Although Natural Language Processing has witnessed several breakthroughs after its inception, recent academia is witnessing a stall in making further progresses in the field. Various techniques such as Stanford Parser and Neural Networking Machine Learning mostly focus on syntactic analysis of the language. However, from linguistic perspective, language is not only about structures of words and grammatical components, but also about structures of sound. This is what’s called phonetics in linguistics. Strangely, studies on the phonetics have been considered minor compared to other parts of linguistics, such as syntax (which has been the focal points of researchers in NLP) and pragmatics. This can be contributed to the difficulty in ‘scientifically’ studying phonetics of languages, yet when we turn our attentions outside the conventional academic field of ‘linguistics’, this might be solved, probably easier than what we’ve thought.
Here, I propose musimatics, or mathmatical studies on music, to be incorporated into natural language processing in its future researches. Working musicians and music industries have already widely adopted mathematical analysis in recording and processing their music. Yet, eerily, it’s extremely hard to find research papers in phonetics which have been studied in the perspective of the mathematics and music combined.
As far as I know, Dr. M Hannumanthapa et al has finished a survey study on the impact of phonetics on NLP. In their survey paper, “Impact of Phonetics in Natural Language Processing: A Literature Survey”, Dr. M Hannumanthapa and his collegues have given a general study approach of phonetics in NLP, an overview of what studies have been done in the subfield of NLP.
But still, the study focuses on how sounds are ‘pronounced’. However, to make machines learn and interpret pronuncination of languages, we need to focus more on how languages are understood. This is the point where we incorporate Musimatics into NLP.