This poster presents an innovative framework for modelling and visualizing translation trajectories by integrating complexity theory (Byrne 2005, Turner & Baker 2019), case-based reasoning (Richter & Weber 2006) and computational text analysis (Licht & Lind 2023) The project addresses a central methodological question in translation studies and the digital humanities: How can we represent the dynamic, context-dependent evolution of a translation process when the relevant causal factors are both qualitative and quantitative? Since translation choices occupy graded continua rather than sharply bounded binary categories, this project uses fuzzy-set modelling to make such gradience operational, explicit and reproducible. In translation, such processes have been represented at a visual level only intuitively (Figure 1).

Figure 1. Modelling translation with aerodynamics – 4D (Marais & Matthee 2019)

In addition to asking how to visually represent translation dynamics, this project also asks: What specific patterns (such as volatility, stability or attractor strength) characterize translations from different historical and ideological contexts? We hypothesize that a translator's trajectory through the multidimensional state space of choices will form a distinct 'signature,' revealing systemic priorities beyond individual lexical and syntactic decisions.

Building on Kobus Marais’s concept of translation trajectory (Marais 2018) – understood as the emergent path produced by the interaction of constraints, attractors and translator habitus – the project proposes a computationally articulated approach for comparative analysis of translations. The case study focuses on three markedly different English translations of The Analects of Confucius (Confucius 1861, 1898, 1979), produced in distinct historical, ideological and institutional contexts. Scholarly analyses of The Analects translations have often relied on either quantitative comparison of textual features or qualitative discussion of translator intent (Min 2015, Guo 2022, Qianni & Yaqin 2025). This project bridges that divide, offering a method to visualize how both dimensions interact dynamically across a full translation trajectory. Each translator’s choices are here treated as expressions of a complex adaptive system shaped by philological training, personal disposition, sociocultural constraints and intended audience.

The method proceeds in four stages.

(1) Dimensional Modelling

The first step defines a multidimensional state space combining textual-practice variables (literal vs. interpretive strategies, foreignizing vs. domesticating key terms, stylistic register, degree of commentary, use of doctrinal reframing) with contextual and habitus-related variables (academic background, ideological commitments, institutional constraints, audience expectations etc.). These dimensions are operationalised using fuzzy membership scores, allowing the integration of qualitative scholarly judgments with measurable textual features.

(2) State Classification

Using these dimensions, the project identifies nine recurrent “translation states,” such as Philological–Foreignizing, Didactic–Domesticating, Analytic–Clarificatory, Syncretic–Reframing, Constraint-Driven Simplification etc. Each chapter or passage of The Analects is assigned to a dominant state, enabling the construction of a sequence of states that reflects how the translator moves through the text.

(3) Trajectory Construction

For each translation, transitions between states are extracted across the ordered progression of the Analects. These transitions form a state-transition network – essentially a Markov-like model – that visualizes the translator’s path through the conceptual and stylistic landscape defined above. Node sizes represent state prevalence, and edge weights represent transition frequency. Differences between translators thus become differences in dynamical structure rather than only in isolated lexical choices.

(4) Interpretive Layer

The trajectories are contextualized through paratextual materials, historical documentation and biographical information. By correlating specific state shifts with documented constraints (e.g., publisher guidelines, ideological frameworks etc.) or attractors (e.g., philological training, religious commitments etc.), the method demonstrates how translation trajectories emerge from interacting qualitative forces. This provides a new way to articulate causality in translation without resorting to deterministic or universalist models. This layer directly correlates peaks in node centrality or frequent state transitions with specific paratextual evidence, such as a translator's prefaces, publisher guidelines or documented ideological commitments.

The poster shows how this approach produces visually interpretable, theoretically grounded and methodologically reproducible representations of translation processes. It highlights the contribution of complexity thinking to translation studies and demonstrates how DH techniques – NLP and ML methods, such as alternately fine and coarse-grained multilingual embedding comparison, fuzzy modelling, network analysis and visualization and comparative trajectory analysis – can make visible the dynamics between textual features and translator habitus. The method offers a generalizable framework for analysing any corpus where qualitative and quantitative factors jointly shape interpretive output, including multilingual editions, literary rewritings and community-informed translation practices.

The poster features a central visual comparing the state-transition networks of our three translators, making their divergent dynamical “signatures” immediately legible. This project’s primary contribution is a middle-range methodological framework that operably bridges high-level translation theory (complexity) and low-level textual data, enabling the systematic comparison of translational dynamics for the first time in translation studies.