Register-based distribution of expressions of modality in COCA

This study, grounded in Systemic Functional Linguistics (SFL), explores the distribution of expressions of modality across various registers of formality. Halliday and Matthiessen's (2014) framework posits that modality is realized lexicogrammatically within or outside the clause. Previous research has either focused on specific expression types (e.g., modal verbs, adjuncts) or limited analysis to a single rank. This study addresses these limitations by examining a broader range of expressions at both word (modal verbal operators, modal adjuncts, modal nominalizations) and clause (explicit subjective and objective orientations) ranks. The research questions are: 1. How are expressions of modality distributed across registers? 2. What is the relationship between modality expressions and register formality, and what explains this relationship? Understanding this relationship is crucial for comprehending how formality influences meaning-making and for the broader application of SFL's interpersonal metaphor of modality theory.

Three lines of research inform this study. The first line examines expressions of modality based on differing theoretical frameworks, often focusing narrowly on subtypes (e.g., modal verbs or adjuncts) without fully accounting for their dynamic nature or register influence. The second line focuses on the diachronic and synchronic distribution of modal verbal operators, neglecting a comprehensive scope of expression types and the effect of register. The third line examines the register-based distribution of expressions of modality but suffers from limitations such as incomplete subtype coverage or exclusion of specific registers (e.g., academic). This current research builds upon these studies to provide a more comprehensive analysis, encompassing a fuller range of expression types and thoroughly considering the influence of register formality.

This study utilizes the Corpus of Contemporary American English (COCA), a large, genre-balanced corpus of American English (1990-2019). The registers analyzed are spoken texts, fiction, magazines, newspapers, and academic texts. Formality is established using Heylighen and Dewaele's (1999) formula, which considers the relative frequencies of different word classes (nouns, verbs, etc.). Data collection involved constructing specific search queries (SQs) in COCA to retrieve instances of each expression type. Seven SQs were used to identify: 1) modal verbal operators, 2) modal adjuncts, 3) modal nominalizations, 4) 'explicit subjective' orientation, 5) 'explicit objective' orientation (and variations). Raw frequencies were normalized to per million words (PMW). Statistical analysis in R utilized the Shapiro-Wilk normality test, Student's t-test, Mann-Whitney test, Pearson's correlation coefficient test, and Spearman's correlation coefficient test to examine variations and correlations between expression types and register formality. Data visualization was conducted using Excel.

Figure 1 displays the normalized frequencies of expression subtypes across registers. Modal verbal operators with high values (e.g., *must*, *shall/should*) showed no significant register preference. Low-value operators (*can/could*) were more frequent in informal registers. Low-value (*may/might*) and median-value (*will/would*) operators showed competition in formal registers. High and median-value modal adjuncts (*certainly*, *probably*) were more frequent in informal registers. Modal nominalizations (*possibility*, *obligation*) were less frequent in fiction but more frequent in formal registers, especially academic texts. 'Explicit subjective' orientations (e.g., *I/we think*) were most frequent in informal registers, while 'explicit objective' orientations were more prevalent in formal registers, particularly academic texts. Table 4 presents the results of variation analysis, showing significant differences between subtypes and register preferences. Significant correlations between subtypes and formality were revealed in Table 5. For example, *will/would* correlated negatively with *must*, while *probably* and *possibly* correlated positively. Modal nominalizations correlated significantly positively with 'explicit objective' orientations. Figure 1f shows that modal verbal operators are the most frequently used type across registers, especially in informal ones, followed by modal adjuncts and then modal nominalizations. The overall findings demonstrate the interplay between modality expression, value, and register formality.

The findings indicate a preference for low and median-value expressions of modality across registers to accommodate various perspectives and mitigate the forcefulness of assertions. The higher frequency of congruent expressions (modal verbs, adjuncts) compared to metaphorical ones (explicit subjective/objective) in most registers, except for academic texts, could be attributed to the accessibility and implicitness of the former. In formal registers, the preference for modal nominalizations and 'explicit objective' orientations reflects a move towards greater objectivity and reduced speaker prominence. This aligns with the established convention of objectivity in academic writing. The choice of modality expressions is driven by a desire to balance the expression of meaning with the maintenance of face and the acknowledgement of multiple viewpoints.

This research offers a comprehensive analysis of modality expression distribution across various registers, highlighting the interplay between modality value, expression type, and register formality. The study's key contribution lies in its expanded scope of modality expression types, a more nuanced treatment of register influence, and the integration of interpersonal metaphor of modality theory. Future research could explore diachronic trends, extend analysis to other English varieties, and include negative formulations of modality for a more complete picture.

The study is limited to the synchronic distribution of modality expressions in COCA (American English), neglecting diachronic aspects and potentially limiting generalizability to other English varieties. Additionally, only positive formulations of modality were considered, potentially skewing the results. Future studies should address these limitations to gain a more comprehensive understanding of modality across registers and time.