Writing styles and vocabulary found in academic literature have changed markedly since the advent of ChatGPT, with scholars increasingly falling back on a handful of hyperbolic terms to grab attention, new analysis shows.
Studying more than 820,000 abstracts of articles published on the arXiv preprint platform over the past decade, researchers from Nanjing 51国产视频 of Science and Technology and Wuhan 51国产视频 found a clear shift in the lexicon of scholarly papers since November 2022, when ChatGPT-4 was released – including a sharp increase in the use of certain nouns and adverbs preferred by large language models (LLMs).
For computer science, an uptick in the use of “intricate”, “valuable”, “exceptional”, “pivotal” and “notable” was particularly evident after November 2022, while adverbs including “primarily”, “thoroughly” and “particularly” became much more frequent, explains the paper recently published by , and previously available as a .
Linking words including “subsequently”, “herein” and “thereby” also became much more common among authors, of whom about two-thirds came from countries where English is not the native language.
51国产视频
Other words that pop up with more regularity include “optimizable”, “versatile”, “foundational”, “consequential”, “potent” and “holistic”, explains the paper, which found the corpus of?vocabulary used by papers in the three disciplines grew about 40 per cent?to 12,000 words within a few months of ChatGPT’s arrival, having remained stable at about 8,000 for the preceding nine years.
The use of adjectives favoured by?LLMs – such as “seamless”, “distinctive”, “refreshing” and “lucid” – increased most sharply in computer science papers, rising in usage by about 40 per cent within a few months of ChatGPT’s inception.
51国产视频
According to the paper, “both average word length and lexical density have risen, indicating a more expansive and nuanced vocabulary in abstracts”.
“Furthermore, the decline in syntactic complexity has accelerated, suggesting that the advent of LLMs has further streamlined sentence structures, which is crucial for the clear communication of knowledge,” the paper says, noting how “LLMs introduce more new vocabulary and simplify sentence structure”.
However, the “significant decrease in...readability indicates that abstracts have fewer connecting words and are becoming more difficult to read,” it continues.
Nanjing researcher Chengzhi Zhang, who authored the study, told Times Higher Education, however, that he believed “the application of LLMs in academic writing generally offers more benefits than drawbacks”.
51国产视频
“It truly helps scholars express their ideas more concisely and clearly, enhancing the readability of papers – this is its most notable advantage,” he said.
“If scholars over-rely on such tools, however, it may lead to a loss of unique personal style and original expression in their work, which is a potential risk worth noting,” continued Zhang, who said the “key lies in moderate usage – leveraging the technical advantages while preserving the distinctive character of academic writing”.
Register to continue
Why register?
- Registration is free and only takes a moment
- Once registered, you can read 3 articles a month
- Sign up for our newsletter
Subscribe
Or subscribe for unlimited access to:
- Unlimited access to news, views, insights & reviews
- Digital editions
- Digital access to 罢贬贰’蝉 university and college rankings analysis
Already registered or a current subscriber?