Keeping balance between loyalty and modification: a Toulminian model as analytical framework

The paper addresses limitations in widely used modifications of Toulmin's argumentation model and the paucity of empirical, logic-focused analyses of modern Chinese argumentative essays. Building on Toulmin's six-component model and integrating insights from pragma-dialectics (viewing argumentation as a two-party critical discussion), the authors aim to develop a revised analytical framework that remains loyal to Toulmin while capturing dialogical and hierarchical structures in real essays. The framework assumes a writer arguing with a skeptical reader who may hold opposing views. The study applies this framework to Chinese argumentative essays to examine: (1) how Toulmin components represent features of Chinese writers' argumentation; (2) preferences in argument width (parallel support) versus depth (hierarchical support); and (3) tendencies in types of warrants used.

The paper reviews extensive use of Toulmin's model in composition-rhetoric, including as a tool for assessing component quality and for teaching argumentative writing. Several modified models are discussed. Crammond (1998) broadened qualifiers to include constraints, split backing into warrant and data backing, added alternative solutions, countered rebuttals, and reservations, classifying claim and data as necessary and others as optional; however, this departs from Toulmin's logical stratification of claim/data versus justification (warrant/backing). Qin and Karabacak (2010) removed warrant, backing, and qualifier and split rebuttal into multiple components at the same level, further diverging from Toulmin by omitting the reasoning link from data to claim. The paper examines conceptual issues around warrant and data: warrant has been variously treated as rules, inference licenses, or intuition-based categories (a priori, empirical, institutional, evaluative). The authors argue warrant in real essays can be sentences, clusters, or paragraphs that guide readers from grounds to claims, adding a descriptive warrant type (Liu, 2020). Prior definitions of data/ground often overlap with warrant (e.g., causal explanations, logical analysis) but Toulmin et al. (1978) clarified ground as particular facts accepted as true. The review also surveys scholarship on Chinese rhetoric, noting debates about similarities and differences with Anglo-American rhetoric and the scarcity of empirical Toulmin-based analyses of modern Chinese argumentative writing.

Design: Content-analytic study applying a newly proposed Toulminian-pragma-dialectic analytical framework to argumentative essays. The framework distinguishes claim, justification-side components (J-ground, J-warrant, J-rebuttal), opposition-side components (O-ground, O-warrant, O-rebuttal), and qualifier. It models argument width (number of parallel justification-qualifier structures) and depth (number of hierarchical supporting levels). Backing is excluded, with deeper J-Q structures serving supportive functions. Warrants may be sentences, sentence clusters, or paragraphs; qualifiers may be words, phrases, or sentences. Data sources: 60 Chinese-language argumentative essays: 30 newspaper column/commentary essays (Southern Weekend, Southern Urban Daily, Guangming Daily) from June 2019 to June 2020 across topics (AI, education, family, consumption, culture); and 30 full-score National Matriculation Chinese Test (NMCT) compositions (2011–2020) sourced from an open-access website (Zuowen Wang). Column writers had topic freedom and no time constraints; high schoolers wrote under timed exam conditions on assigned topics. Identification and coding: Components were identified primarily by logic and meaning, aided by linguistic cues: claims by explicit assertions or metadiscourse (e.g., "personally, I think"), grounds by example markers (e.g., "for example"), warrants by causal/connective markers (e.g., "since", "because"). Due to parataxis in Chinese, semantic/pragmatic judgment predominated. Annotation order: claim > J-ground > J-warrant > J-rebuttal > O-ground > O-warrant > O-rebuttal > qualifier. Claim frequency per essay was counted as 1 if present; other components were counted in total per essay. J-width (parallel J-Qs) and J-depth (levels of hierarchical support) were computed; analogous measures for opposition were O-width and O-depth. In cases of inconsistent identification, double checking was conducted. Reported agreement included O-warrant 95% and O-rebuttal 100%. Analysis: Descriptive statistics of component frequencies; inter-group comparisons via Mann–Whitney U tests; intra-group width vs depth comparisons via Wilcoxon tests. Warrants were categorized into five types: a priori, empirical, institutional, evaluative, and descriptive.

- Component frequencies across all essays (Mean, SD): Claim 0.99 (0.09); J-ground 7.82 (3.41); J-warrant 2.42 (0.92); J-rebuttal 0.70 (0.90); O-ground 0.52 (0.95); O-warrant 0.23 (0.60); O-rebuttal 0.40 (0.76); Qualifier 0.23 (0.45); J-width 2.81 (1.09); J-depth 1.07 (0.17); O-width 0.68 (0.68); O-depth 0.48 (0.51). - Writers provided far more J-grounds than J-warrants (7.82 > 2.42). Claims were almost always present. Oppositional components and qualifiers were infrequent overall, with O-warrant and qualifier the lowest. - Width vs depth: J-width significantly exceeded J-depth for column writers, high schoolers, and all writers combined (Table 4; e.g., All writers: test statistic 6.61, p < 0.001). No significant difference between O-width and O-depth. - Group comparisons: No significant inter-group differences for most components except that column writers produced more J-warrants and more qualifiers than high schoolers. - Preference patterns: Writers favored parallel arguments (greater J-width) over hierarchical chains (J-depth), and emphasized justification over opposition. - Warrant types (frequencies, percentages across 145 warrants): Descriptive 52 (36%); Evaluative 36 (25%); Empirical 24 (17%); Institutional 24 (17%); A priori 9 (6%). Column writers used more institutional and empirical warrants; high schoolers favored evaluative and descriptive warrants. Descriptive warrant was the most frequent type for both groups. - Illustrative analyses showed qualifiers can span sentences and that rebuttals can form hierarchical J-Q structures, supporting the framework's ability to capture depth and dialogic anticipation.

Findings indicate that modern Chinese argumentative essays often leave the warrant-claim linkage implicit, despite multiple grounds, aligning with a reader-responsible style that invites readers to infer reasoning. This addresses RQ1 by showing distinctive use of Toulmin components: abundant grounds, fewer explicit warrants, sparse opposition, and limited qualifiers overall. Regarding RQ2, the pronounced preference for width over depth demonstrates a strategy of piling parallel supports rather than developing extended chains, especially among expert columnists who can marshal more distinct reasons. For RQ3, the dominance of descriptive warrants suggests that Chinese writers frequently blend inference with narration/description and subjective commentary to guide readers toward claims, a practice connected to linguistic and cultural patterns (e.g., topic-prominent sentence structures and valuing moderation and indirectness). The low frequencies of opposition components should not be taken as a lack of persuasiveness in the Chinese context; instead, writers tend to reinforce claims through justification rather than explicit confrontation, and use qualifiers to moderate tone and acknowledge limited concessions. Overall, the proposed framework, integrating pragma-dialectic interaction with Toulminian structure, effectively captures these dialogic and hierarchical features in real essays.

The study proposes a Toulmin-loyal yet pragma-dialectically informed analytical framework that models argumentative essays with parallel and hierarchical structures and dialogic anticipation of opposition. Applying the framework to 60 Chinese essays reveals: (1) writers do not consistently provide explicit warrants for all grounds, fostering reader engagement; (2) preference for justification over opposition and for argument width over depth; and (3) frequent use of descriptive warrants that blend inference with description. These patterns reflect evolving Chinese rhetorical practices distinct from both ancient conventions and fully Anglophone norms. The framework serves as a descriptive tool rather than an evaluative metric and has potential applicability to other languages and cultural contexts. Future research should test and compare the framework across different contexts and languages to assess generalizability and cultural variability.

Major limitation is the lack of cross-context comparative analysis; results are based on Chinese-language essays from specific sources and time spans. Group differences (topic freedom, time constraints, writing purposes) mean observed differences (e.g., in qualifiers and warrants) cannot be attributed solely to proficiency. Warrants were identified primarily via semantic/pragmatic judgment in a paratactic language, which may introduce subjectivity despite double checking. Opposition component usage may be culturally influenced, limiting generalizability to other contexts.