Q13 - Columnist: Computer voice-recognition

[smiller]

Question Type:
Necessary Assumption

Stimulus Breakdown:
All of the statements in the premise and conclusion are about voice-recognition technology. We can simplify the argument by ignoring that, and focusing on what's unique in each statement.

Premise: cannot distinguish between homophones
Conclusion: cannot recognize and use grammatical and semantic relationships → will not accurately translate spoken words into text

Answer Anticipation:
Even if we look past "voice-recognition technology" and focus only on the unique ideas in each statement, this argument is still a little strange. There doesn't seem to be any clear connection between the premise and conclusion. In fact, voice-recognition technology seems to be the only concept they share. We don't know exactly how being unable to "distinguish between homophones" is related to "grammatical and semantic relations among words," or what it has to do with translating spoken words into text.

(Okay, so you happen to be a linguist who writes voice-recognition programs for a living, and the connections are crystal clear to you. Excellent! But the LSAT doesn't require that kind of specialized knowledge. The connections wouldn't be clear just based on common knowledge.)

We can use our understanding of arguments, though. The argument states that the conclusion is true as a consequence of the premise. This means that the premise provides some link between the two concepts in the conclusion, "not recognizing and using grammatical and semantic relationships" and "not accurately translating spoken words into text."

We could link all of these concepts together with two assumptions:

cannot recognize and use grammatical and semantic relationships → cannot distinguish between homophones

and

cannot distinguish between homophones → will not accurately translate spoken words into text

In fact, both of these statements are necessary in order for our premise to support the conclusion.

It might help to look at a simplified version of the argument:

Premise: A
Conclusion: B → C

Necessary Assumption 1: B → A
Necessary Assumption 2: A → C

Now that you've had your conditional logic workout for the month, let's look at answer choices.

Correct answer:
(A)

Answer choice analysis:
(A) Correct: This is the contrapositive of one of the assumptions we mentioned above. It give us "can distinguish between homophones → can recognize and use grammatical and semantic relationships."

(B) This is an Illegal negation of the conclusion.

(C) Out of scope: The entire argument is about voice-recognition technology. We don't need to assume anything about the way humans distinguish between homophones.

(D) Illegal reversal: This give us "cannot distinguish between homophones → cannot recognize and use grammatical and semantic relationships."

(E) Out of scope. The argument is about translating spoken words into written text, not checking the spelling and grammar of written text. We don't need to assume anything about the latter.

Takeaway/Pattern:
When both the premise and conclusion are about the same subject, like voice-recognition technology, we can ignore that subject momentarily to get a better sense of the gap in the argument. To handle Logical Reasoning like a machine, you need to be well-versed in conditional logic and ready to translate "unless" statements into more basic conditional notation. Even with that being said, this is a tough question. Under timed conditions, you have to make wise choices and invest your time where you need it the most.

#officialexplanation

[JoeH98]

Hi,

I approached this question differently.
~DBH (Cannot distinguish between homophones)
IRU (Improved to recognize and utilize blah blah)
AT (Accurately translate blah blah)

Premise: ~DBH

Conclusion: AT -> IRU

I think so far, my thought process was same as yours, but I equated DBH and IRU. I mean if you think about it..if it cannot distinguish between homophones such as "their" and "there," doesn't that also mean it will not accurately translate a computer user's spoken words into written text?

So the two ideas are essentially saying the same thing, so new diagram is as follows:

Premise: ~DBH

Conclusion: DBH -> IRU (or ~IRU -> ~DBH)

Therefore, the necessary assumption needed is that without IRU, ~DBH, which is what answer choice (A) states.

Is this a okay reasoning? I feel like this question was atypical since the assumption required was just showing that the necessary condition of the conditional conclusion was indeed necessary whereas other N/A question answers normally bridge the gap between premise and conclusion.

Thanks.

JH

[andrewgong01]

Can we also say that Choice "D" is too extreme? I chose Choice D on the real test but on review D is basically the flipped version of "A", the credited response. However, the argument never said grammar recognition was sufficient for for homophones ; rather it was introduced as a necessary condition but "D" makes it a sufficient condition, which makes it too extreme.

[smiller]

I think so far, my thought process was same as yours, but I equated DBH and IRU. I mean if you think about it..if it cannot distinguish between homophones such as "their" and "there," doesn't that also mean it will not accurately translate a computer user's spoken words into written text?

JoeH98, I’m totally on board with one part of your explanation. It’s probably safe to say that if a program can’t distinguish between homophones such as "their" and "there," it won’t accurately translate spoken words into written text. I described that as an assumption in the argument, one of the two assumptions that are necessary to link the premise to the conclusion. We could probably call this an inference instead of an assumption, but even then we’re not “equating” those terms.

“Equate” has a specific meaning. It means stating that two things are equal—that two terms have the exact same meaning. The relationship has to work in both ways. “Distinguishing between homophones” does not have the exact same meaning as “accurately translating spoken words into written text.” We can’t say that these two terms are equal.

What you’re describing is actually a conditional relationship:

~DBH → ~AT

Note that this doesn’t work in the opposite direction. If a computer can’t accurately translate spoken text, it doesn’t guarantee that the computer can’t distinguish between homophones. There could be another reason for its inability to translate accurately.

We also can’t equate DBH and IRU. They don’t mean exactly the same thing. That might have been a typo in your explanation, though. I think what you’re trying to do in your explanation is equate DBH and AT, allowing you to directly substitute one for the other. But as I just stated above, we can’t do that. The terms don’t mean exactly the same thing, so we can’t directly substitute one for the other.

It just so happens, in this one argument, that you can get to the correct answer by doing that. But it’s not a valid process. There’s no valid logic that allows us to do that kind of direct substitution. If you try to apply the same process to other questions, you're likely to get the wrong answer.

If we look at a complete link between the premise and conclusion, we have

~IRU → ~DBH → ~AT

As you noted, we can probably accept ~DBH → ~AT. But it is still necessary to assume ~IRU → ~DBH. In order for the premise to support the conclusion, we must assume that this particular conditional relationship exists.

[smiller]

Can we also say that Choice "D" is too extreme?

Andrewgong01, what’s important is that you’re seeing the reversed logic in choice (D). As you wrote, that answer treats a necessary condition as if it were a sufficient condition.

When we talk about an answer being too “extreme,” we're usually referring to a quantifier or modifier. For example, suppose we're looking at an Inference question, and the stimulus allows us to infer that Red Bull is "a delicious beverage." If an answer choice states that Red Bull is "the most delicious beverage ever," that would be too extreme.

It’s not wrong to call (D) “extreme” if that’s your way of thinking about the the reversed logic. Some people think about illegal reversals that way. It’s a different way of articulating why the answer is wrong. Just keep in mind that sometimes we use the word “extreme” to describe a different type of wrong answer, one that doesn’t involve a conditional logic error.

[SJK493]

Premise: A
Conclusion: B → C

Necessary Assumption 1: B → A
Necessary Assumption 2: A → C


So this is what was posted for the official explanation. I had always thought for Sufficient Assumption/Necessary Assumptions were the same that when given a premise with a conclusion that is a conditional statement.

Premise: A
Conclusion: B → C

Sufficient Assumption: A → B
Necessary Assumption: A → B

This creates A → B → C. But this does not seem to work for this question, and I suppose for Necessary Assumptions. So is the approach you outlined in the official explanation for Necessary Assumptions, and the approach that I outlined below for Sufficient Assumptions?

[MingL143]

When I read the stimulus, I thought this is a conditional argument, so I was trying to look for a missing link or some new information. But I can’t find any. Then I was trying to think of an objection to the argument, I couldn’t come up one. The correct answer is just negation of the conclusion. So how do you work on this kind of argument? Is this argument “a phenomenon – causation”?

[ohthatpatrick]

It seems like you're expecting every argument to be Conditional, Causal, or Comparative.

I don't harbor that expectation or find it super useful to try to force everything into those categories.

To me, a broader archetype at play here is
1. Here's problem X
2. Thus, until we take action A, we won't get result B.

That argument is assuming
- problem X is currently preventing us from getting result B
- action A would help us to solve problem X (or get result B)
- no other action would help us to solve problem X (or get result B)

So the correct answer could sound like
- an inability to distinguish between homophones is creating inaccurate translations of voice to text

- recognizing and utilizing grammatical/semantic relations among words would help us to distinguish between homophones (or to get accurate translations)

- doing this other thing (action C) would NOT help us to distinguish between homophones (or to get accurate translations)


If I were arguing the anti-conclusion, I'd be thinking, "How can I argue that even if we haven't started to recognize and utilize grammatical and semantic relations among words, we can still get accurate translations?"

My first thoughts there become
- maybe there's some other way besides using grammar/semantic relations that we could distinguish between homophones.

And then the correct answer rules out that objection.

I'm not sure what you mean by "the correct answer is just a negation of the conclusion". The conclusion is the final sentence, and it doesn't mention distinguishing between homophones. Meanwhile, (A) does.

It's a bridge between the evidence (the first sentence) and the conclusion (the second sentence).

Hope this helps.

[abrenza123]

I understand that D is a reversal of correct answer A, but I'm hoping for a more in depth reason why it is incorrect...


I was attracted to D because it seemed to "chain" the conditionals together, but I also knew because it didn't fully explain the relationship in the conclusion. I knew it would link not distinguishing between homophones to improved recognition but still didn't explain the link in the conclusion between recognizing semantics and accurate translation. I I had a difficult time because I wasn't sure what to do with a premise (problem X) and then two different items (action A and result B) in the conclusion. since they seemed driven by conditional logic I think I was almost making it into an SA question and trying to find a way to trigger the conclusion, if that makes sense?

[ohthatpatrick]

You should not be trying to prove the conclusion, as in Sufficient Assumption. But given the conditional elements in the argument and in the answer choices you're considering, it wouldn't be crazy to think about how the argument is trying to work, conditionally.

When you see conditional logic answers in NA, just diagram them (mentally or physically) and see if it matches a move the author was making.

CONSIDER THIS ARGUMENT:
Joey just went to 7/11. Thus, he's probably got a new pack of cigarettes.

Which of these are we assuming:
(A) If Joey has a new pack of cigs, he probably went to 7/11
or
(B) If Joey went to 7/11, he probably bought a pack of cigs
or
both?



Just (B). That's the flow of the argument, from PREM --> CONC.

The author could conversationally defend herself against (A), and say, "I wasn't assuming that! I don't think that EVERY time I see Joey with cigs he must have bought them from 7/11. I just know that every time he goes to 7/11 he buys cigs."

In Q13's argument, the flow of logic of the last sentence hinges on "until".

UNTIL / WITHOUT / UNLESS = "if not"

So that last sentence is saying
"IF voice recog. can't do grammar/semantics, THEN it won't accurately translate."

The first sentence is synonymous with "not accurately translating", because not being able to distinguish between homophones would be a lack of accurate translation.

So it makes the most sense to see it as:
Can't do gram/seman -> can't distinguish h-phones -> no accurate trans.

Thus, when we're considering (A) vs. (D), we would want to see:
"Can't do gram/seman --> can't distinguish h-phones"
or
"Can distinguish h-phones --> can do gram / seman"


(D)'s logic would look more like
If voice recognition can't distinguish homophones,
then it can't recognize grammar / semantics,
and thus it will not accurately translate spoken words

(A)'s logic looks like
If voice recognition can't recognize grammar / semantics,
then it can't distinguish homophones,
and thus it will not accurately translate spoken words

Conversational example:
Brad cannot currently get above 171 on the LSAT. As a consequence, until he aces Logic Games, he will not get a 99th percentile score.

Is that saying
"If he doesn't ace Logic Games, he can't get above a 171"
or is that saying
"If he can't get above a 171, he doesn't ace Logic Games"
?

It's saying the first one.