Assessing Cultural Differences in Translations: A Semantic Network Analysis of the Universal Declaration of Human Rights

Abstract

Intercultural communication, in many cases, is cross-lingual communication. Effective cross-lingual communication requires successful translation processes. Translation quality involves two factors, the technical and the linguistic. Focusing on the influence of language factor, this study demonstrates the application of semantic network analysis and spatial modeling to examine translation equivalence. The examined texts are seven different linguistic versions of the Universal Declaration of Human Rights (six official languages and Korean). The results suggest that translations are roughly equivalent but with subtle differences reflective of each language's cultural predispositions. The paper concludes by discussing the importance of translation and language issues for intercultural communication.

Keywords:

• Translation
• Cultural Difference
• Semantic Network Analysis
• Spatial Modeling
• Language

Notes

1. “The Documentation Division comprises the Translation Services for the six official languages of the United Nations; the Editorial, Terminology and Reference Service; the German Translation Section; and the Contractual Translation Unit” (Department for General Assembly and Conference Management, n.d.). For more information, refer to the department's official website: http://www.un.org/Depts/DGACM/functions.html

2. See article about Carolyn Riding (Anonymous, 2003) and Samia Montasser's (2003) note.

3. Quadratic Assignment Procedure (QAP) is used to examine the similarity between two matrices through measuring correlation of corresponding cells. The advantage of QAP analysis is that it enables direct tests of equivalence of two relational entities, retaining dyadic value of each cell, which is the product of row and column interdependence. The analysis tests the null hypothesis that two networks are uncorrelated. By a permutation procedure, referred to as the quadratic assignment procedure, one can determine the distribution of all possible correlations given the structures of the two matrices. See Krackhardt (1987).

4. Eigenvector centrality is a more sophisticated version of the degree centrality (Freeman, 1979). While degree centrality considers only the number of connections or the sum of the strengths of the links a node has, eigenvector centrality acknowledges that the quality of connections should be regarded in addition to the quantity of connections (Newman, n.d.). The connection to the node which has higher centrality is given more weight in eigenvector centrality.

5. There were 21 comparisons because there were seven different languages in this study. Thus, all possible combinations is equal to (N(N-1)/2) where N = 7.

6. The coordinate matrices, rotated spaces, and the distance of each concept between the two datasets are not presented in this paper. However, they are available from the authors upon request.

7. Adding 16% accounted for by the third dimension, 93.89% of variance was accounted for by the first three dimensions. Although it was not explicated in the article, the third dimension differentiates Arabic and Spanish.

Additional information

Notes on contributors

Kyounghee Kwon

Kyounghee Kwon is a Doctoral Student in the Department of Communication, State University of New York at Buffalo

George A. Barnett

George A. Barnett is a Professor in the Department of Communication, State University of New York at Buffalo

Hao Chen

Hao Chen is affiliated with the Department of Communication, State University of New York at Buffalo