PrepTest 154, Section 1, Question 13
Columnist: Computer voice-recognition technology currently cannot distinguish between(38 more words not shown)
Which one of the following(8 more words not shown)
- AIn order for(19 more words not shown)
- BIf voice-recognition technology(21 more words not shown)
- CHumans can distinguish(13 more words not shown)
- DUnless voice-recognition technology(16 more words not shown)
- EComputer programs that(15 more words not shown)
Reading the argument
“I'm going over there.” The sound of the last word doesn't distinguish “there” from “their,” but the rest of the sentence helps us choose the spelling. That's the proposed solution to the computer's problem. Homophones sound alike but have different meanings and spellings. Recognizing grammatical and semantic relationships means using how words fit together and what they mean, rather than relying only on the sound in isolation.
I can see why that would help. But the columnist says accurate translation won't happen until the technology gains this ability. Is there another way to distinguish the words? I don't have a definite alternative method in mind. Still, not being able to name another solution isn't proof that the proposed ability is required.
What we’re looking for
For the required-assumption question, focus on that possible alternative. The argument needs to rule out distinguishing homophones without recognizing those relationships. Otherwise, the present homophone problem wouldn't establish that accurate translation must wait for this particular improvement. We need the relationship as a requirement, not a promise that adding it solves every problem in speech recognition.
Working the answers
- ACorrect
To distinguish homophones, the technology must recognize grammatical and semantic relationships. That supplies the dependence the columnist needs. If the technology could distinguish the words without that ability, the evidence about homophones wouldn't establish that this improvement must come first.
A doesn't prescribe a single algorithm or say this ability will be enough on its own. It says the ability is required, which is the claim the argument relies on.
- B
This guarantees accurate translation once the improvement is made. The conclusion doesn't require that guarantee. Grammar and meaning could be necessary while other problems still need work. Needing an improvement and being guaranteed success by that improvement are different claims.
- C
The conclusion concerns what computer voice-recognition technology requires. It doesn't depend on humans being able to distinguish homophones without grammatical or semantic information. Whether people have another method would be a separate question; C doesn't establish the requirement for the programs.
- D
This makes distinguishing homophones a requirement for recognizing grammatical and semantic relationships. We need the dependence the other way around: recognizing those relationships must be required for distinguishing homophones. A program could recognize some relationships among words while still struggling to choose between words that sound alike. D isn't something the argument must assume.
- E
A spelling-and-grammar checker starts with written text. The columnist is discussing how to translate spoken words into that text. We aren't told what the existing checkers can do, and the argument doesn't require them already to distinguish homophones. Their current capabilities don't supply the proposed dependency.
More Necessary Assumption questions
Learn the LSAT our way
Our platform drills Necessary Assumption questions at your level and adjusts as you improve.