Machine Translation as an Alternative to Language-Specific Dictionaries for LIWC

This research presents an alternative approach for applying the text analysis program, Linguistic Inquiry and Word Count (LIWC) to non-English text: using machine translation (MT) to translate the text into English and then apply the English LIWC dictionary, rather than applying a translated LIWC dictionary to the original texts. We evaluate the alternative method using the output from two open-source MT systems and Google Translation on a set of TED Talk subtitle data. The method is tested on three language pairs: Dutch-English, German-English and Spanish-English. Provisionally, MT seems to lead to better results in the three language pairs than language-specific dictionaries.