mistranscriptions

Mistranscriptions are transcripts that fail to faithfully reproduce the original spoken or written material. They can occur in oral-to-text transcription, captions and subtitles, optical character recognition of printed documents, and the copying of manuscripts. A mistranscription differs from intentional edits and from translations, as the aim is to represent the source exactly as it appeared or sounded.

Common forms include incorrect words or phrases, omissions, insertions, misordered sentences, mispunctuation, and failures to render diacritics or capitalization. Errors often involve homophones or letters that look similar in a given typeface, leading to substitutions such as words that do not match the intended meaning. The result is a text whose meaning may be distorted without obvious signs of manipulation.

Causes range from human error in manual transcription to limitations of automated systems. Poor audio quality, strong accents, or ambiguous handwriting can lead to mishearing. OCR and speech-to-text programs may misinterpret characters or sounds, especially with noisy sources or unusual fonts. Time pressure and lack of standardized conventions also contribute to mistranscriptions.

Impact and mitigation: Mistranscriptions can propagate misinformation, misattribute quotes, complicate legal or academic work, and undermine trust in media records. Correction often requires verification against the original source, issuance of errata, or re-transcription. Quality-control measures include independent review, careful proofreading, version tracking, and retaining source recordings for future checks.