Glossary
Code-switching
Alternating between two or more languages within a conversation or sentence; common in multilingual speech and hard for models to handle.
Speakers may switch at sentence boundaries or mid-phrase, as in Spanglish or Hinglish. It breaks monolingual assumptions: language-identification front ends misfire, lexicons miss mixed words, and systems trained one language at a time degrade sharply on mixed speech. Natural code-switching is hard to elicit in a studio, so authentic recordings are scarce.
Why it matters
Bilingual conversational archives carry a signal that scripted single-language collections cannot supply.