Assessment and Evaluation Flashcards
7 cards from real CART practice questions. Tap to flip, then mark Knew It or Still Learning โ missed cards come back until you master them.
Read the first 7 Assessment and Evaluation flashcards as text
Which four elements did the FCC's 2014 caption quality order identify for television captions?
Answer: Accuracy, synchronicity, completeness, placement
The FCC caption quality standards cover accuracy, synchronicity, completeness, and placement.
In caption quality assessment, 'synchronicity' refers to what?
Answer: Captions coinciding with the spoken audio as closely as possible
Synchronicity measures how closely captions track the timing of the spoken content.
Which measurement model scores live captions by weighting errors by severity rather than simply counting wrong words?
Answer: The NER model
The NER model rates edition and recognition errors as minor, standard, or serious based on their impact on meaning.
A major limitation of Word Error Rate (WER) for evaluating CART output is that it:
Answer: Treats all errors equally regardless of impact on meaning
WER counts substitutions, deletions, and insertions equally, so a meaning-changing error weighs the same as a trivial one.
What does 'latency' measure in a CART session?
Answer: Delay between the spoken word and its appearance on the display
Latency is the lag between speech and the displayed text.
Which is the most objective way for a CART provider to self-assess realtime accuracy?
Answer: Compare the unedited realtime file against an accurate verbatim transcript
Comparing raw output to a verified transcript yields a measurable error count.
In a completeness evaluation, which omission is most significant?
Answer: A dropped 'not' that reverses the sentence's meaning
Dropping a negation changes the message and is a serious meaning error.