Identify the production method of the sampled television captions

Identify whether the sampled U.S. broadcast television captions were generated by steno captioning, respeaking, or automatic speech recognition, in order to determine how caption-generation method contributes to the observed metric scores and viewer experience.

Background

The study compares broadcast television captions with vendor-generated ASR captions, but the production methods of the broadcast captions were not available to the researchers. Consequently, the broadcast sample may combine captions produced through steno captioning, respeaking, and ASR without allowing the results to be attributed to a specific generation method.

Determining these production methods would enable more precise comparisons among captioning technologies and would help evaluate claims about whether caption-quality metrics are neutral with respect to caption-generation source.

References

We also do not know which of the TV captions in our stimuli were generated by steno, re-speaking and ASR, respectively.

— Are Caption Metrics Broken? Latency, Deaf and Hard of Hearing User Ratings, and Bias across Technologies  (2609.11408 - Thompson et al., 10 Sep 2026) in Section 7, “Limitations”