NLP4NLP+5: The Deep (R)evolution in Speech and Language Processing

Area of Science:

• Computer Science
• Computational Linguistics
• Artificial Intelligence

Background:

• The study analyzes the evolution of speech and natural language processing (NLP) research.
• It builds upon previous analyses of the NLP4NLP corpus, extending the timeframe to 55 years (1965-2020).
• The extended corpus (NLP4NLP+5) includes nearly 90,000 documents, 67,000 authors, and 590,000 references.

Purpose of the Study:

• To analyze changes in speech and NLP research from 2016-2020.
• To identify shifts in research topics and author demographics.
• To compare recent trends with historical data and general scientific literature.

Main Methods:

• Analysis of the extended NLP4NLP+5 corpus (55 years of publications).
• Quantitative analysis of document volume, author numbers, references, and word count.
• Comparative analysis across different publication sources and with general scientific literature.

Main Results:

• A substantial increase in publications, with 2020 matching the output of 1965-1989.
• Significant growth in authors, references, and total words analyzed.
• Profound changes in research topics, with a notable rise in artificial intelligence, neural networks, machine learning, and word embedding.

Conclusions:

• The fields of speech and NLP have undergone rapid transformation in recent years.
• A new generation of researchers is emerging, focusing on AI-driven topics.
• The landscape of NLP research is rapidly evolving, marked by new publication trends.