Network motifs for translator stylometry identification

Abstract

Despite the extensive literature investigating stylometry analysis in authorship attribution research, translator stylometry is an understudied research area. The identification of translator stylometry contributes to many fields including education, intellectual property rights and forensic linguistics. In a two stage process, this paper first evaluates the use of existing lexical measures for the translator stylometry problem. Similar to previous research we found that using vocabulary richness in its traditional form as it has been used in the literature could not identify translator stylometry. This encouraged us to design an approach with the aim of identifying the distinctive patterns of a translator by employing network-motifs. Networks motifs are small sub-graphs which aim at capturing the local structure of a complex network. The proposed approach achieved an average accuracy of 83% in three-way classification. These results demonstrate that classic tools based on lexical features can be used for identifying translator stylometry if they get augmented with appropriate non-parametric scaling. Moreover, the use of complex network analysis and network motifs mining provided made it possible to design features that can solve translator stylometry analysis problems.

On Explainable Features for Translatorship Attribution: Unveiling the Translator’s Style With Causality

“…The order of the experiments is the following: 1) Subsection IV-A presents the results of the experiments performed on the Don Quixote corpus. We contrast them with the results in [21]. 2) Then, in Subsection IV-B, we present the results for the experiments on the Ibsen parallel corpus (i.e., the two translations of the play Ghosts) and compare them with the results on the same corpus of [20].…”

“…The two most recent works, which are also closer to what we attempt here, are [20] and [21]. The authors of both papers use the methodology suggested by Baker [5] of using parallel corpora-same source text translated by different people into the same target language.…”

“…It is worth mentioning in order to contrast their results against ours, that, as part of their data preprocessing, they cleaned and lemmatized the text leaving only alphanumeric characters (thus removing punctuation). Contrasting with [20] and even though they did use a feature selection technique by ranking, El-Fiqi et al [21] did not present the network motifs most distinctive for each translator nor provide an stylistic interpretation of their results.…”

“…In the second work, El-Fiqi et al [21] also used two corpora: the last six (out of 30) parts of seven translations into English from Arabic of the Quran and three translations into English from Spanish of the two parts of the novel Don Quixote by Miguel de Cervantes. After obtaining negative results using the methods in [14], they use network motifs (repeated subgraphs on the word adjacency network), which, they argue, capture some syntactic patterns in the writing of each translator.…”

“…We experimented with one corpus from each of the two works reviewed in Subsection II-B: the Ibsen corpus from [20] and the Don Quixote corpus from [21]. We downloaded the Ibsen plays (translations of theater plays from Norwegian into English) from the Project Gutenberg 3 website.…”