicit subjective orientation indicates that academic writers are gradually avoiding using explicit subjective orientation to realize interpersonal metaphor of modality during the past 30 years. The plausibility of this diachronic change lies in that the concealment of the academic writer’s subjective judgment of the proposition in question is to foreground the objectivity of the writer’s ideas.

Table 5 Preferences of explicit subjective orientation in three phasesFootnote

Numbers in bold in all tables of this paper stand for statistical significance and negative numbers stand for repulsion; N/A stands for being non-attestable.

.
Full size table

For the purpose of testifying the necessity of implementing such a complicated computational procedure instead of simply relying on the observed or normalized frequencies, a correlation test between observed frequencies (observed frequencies and normalized frequencies do not make any difference at issue in that both are considered under the same sub-corpus, i.e., the first phase in COCA) and collostructional strength in phase 1 is conducted. The result shows that they are not significantly correlated (ρ = 0.8286, p = 0.0583 > 0.05), indicating that the observed frequencies do not suffice to tell the association between mental process verbs and explicit subjective orientation. Therefore, our extra effort of computing the collostructional strength between the two is warranted.

Diachronic preferences of explicit objective orientation

Diachronic collexeme analysis is implemented in this subsection to investigate changes of preferences of explicit objective orientation realizing interpersonal metaphor of modality. Results are tabulated in Table 6.

Table 6 Preferences of explicit objective orientation in three phases.
Full size table

Table 6 demonstrates that modal adjectives such as possible (coll.s = 1620.7 in phase 1, 1239.04 and 737.8 in phases 2 and 3 respectively), important (659.32, 1060.78 and 363.5 in phases 1, 2 and 3 respectively), and necessary (205.53, 17.88 and 103.5 in phases 1, 2 and 3 respectively) are significantly attracted to the grammatical pattern of explicit objective orientation in three phases (probable is significantly attracted in phases 1 (84.28) and 2 (30.79), while insignificantly attracted in phase 3 (1.51)). Concerning types of modality (cf. Table 1), probability with median (i.e., probable) and low (i.e., possible) modal values and obligation with median modal value (i.e., important and necessary) are significantly attracted to explicit objective orientation in three phases except for the insignificant attraction of probable in phase 3. This indicates that academic writers prefer to use probability with low and median modal values and obligation with median modal value to present their judgement on a proposition in question while concealing their subjectivity. With respect to usuality, all modal adjectives in three phases (−1641.1, 1384.4 and 697 in phases 1, 2 and 3 respectively for common; −205.71, −133.6 and −44.5 in phases 1, 2 and 3 respectively for usual; −10.53, −6.63 and −9.42 in phases 1, 2 and 3 respectively for customary) are significantly repelled to the grammatical construction, indicating that usuality at present is not a sufficient means of concealing academic writers’ subjective judgement on the proposition discussed.

It is intriguing to know whether this diachronic attraction and repulsion of modal adjective to explicit objective orientation will be transferred to their derived nominalizations denoting implicit objective orientation, and this will be examined in the following subsection.

Diachronic preferences of implicit objective orientation

Collexeme analysis in the lens of diachrony is employed to examine which modal nominalizations are preferred to the implicit objective orientation that realizes interpersonal metaphor of modality. Results are shown in Table 7.

Table 7 Preferences of implicit objective orientation in three phases.
Full size table

Table 7 informs us of three aspects. First, modal nominalizations such as importance (coll.s = 6818.7 in phase 1, 1764.4 and 787.5 in phases 2 and 3 respectively), obligation (2684.3, 200.8 and 510.2 in phases 1, 2 and 3 respectively), possibility (1236.6, 23.9 and 198.9 in phases 1, 2 and 3 respectively), and probability (332.8, 724.7 and 1660.5 in phases 1, 2 and 3 respectively) are significantly attracted to implicit objective orientation in three phases of COCA, and vitality is significantly attracted to this grammatical pattern in phase 1 (22.5) but subsequently repelled in phases 2 (−10) and 3 (−11.6). Second, concerning types of modality, probability with median and low values, i.e., probability and possibility, is significantly attracted to this pattern in three phases; obligation with high and median modal values, i.e., obligation and importance, (inclusive of vitality in phase 1) is significantly attracted to this pattern in three phases too; no modal nominalizations denoting usuality is attracted to implicit objective orientation. This significant attraction indicates that probability and obligation are still the most commonly employed means for academic writers to express their judgments on the proposition in question. In addition, even within the same type of modality, competition among modal nominalizations still exists, e.g., competition between possibility and likelihood within probability. Third, compared with explicit objective orientation, only probability with low value, i.e., possible or possibility, and obligation with median value, i.e., important or importance, keep being significantly attracted to their corresponding explicit objective orientation and implicit objective orientation respectively. This intact attraction through time indicates possible and important in explicit objective orientation and possibility and importance in implicit objective orientation are two commonly used means for academic writers to hide their subjective judgment on the proposition at issue. It also indicates that these writers usually express their judgment in an indirect manner (e.g., by using importance to indirectly denote the necessity of preforming the state of affairs) with a median or low modal value (e.g., importance or possibility).

Discipline-specific preferences of interpersonal metaphor of modality

The previous section has investigated the diachronic preferences of interpersonal metaphor of modality in the academic genre by employing diachronic collexeme analysis in accordance with the first research question. For the purpose of addressing the second research question, the preferences of interpersonal metaphor of modality in specific academic disciplines are highlighted in this section.

Preferences of explicit subjective orientation in academic disciplines

In the analysis of the preferences of explicit subjective orientation in academic disciplines, a synchronic collexeme analysis has been executed. The findings of this analysis are delineated in Table 8. It shows that academic writers seldomly employ explicit subjective orientation to realize interpersonal metaphor of modality in natural sciences because only the mental process verb believe (coll.s = 248.54) in the mental projecting clause is significantly attracted to this grammatical pattern. Pertaining to social sciences and human sciences, process verbs such as think (161.22 in social science and 12.61 in human science), guess (66.51 in social science and 5.96 in human science), and believe (14.86 in social science and 96.66 in human science) in mental projecting clauses are significantly attracted to explicit subjective orientation, indicating that writers from the two academic disciplines are likely to impose their subjective judgment while commenting on a proposition in question. Their lacking empirical evidences prompt them to use more negotiable expressions such as I/we think, I/we guess; in doing so, academic writers themselves are more liable to entertain other voices so as not to take the responsibility.

Table 8 Preferences of explicit subjective orientation in academic disciplines.
Full size table

Preferences of explicit objective orientation in academic disciplines

Implementation of synchronic collexeme analysis on preferences of explicit objective orientation in different academic disciplines (see Table 9) shows that probability with low modal value, i.e., possible (827.3, 1931 and 930.4 in natural sciences, social sciences and human sciences respectively), and obligation with median value like important (439, 1393.8 and 227.7 in natural sciences, social sciences and human sciences respectively) and necessary (91.9, 42.2 and 188.4 in natural sciences, social sciences and human sciences respectively) are significantly preferred in all academic disciplines. This common preference indicates that types of modality as such are not discipline-sensitive and used as a common means for academic writers to conceal their judgment on the proposition being discussed. The three academic disciplines also display some differences in terms of preferences of explicit objective orientation. Academic writers in natural sciences, on the one hand, significantly use the usuality of modality, i.e., customary (4.7) to realize interpersonal metaphor of modality, and this preference is absent in other academic disciplines. On the other hand, probability with median value, i.e., probable (34.1 and 101.4 in social sciences and human sciences respectively), are significantly attracted to explicit objective orientation in both social sciences and human sciences but not in natural sciences (1.9). These discipline-sensitive types of modality as mentioned above are actually in accordance with the intrinsic natures of academic disciplines in that academic writers in social sciences and human sciences rely more on their own subjective judgment on the proposition while lacking empirical evidence.

Table 9 Preferences of explicit objective orientation in academic disciplines.
Full size table

Whether the preferences of explicit objective orientation remain consistent in academic disciplines when realizing the interpersonal metaphor of modality merits further investigation, as will be detailed in Preferences of implicit objective orientation in academic disciplines.

Preferences of implicit objective orientation in academic disciplines

By implementing a synchronic collexeme analysis, preferences of implicit objective orientation in academic disciplines are uncovered and the results are presented in Table 10.

Table 10 Preferences of implicit objective orientation in academic disciplines.
Full size table

According to Table 10, modal nominalizations denoting probability with low modal value such as possibility (36.4, 282.1, and 1362.1 in natural sciences, social sciences and human sciences respectively) and obligation with high or median modal value such as obligation (262.4, 1647.2, and 1146.5 in natural sciences, social sciences and human sciences respectively) or importance (1169.6, 5173.4 and 2630.5 in natural sciences, social sciences and human sciences respectively) are significantly attracted to implicit objective orientation in all academic disciplines. These commonly preferred modal nominalizations denoting implicit objective orientation as such are in line with the common preferences of explicit objective orientation in that both belong to probability with low modal value, i.e., possible and its nominalized counterpart possibility, and obligation with median modal value, i.e., important and its nominalized counterpart importance. The inconsistency lies in that the commonly preferred obligation in explicit objective orientation is the modal adjective necessary that denotes median modal value while that in implicit objective orientation is the modal nominalization obligation that denotes high modal value. The phenomenon may be attributed to the propensity of academic authors to obscure their subjective inclinations through the strategic suppression of interpersonal meanings, while simultaneously enhancing the salience of ideational meanings. Interpersonal meaning enacts the interaction between interlocutors and hence is subjectivity-oriented while ideational meaning construes experiences in human world and thus is state of affairs- or proposition-oriented.

There are also some differences in employing implicit objective orientation to realize interpersonal metaphor of modality in three academic disciplines. Probability that denotes probability with median value in both natural sciences (2403.2) and social sciences (1079) is significantly attracted to implicit objective orientation, but this significant attraction is absent in human sciences (−61.6). Additionally, likelihood that expresses low modal value of probability is significantly attracted to this grammatical pattern in natural sciences (6.3) while significantly repelled in the other two academic disciplines (−91.7 in social sciences and −243.9 in human sciences); vitality and necessity that denote median modal value of obligation are significantly attracted to implicit objective orientation in human sciences (19.6 and 5.4) while (significantly) repelled in natural sciences (−20.4 and −566) and social sciences (−0.1 and −475.6).

Discussion

In this section, diachronic preferences of interpersonal metaphor of modality and their synchronic discipline-sensitive natures in terms of three means of realizations are discussed and the underlying reasons are subsequently expounded.

On diachronic preferences

A VNC analysis clusters the original six phases in COCA into three ones concerning the three means of realizations. The significantly attracted process verbs in projecting clauses that denote explicit subjective orientation in phase 1 include think, believe, and guess, and during the second phase only believe keeps the significant attraction. In phase 3, no instances of explicit subjective orientation are attested in the corpus. Explicit subjective orientation like I/we think present the writer’s subjective judgment on a state of affairs directly, which is inconsistent with the fact that ideas in academic discourse should be articulated in an objective way so as to highlight the non-negotiability (Yang, 2019) and authoritativeness (Schleppegrell, 2004) of academic texts. The reason might be that academic writers in the early phases of research may focus less on the objectivity of their writings due to the challenges inherent in conducting empirical studies. The scarcity of empirical data can lead writers to utilize alternative methods to convey their assessments, such as employing subjective language or explicit stance markers to articulate their perspectives on the propositions under discussion (Barbara et al., 2024). However, with the development of science and technology, many studies are gradually equipped with advanced technological devices (e.g., empirical studies drawing on Magnetic Resonance Imaging in natural sciences), which significantly foregrounds the factuality and accuracy of the empirical results. Consequently, explicit subjective orientation is diminished gradually in phase 2 and even avoided in phase 3 (cf. Table 5).

Modal adjectives that are significantly attracted to explicit objective orientation in three phases include probability with low value, i.e., possible, and obligation with median value, i.e., important and necessary. This significant attraction throughout the phases shows that the three modal adjectives denoting explicit objective orientation are used as a common means by academic writers to conceal their subjective judgments. What difference lies in the three phases is that the modal adjective probable is significantly attracted to explicit objective orientation in phases 1 and 2 but not in phase 3; and the degree of significant attraction is diminishing gradually from phase 1 (84.28), through phase 2 (30.79), to phase 3 (1.51). This gradual reduction is mirrored by the use of explicit subjective orientation (e.g., I think) in academic discourses, because probable that denotes probability with median value could not effectively mask the subjective judgments of scholars regarding the proposition under scrutiny (Zhou, 2023b). Another possible reason might be that modal adjectives that denote median or high modal value are replaced by other means such as modal nominalizations, because modal adjectives are important evaluative linguistic means expressing interpersonal meanings while the evaluative interpersonal meaning is sufficiently concealed by modal nominalizations, denoting implicit objective orientation, which express ideational meanings in nature. This possible reason is verified in the following paragraph while discussing implicit objective orientation.

Modal nominalizations that are significantly attracted to implicit objective orientation in all phases at issue include possibility denoting probability with low value and importance denoting obligation with median value. This significant attraction is consistent with possible and important in explicit objective orientation, indicating that one of the common features of implicit objective orientation is to express academic writers’ judgment by means of low modal value of probability, i.e., possibility, and median modal value of obligation, i.e., importance. Another paramount feature in using implicit objective orientation (to realize interpersonal metaphor of modality) is to highlight the median modal value of probability and high modal value of obligation. Specifically, probability that denotes median modal value of probability is significantly attracted to implicit objective orientation in phase 1 (332.8), and this significant attraction is enhanced in phase 2 (724.7) and peaked at phase 3 (1660.5); obligation that denotes high modal value of obligation is significantly attracted to the grammatical pattern in all phases. The underlying reason might be that the adjectival counterparts of probability and obligation, i.e., probable and obligatory (which denote interpersonal meaning in nature), are taken as evaluative resources (cf. Martin & White, 2005) that directly express academic writers’ assessment of the proposition, while these evaluative resources that express high value of writers’ assessment are sufficiently concealed by such modal nominalizations as probability and obligation which denote ideational meaning in nature. Consider examples (9a-b).

  1. (9)

    a. The administration had a professional obligation to inform the judiciary about those false allegations but failed to do so. (COCA_academic_2019)

    b. It was professionally obligatory for the administration to inform the judiciary about those false allegations but failed to do so.

The academic writer’s assessment of the proposition to inform the judiciary about those false allegations in (9b) is directly articulated by the modal adjective obligatory in the projecting clause it was professionally obligatory, which to some extent violates the “maxim” that academic discourses should be presented objectively. Nevertheless, this violation is sufficiently avoided by the use of the modal nominalization obligation in (9a) which is integrated into the proposition in question. In other words, the interpersonal meaning denoted by the modal adjective obligatory in the projecting clause is expressed by the modal nominalization obligation that denotes ideational meaning metafunctionally.

On synchronic preferences

Synchronic collexeme analysis of preferences of explicit subjective orientation in terms of academic disciplines demonstrates that process verbs such as believe, think, and guess in mental projecting clauses are significantly attracted to the grammatical pattern. Specifically, only believe is discipline-insensitive to all academic disciplines in question, and think and guess are discipline-insensitive to both social sciences and human sciences. Taking diachronic preferences of explicit subjective orientation into consideration, the preference of believe to this pattern in all academic disciplines occurs generally in the first phase of academic genre of COCA; preferences of think and guess to explicit subjective orientation probably occur in the first and second phases. Concerning the fact that think and guess are significantly attracted to explicit subjective orientation in social sciences and human sciences but not in natural sciences, academic writers in natural sciences report experiment-based results objectively and they seldom use this pattern to impose their subjective judgment which will doubtlessly attenuate the authenticity and reliability of their statements.

The modal adjectives with significant attraction to explicit objective orientation in all academic disciplines incorporate possible, important, and necessary, according with the significant attraction of modal adjectives in the three phases concerned diachronically. This insensitivity in both academic disciplines and the three phases prompts the conclusion that explicit objective orientation by and large denotes probability with low value, i.e., possible, and obligation with median value, i.e., important and necessary. The discipline-sensitive modal adjective that needs to be expounded in explicit objective orientation is probable. It is significantly attracted to this pattern in both social sciences and human sciences but not in natural sciences chiefly because of, as mentioned in Significance and implications, the advancement of empirical devices that facilitate academic writers to report findings confidently and objectively or the employment of other objective means like modal nominalizations. Associating the preference of probable to this pattern in academic disciplines with that in three phases of COCA, it might be concluded that the preference of probable to explicit objective orientation in social sciences and human sciences primarily occurs in earlier phases of COCA, and this preference gradually fades away as the research in the two academic disciplines is also conducted empirically. Considering the example in (10) which is sourced from the social science International Journal of Entrepreneurship, the writer presents the proposition the results of the study provided an overview of the relationship between the personality traits of the participants and the entrepreneurial intention objectively without using any modal elements in that this proposition is supported by the empirical evidence the locus of control and the entrepreneurial intention has a positive relationship between them as the correlation coefficient is 0.850 that is positive.

  1. (10)

    The results of the study provided an overview of the relationship between the personality traits of the participants and the entrepreneurial intention. For example, the locus of control and the entrepreneurial intention has a positive relationship between them as the correlation coefficient is 0.850 that is positive. (COCA_academic_2019_International Journal of Entrepreneurship).

Modal nominalizations that are significantly attracted to implicit objective orientation in academic disciplines include possibility and probability that denote probability with median or low value, and importance and obligation that denote obligation with median or high value. The fact that the significant preferences of implicit objective orientation in academic disciplines are consistent with those in three phases of COCA prompts the conclusion that implicit objective orientation generally denotes such modal meanings as probability with median or low value, i.e., possibility and probability, and obligation with median or high value, i.e., importance and obligation. As explained in Significance and implications, probability and obligation are used as a sufficient means to conceal academic writers’ subjective assessments of their statements in that modal nominalizations generally realize ideational meaning while their adjectival forms generally realize evaluative meanings which belong to the scope of interpersonal meaning. Another interesting phenomenon is the competition between importance and necessity. Although both modal nominalizations denote obligation with median value in implicit objective orientation, it seems that academic writers usually prefer importance to necessity (Table 10). The possible reason is that necessity implicates the directness of fulfilling a performance and importance highlights the indirectness; indirectness, compared with directness, is a better way for academic writers to conceal their subjective judgment and also more feasible for them to align with potential readers.

Significance and implications

Significance of this research is at least four-fold. Firstly, it contributes to the field of SFL by examining the use of grammatical metaphor, specifically interpersonal metaphor of modality, in academic discourse. It builds on the work of scholars like Halliday and Matthiessen (2014), who have discussed the concept of grammatical metaphor, and extends this research by looking at how it manifests in different academic disciplines and over time. Secondly, it employs both diachronic and synchronic collexeme analyses to investigate the preferences of interpersonal metaphor of modality in academic writing. This approach provides a nuanced understanding of how language use in academic writing evolves and how it varies across disciplines, which is a valuable addition to research on academic literacies and language variation (Zhou, 2023a). Thirdly, the paper’s insights into how different disciplines use interpersonal metaphor of modality could help improve interdisciplinary communication. Understanding the linguistic preferences of various fields can facilitate clearer exchanges and more effective collaboration across disciplinary boundaries. Fourthly, the use of corpus linguistic methods, such as VNC and collexeme analysis, represents a methodological advance in the study of academic writing. These techniques allow for a detailed examination of language use that can reveal patterns and preferences that might not be apparent through more qualitative analyses (e.g., Gries & Hilpert, 2008; Schönefeld, 2013; Zhou, 2023a, 2023b).

This research also has at least four aspects of implications on academic writing. First, diachronic and synchronic preferences of explicit subjective orientation that realizes interpersonal metaphor of modality implicate that it is not appropriate for academic writers to employ explicit subjective orientation to conceal their personal assessment of the proposition being discussed. They should use other more impersonal means such as explicit objective orientation and/or implicit objective orientation. Second, academic writers should be aware that, while employing explicit objective orientation, probability with low modal value like possible and obligation with median modal value such as important and necessary are insensitive to both diachronic academic discourses and academic disciplines. In other words, these means do not distinguish academic disciplines such as natural sciences, social sciences, and human sciences. What actually distinguishes them is the use of probable that denotes median modal value of probability in that it is significantly preferred by academic writers from social and human sciences while significantly repelled by those from natural sciences. Third, it is suggestive for academic writers to employ implicit objective orientation while expressing their subjective opinions on what is evaluated. Specifically, probability and obligation are suggested in that both are capable of sufficiently hiding academic writers’ intensions to denote median modal value of probability and high modal value of obligation. Finally, diachronic and synchronic competitions between importance and necessity that denote median modal value of obligation imply that importance, because of its indirectness, is a better means for academic writers to conceal their subjective judgments than necessity.

Conclusion

Diachronic preferences of interpersonal metaphor of modality in academic genre of COCA demonstrate that explicit subjective orientation gradually fades away and even is avoided as a means to realize interpersonal metaphor of modality because of its foregrounding of the academic writers’ subjective assessment of the proposition; explicit objective orientation significantly attracts such modal adjectives as possible that denotes low modal value of probability and important and necessary that denote median value of obligation; and implicit objective orientation in question significantly attracts such modal nominalizations as possibility and probability that denote low and median values of probability and importance and obligation that denote median and high values of obligation in three phases, because modal nominalizations which primarily realize ideational meaning are employed as a sufficient means of concealing academic writers’ evaluative meanings within the interpersonal meaning.

Synchronic preferences in academic disciplines demonstrate that explicit subjective orientation in natural sciences only significantly attracts believe in the projecting clause and it also significantly attracts think and guess in social sciences and human sciences owing to the fact that natural sciences are more empirically evidenced than the other two types of academic disciplines. Explicit objective orientation seems to be discipline-insensitive in that all types of academic disciplines significantly attract possible that denotes low value of probability and important and necessary that denote median value of obligation. What implicit objective orientation significantly attracts in academic disciplines includes not only possibility and importance that denote low value of probability and median value of obligation respectively, but also probability and obligation that denote median value of probability and high value of obligation respectively. The reason of this is that modal nominalizations in implicit objective orientation realizing ideational meanings are more sufficient to cover academic writers’ subjective assessment than modal adjectives in explicit objective orientation realizing evaluative meanings in terms of interpersonal meanings.

There are at least two issues that this paper fails to cover. One is that it delimits the discussion within data retrieved from COCA which is typical of American English and leaves other dialectal Englishes such as British English untouched. The other is that the time period for accounting for preferences of interpersonal metaphor of modality in academic disciplines covers only thirty years as diachronic changes might take even longer years to complete. Future researches are therefore suggested to consider other dialectal Englishes and longer years while examining preferences of interpersonal metaphor of modality.