Checking the checkers

Tihs blog was created to help me keep track of the materials for a thesis in linguistics that I'm working on. The essay is about figuring out a truth-scale somewhat like the one used by fact-checkers such as politifact, but that is based in a semantic/pragmatic perspective on language and truth. If you've managed to find this blog you're very welcome to get in touch with me with ideas, thoughts, critique or questions of any kind.

[yeli-renrong]

Klingon isn’t that phonologically weird – it just shows signs of a substrate. 

Presumably, Old Klingon had a two-way stop MOA contrast between unvoiced and implosive consonants, and the implosives were lost in various ways due to substratal influence: some dialects apparently have prenasalized stops or nasals, but in the standard these are voiced stops. The development of a voiced retroflex plosive from an alveolar plosive is known from Somali.

(It’s not unheard of for languages with a contrast between /t/ and /d/ to realize these two plosives at different MOAs. I don’t know how I’d go about finding the paper on this in [IIRC] Austronesian, but I know I saw one once.)

It’s also possible that the substrate language had a dental/retroflex contrast of the sort common in India. This would explain why the only sibilant is retroflex. But retroflex symbols are commonly used for apical postalveolars, and an apical (post)alveolar as the only sibilant is known from Icelandic and Greek.

The contrast between /qʰ/ and /q͡χ/ presents difficulties, but this may be an artifact of the mapping between Klingon phonemes to IPA symbols – it’s possible that a ‘velar’ consonant in Klingon sounds to humans like a postvelar (‘uvular’). (Is the soft palate smaller in Klingons than in humans?) This would, of course, create pressure on the true uvular to develop phonetic affrication.

Really the weird thing about Klingon is that it contrasts /v/ and /w/ but doesn’t have /f/.

[yeli-renrong]

Maybe that’s not so weird – Old Klingon *b *d *g > v r ɣ?

So, Old Klingon:

p t č k q ʔ b d ǯ g ɓ ɗ ƛ s x m n ŋ w l j

Which isn’t unreasonable. The laterals could maybe be paired with the stops; the voiced stops presumably had fricative/approximant allophones, given their outcomes.

Then again, unconditional lenition of voiced stops seems improbable. Preceding nasals blocked lenition of voiced stops in Greek, right?

[danbensen]

I propose the substrate was the language of the pre-contact pre-genetic engineering Klingons.

[linguistifacts]

Reblogging because I want to remember to look into this

[linguistifacts]

PolitiFact too hard on Walker?

Today I’ll investigate a case where I think that PolitiFact may be a little too harsh in their text. The case in point is this: http://www.politifact.com/wisconsin/statements/2017/jan/13/scott-walker/scott-walker-says-wisconsin-property-taxes-percent/

Walker said Wisconsin property taxes, as a percentage of personal income, "are the lowest that they've been since the end of World War II."

It’s worth noting that one change Walker made reduces income tax refunds given to lower-income taxpayers through the state Homestead Credit, which is aimed at softening the impact of property taxes.

But an analysis by the respected Wisconsin Taxpayers Alliance shows that on the measure Walker cited, property taxes are the lowest they have been since 1946.

We rate the statement True.

Even though the statement is rated true, the caveat about Walker’s politics makes it seem like PolitiFact still judges the claim as somewhat misleading. And even though that caveat may be of importance to public opinions about Walker and his policies, it is not necessarily relevant to the level of this income tax, in this case.

Of course, relevance is a somewhat vague term, and can be hard to evaluate in a general way. But the truth of the actual proposition, that the income taxes are as low now as they were in 1945, is not changed in any way by the fact Walker hasn’t changed the tax refunds for some citizens. This might be interesting from a political point of view, but not for judging truth, in this case.

The only way this would be relevant is if PolitiFact again assumes that something slightly different than what is said is the “point” of the utterance. In this case, it could be something like “the property taxes are as low as they can possibly be”. It is not absurd to assume that Walker probably wants to invoke a state where people believe something along that line, but it’s not what he’s saying, and it’s not a necessary, calculable implication from his utterance. So even though this could be contained in the utterance in some way, it would be too weak to let that affect the judgement of it. Perhaps this could be analysed as a class of weak, probably intention-based meanings from an utterance, but I’m not sure it would be strong or calculable enough for that to be fruitful.

So “true” seems to be the most appropriate rating here. Since nothing contextually relevant that would change the claim is missing.

[linguistifacts]

How strongly does Trump hint?

This example is a case where speaker and context most definitely are of great importance. The case and ruling is the following: http://www.politifact.com/truth-o-meter/statements/2017/feb/16/donald-trump/trump-hints-stock-market-record-high-linked-presid/

Trump said, "The stock market has hit record numbers, as you know. And there has been a tremendous surge of optimism in the business world."

The three major stock indexes, Dow, S&P 500 and Nasdaq all closed at record highs for five consecutive days. While investors are optimistic about Trump’s plans to cut taxes and eliminate regulations, experts say other factors play influential roles in the stock market. After rising in December, consumer confidence dropped in January, though consumers are confident the economy will expand in coming months.

Trump’s statement is accurate but needs clarification or additional information. We rate it Mostly True.

As PolitiFact notes, this was said as Trump “lauded his first month in office”. 

The actual proposition, what is literally said here does seem to be true, from the text that PolitiFact provides us with. What does not seem to be true, at least not necessarily, is that this is connected to Trump or his presidency. But this is never explicitly claimed in the utterance. I’d say that PolitiFact actually provides too little context for us to decide how strongly the claim is connected to his politics. It certainly could be, but in this quote there really is no element that implies this. However, since it is said when Trump talks about the first month of his presidency, it is likely that the two are connected. But this would be purely from context, not from implications contained in the utterance in some way, which often is the case otherwise. So I’d say that this connection is quite weak, and if it’s strong enough to make the statement contain some falsity, is up for discussion.

[linguistifacts]

Condemned for Whom?

Today I’m taking a closer look at a claim from Kathleen Vinehout (http://www.politifact.com/wisconsin/statements/2017/mar/03/kathleen-vinehout/are-kids-rural-wisconsin-crossing-condemned-bridge/), rated half true by Politifact:

Vinehout says Wisconsin children are crossing "condemned bridges in the middle of the winter to get to the school bus."

Two bridges in Vinehout’s home county of Buffalo have been "condemned" in that they are too badly damaged for any vehicle traffic. But officials have determined that they remain safe for pedestrians, and children do use one of the bridges to get to and from their school bus.

For a statement that is partially accurate but leaves out important details, our rating is Half True.

This is interesting because the appropriate rating depends on how you connect the sentence, basically. It is true that there are condemned bridges, and it is also true that children are crossing the same bridges as “condemned” refers to in order to get to school. But it is not true that these bridges are condemned for the children. They are condemned for cars, but deemed safe for pedestrians. But when put this way, it certainly sounds like they are condemned in the relevant context, that is, for children to walk on.

So, while technically true, this utterance is surely misleading. I’d argue that you could claim that it’s due to the maxim of relevance (or maybe manner). In order for the information that the bridges are condemned to be relevant in this context, they must be condemned in a way that is relevant to children passing them, thus implying that they’re condemned for the walking children. So this illustrates a technically true sentence that carries false and misleading implications.

With this in mind, half true may be an appropriate rating. However, the definition for this is that the statement “is partially accurate but leaves out important details or takes things out of context “. I’d say that its truth depends on the definition of “condemned”, but it definitely could be argued that it’s completely true, not just partially.

Perhaps this statement is best described as having two different, both possible, interpretations: one that renders it true, and one that renders it false. Assuming that Vinehout adheres to the maxim of relevance, the second reading seems more likely, but the first one is equally possible, I’d say. Perhaps this example can be seen as indicating that a more sophisticated and descriptive scale is needed, one where this can be described in a good way?

[linguistifacts]

HB2′s Hurting N.C. by “One-tenth of One Percent of Annual GDP” - True, but More than it Sounds

Moving on to a statement from Dan Forest that’s been classified as simply “True”.  The whole thing is at http://www.politifact.com/north-carolina/statements/2017/mar/07/dan-forest/hb2-has-cost-north-carolina-hundreds-millions-doll/, and it amounts to this:

Forest said HB2’s economic harm to North Carolina has amounted to only “one-tenth of 1 percent of our annual GDP.”

He’s right.

While $500 million in economic losses (not to mention at least 1,400 jobs lost) is not insignificant, the state’s overall economy is large enough that the losses are only about 0.1 percent of the total GDP.

We rate this claim True

Though I agree with the judgement that this is true, there’s an interesting aspect of it that’s highlighted by the title that politifact has chosen for the piece: “HB2 has cost North Carolina hundreds of millions of dollars. Dan Forest says that isn't very much“. Though it is true that that amount of money is that small a percentage of the GDP, to put it the way Forest does makes it sound like it’s not that much money. When we hear very tiny percentages like this we seem to draw the conclusion that that money does not amount to anything important,  so this could be seen as an implication in this case. How we would judge that implication depends on what we compare it to, but most people would probably agree that $500 millions is a lot of money. Had Forest said this amount instead of a percentage, the effect of the utterance would probably be different.

You could argue that this is enough to claim that this utterance is devoid of context that would change the interpretation of it, thus making it mostly true instead of wholly true. This is also implied by politifact’s title, even though this isn’t reflected in their ruling. All in all, I think this utterance is a good example of how fuzzy these things can be. It is true, but there are grounds for rating it as mostly true, probably depending on how strong you judge this implication to be.

[linguistifacts]

Michelle Obama “not Proud of Her Country” Before Her Husband’s Election - no Grain of Truth?

This is an example of an utterance that politifact has rated as false, thus being “not accurate”. You can find the whole thing on http://www.politifact.com/wisconsin/statements/2017/feb/23/david-clarke-jr/david-clarke-says-michelle-obama-said-she-was-prou/, but let’s look at the sum up:

“Clarke says Obama "said she was never proud of her country til they elected her husband POTUS."

He provided us and we could find no evidence of such a statement.

What Obama said -- nine months before her husband was elected president -- was that for the first time in her adult life, she was proud of her country not just because Barack Obama had done well in pursuing the presidential nomination, but also "because I think people are hungry for change."

We rate Clarke’s statement False.”

I’d argue that there’s a little bit of truth in this, given that Obama actually said that she was proud of “her country” for the first time during Barack’s election campain, thus events very closely connected to his election. Though this is not completely the same thing, or at the same time, as him actually being elected, but for my intuition, I’d say it’s close enough for this to be considered mostly false. Granted, the literal statement is indeed false, there are things that are extremely similar to this statement that are true. It is reasonable to expect that Clarke’s purpose with saying this was that Obama did not feel patriotic pride until it made her husband a big shot (at least this is what he seems to imply). And this does seem to have at least some truth to it, at least it’s possible to glean from the utterance mentioned in politifact’s ruling (though she herself would probably not agree with putting it in those words). Then this utterance would be more at home in the mostly false-category

The clever reader might object that this can be seen as the same kind of excuse that I made in the media/congress approval rate-post, but the other way around. There I may have seemed a bit nitpicky, while basically accusing politifact to be overly nitpicky in this post. That objection would indeed be correct (I do think that there’s more reason to interpret things a little looser here than there though). But a point of this is to show that politifact’s rulings seem to be made inconsequently. Part of my whole point is that it seems strange that this utterance is judged so harshly, while Trump’s comparison less so, even though there’s similar excuses or accusations to be directed at both of the examples.

[linguistifacts]

“Media has a Lower Approval Rate than Congress” - Just Mostly False?

Ok, let’s start off with one of my favourite examples of weird ratings yet. The case in point is this evaluation of a Trump utterance: http://www.politifact.com/truth-o-meter/statements/2017/feb/16/donald-trump/news-media-less-trusted-congress/

Politifact’s ruling goes as follows:

“Trump said he thought that the media had lower favorability ratings than Congress. But Congress actually ranks below the news media, according to surveys from three different research groups spanning several years. In two polls, mistrust in the media broke 40 percent, which is hardly anything to brag about. But in those studies, mistrust in Congress was over 50 percent.

Trump had a point that the media has a trust issue, but he got the ranking wrong in terms of Congress.

We rate this claim Mostly False”

So the interesting thing is that the actual claim is false - media by no means has lower approval rate than congress, it actually has about 10 percentage points higher, so it’s not even that close - yet it’s not judged as completely false. Politifact’s criteria for “mostly false” is “The statement contains an element of truth but ignores critical facts that would give a different impression”. Yet there is no element of truth here. The proposition made in this particular utterance is a comparison, which doesn't hold. 

So why is it just mostly false? One explanation is that politifact might have seen this as some kind of hyperbole, so that in some way, the point of the utterance would be that media has very low approval ratings. In this case the comparison wouldn’t be the central part or even the point, so then the point of the sentence would be true - 40% mistrust is not too good. This would hold even though the literal interpretation is false.

In my view, this might be too kind if an analysis. I don’t see why we shouldn’t interpret this literally - it doesn’t obviously flout any maxims, since the literal statement is a reasonable enough statement to make, and as such shouldn’t give rise to any implicatures. If this was meant to be a hyperbole it’s nowhere near outrageous enough, if you ask me. Even if the utterance is meant literally, but a “side-point” of it being that media has low favorability, that’s actually not what is proposed by this particular utterance. It might be said in other utterances in the same context, and they might be true, but this utterance is not.

[linguistifacts]

Greetings and welcome!

I am a linguistics student currently writing a thesis for my one-year masters degree. I will mainly use this blog to organise and tag my material, which will consist of truth-value judgements from (mainly) politifact.org. 

The purpose of the thesis is to investigate the kind of truth-scale that is used by fact-checkers such as politifact, how these are handled by actual people, which parameters that seem to be vital for the classification, and if a linguistic theory can be sketched around all of this.

If you’ve made your way here for whatever reason, you’re very welcome to get in tough with me about these things, or basically anything else for that matter. Just comment on the post in question, message me or submit an ask.

(If anything weird turns up here, like stupid memes or pictures of Jeff Goldblum and/or aliens, it’s probably because I meant to post them to my personal blog over at mstealady.tumblr.com (if you like those things check it out, otherwise don’t) and tumblr mobile let me down. I apologize for this in advance)