Investigating syntactic priming cumulative effects in MT-human interaction

Area of Science:

• Second Language Acquisition
• Computational Linguistics
• Psycholinguistics

Background:

• Investigates the impact of Google Neural Machine Translation (GNMT) on English language learners (ELLs).
• Examines the acquisition of challenging English syntactic structures that differ from learners' native languages.

Purpose of the Study:

• To determine if interaction with GNMT influences L2 English learners' production of a specific syntactic structure.
• To assess syntactic priming and cumulative effects from continuous exposure to GNMT output.

Main Methods:

• Tested 30 Brazilian Portuguese L2 English speakers on image descriptions.
• Analyzed the use of prepositional noun phrases versus GNMT-influenced structures for possession.
• Measured adaptation to L2 structures through continuous exposure to Google Translate output.

Main Results:

• Demonstrated a robust syntactic priming effect in L2 English learners.
• Observed a significant cumulative effect, indicating adaptation over time.

Conclusions:

• GNMT output can significantly influence the linguistic behavior of L2 English learners.
• ELLs unconsciously acquire and adapt to new syntactic structures through sustained interaction with GNMT.