Abstract
When two or more non-native language speakers or learners converse, they may often encounter words or grammatical structures they do not understand. In the absence of a native speaker, learners tend to pause their conversation to look up unfamiliar words in a dictionary, causing disruption to their conversation flow. This disclosure describes techniques that assist non-native language speakers or learners during real-time conversations without disrupting conversational flow. With user permission, an ongoing conversation is analyzed in real time by a large language model (LLM) based language assistant to identify non-idiomatic speech and grammar, vocabulary, and/or pronunciation errors. When a user is detected to experience a language difficulty, e.g., appears to search for a word or phrase, asks for help, etc., the language assistant makes context-sensitive interjections. Effectively, the techniques simulate the presence of a third person in the conversation who is a native speaker and who assists the conversations between the two non-native language speakers.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Guajardo, Jaime and Sutasirisap, Sriratana (Daowz), "Real-time Context-aware Assistance for Non-native Language Speakers", Technical Disclosure Commons, (July 17, 2025)
https://www.tdcommons.org/dpubs_series/8373