Co-occurrence network analysis of translation: a case study of Jabberwocky

Queiroz, João

doi:10.5281/zenodo.14532428

Published December 19, 2024 | Version v1

Preprint Open

Co-occurrence network analysis of translation: a case study of Jabberwocky

Queiroz, João¹

1. Federal University of Juiz de Fora

This work introduces a new framework for the quantitative and comparative analysis of literary translation phenomena using graph attributes derived from SpeechGraphs software. SpeechGraphs (SG) takes text as input and produces graph features as output. In the context of natural languages and word-level networks, SG belongs to the category of "co-occurrence graphs," which model co-occurrence patterns between successive words. SG was initially developed to help diagnose schizophrenia and bipolar disorder, achieving success through graph analysis. Although numerous mathematical quantification approaches have emerged in recent years, graph analysis has yet to be explored in Translation Studies. Graph attributes are employed to visually and statistically reveal hidden patterns within the word-level co-occurrence language network. This approach is particularly effective for identifying patterns that are difficult to detect without digital enhancements. We chose Lewis Carroll’s widely translated nonsense poem, Jabberwocky, as the focus of our case study.

Notes

The analysis of Jabberwocky and its translations using graph attributes from SpeechGraphs software might offer impactful methodological implications for Translation Studies. By analyzing graph attributes such as Hub Degree, Average Total Degree (ATD), and the number of edges, researchers can objectively model the structural similarities between the original text and its multiple translations. This quantification offers a rigorous empirical framework for analyzing how a translation reflects the structural properties of the source text. Secondly, the analysis enables the identification of linguistic patterns across translations. For instance, the dominance of conjunctions as hub nodes in some translations compared to nouns in others highlights distinct stylistic, semiotic, or metasemiotic choices. Recognizing these patterns provides valuable insight into the linguistic and cultural strategies underlying the translation of complex texts like Jabberwocky. Additionally, this method enables the identification of diverse translation strategies. Variations in attributes such as the WC/Nodes Ratio and Hub Degree can reveal distinct methodological approaches. For instance, a higher WC/Nodes Ratio and Hub Degree in the source text indicate a dense, repetitive structure, whereas lower values in a translation may reflect an effort toward simplification or clarification. These insights into translation strategies provide a deeper understanding of the translator’s approach and intent, enriching the evaluation process.

This approach, which we outline here only in an introductory manner, holds particular relevance for Translation Studies as a new alternative, bridging quantitative analysis with traditional qualitative methods. By providing objective metrics to evaluate structural features, it complements existing frameworks and opens new pathways for exploring equivalence across translations. Moreover, its applicability to a wide range of texts and languages underscores its potential as a versatile tool for advancing theoretical and practical insights in the field.

Files