You are currently browsing the category archive for the ‘tests’ category.
I’d presumed it’s trivial to show that good grammar can improve your chances of success — not that good grammar is an indication of ability, but merely that having good grammar skills lends an appearance of credibility and competence that may or may not be backed up with actual skills for the task at hand. I strongly suspect, for instance, that a resume written in accordance with the basic rules of English grammar will be more likely to bring its writer an interview, all else being equal. Rather like legacy status in an application to an Ivy League school — except with an at-least-tenuous link to ability — I’ve imagined it serves as a little bonus.*
But having recently seen a few ham-handed attempts at this yield results approximately as convincing as a child’s insistence that their imaginary friend was the one who knocked over the vase, I’m beginning to re-think my presumption.
For instance, I’ve recently found this terrible post and infographic from Grammarly that purports to show that — well, it’s a little hard to say, because they’ve managed to write 500-some words without ever having a clear thesis. The infographic reports the grammatical error rates for three pairs of competing companies, and juxtaposes this with corporate data on the three pairs, presumably to look for correlations between the two.
I believe their claim is that fewer grammar mistakes are made by more successful companies. That’s a pretty weak claim, seeing as it doesn’t even require causation. We’d see this pattern if greater success led to improved grammar, perhaps by having money to hire editors; we’d see it if better grammar increased the company’s performance; we’d see it if the two were caused by an unobserved third variable. That said, the study won’t even find evidence for this tepid claim, and perhaps that is why they carefully fail to make the claim explicit.
The post tells the reader that “major errors undermine the brand’s credibility” and that investors “may judge” them for it, but even these weak statements are watered down by the concluding paragraphs. This restraint from overstating their case is hardly laudable; it’s clear that the reader is intended to look at these numbers and colors, this subtle wrinkled-paper background on the infographic, and draw the conclusion that Grammarly has stopped short of: you need a (i.e., their) grammar checker or you will lose market share!**
It might not seem worth bothering with a breakdown of the bad science going on in this infographic. Alas, the results were uncritically echoed in a Forbes blog post, and the conclusions were only strengthened in the re-telling. So let’s look at exactly why this analysis fails to establish anything more than that people will see proof of their position in any inconclusive data.
Let’s start by looking at the data underpinning the experiment. The company took 400 (!) words from the most recent LinkedIn postings (!) of three (!) pairs (!) of competing multinational corporations. We’re not even looking at the equivalent of a single college admission essay from each company, in an age where companies are producing more publicly consumable text than ever before.
Not to mention, I looked at the LinkedIn posts from Coke, one of the companies tested. Nine of their last ten posts were, in their entirety: “The Coca-Cola Company is hiring: [position] in [location]”. The tenth was “Coke Studio makes stars out of singers in India [link]”. How do you assess grammaticality from such data?
Well, let’s suppose the data is appropriate and see what results we get from it. Remember: the hypothesis is that lower error rates are correlated with higher corporate success (e.g., market share, revenue). Do we see that in the head-to-head comparisons?
- The first comparison is between Coke and Pepsi. Pepsi has more errors than Coke, and, fitting the hypothesis, Coke has a higher market share! But Pepsi has higher revenues, as the infographic notes (and then dismisses because it doesn’t fit the narrative). So we start with inconclusive data.
- The second comparison is between Google and Facebook. Google makes fewer errors and has higher corporate success. Let’s take this one at face value: evidence in favor.
- The third comparison is between Ford and GM. Ford makes fewer errors but is worse on every financial metric than GM. “However, these numbers are close”, the infographic contends. Evidence against.
So we have three comparisons. In one, which company is more successful is ambiguous. The two “decisive” comparisons are split. The data is literally equal in favor and in opposition to the conclusion. It is insulting that anyone could present such an argument and ask someone to believe it. If a student handed this in as an assignment, I would fail them without hesitation.***
What’s richest about this to me is that the central conceit of this study is that potential consumers will judge poor grammar skills as indicative of poor capability as a company. I’ve never found convincing evidence that bad grammar is actually indicative of poor ability outside of writing; the construction crew that put together my house probably don’t know when whom can be used, but my house is a lot more stable than it would be if Lynne Truss and I were the ones cobbling it together. But for all those people out there saying that good grammar is indicative of good logic, this clearly runs counter to that claim. Grammarly’s showing itself incapable of making an reasoned argument or marshalling evidence to support a claim, yet their grammar is fine. How are poor logic skills not a more damning inability than poor grammar skills, especially when “poor grammar” often means mistakenly writing between you and I?
The Kyle Wienses out there will cluck their tongues and think “I would never hire someone with bad grammar”, without even thinking that they’ve unquestioningly swallowed far worse logic. Sure enough, the Forbes post generated exactly the comments you’d expect:
“I figuratively cringe whenever grammar worthy of decayed shower scum invades my reading; it makes you wonder just how careful the company is of other corporate aspects (oh, gee, I don’t know, say, quality as well)”
With comments like that, maybe these people are getting the company that best reflects them: superficial and supercilious, concerned more with window-dressing to appear intelligent than with actually behaving intelligently.
*: I, of course, don’t mean that being obsessive about different than or something is relevant, but rather higher-level things like subject-verb agreement or checking sentence structures.
**: Though Grammarly makes an automated grammar checker, it wasn’t used to assemble this data. Nor was it run on this data, so we don’t know if it would even provide a solution to help out these grammatically deficient brands.
***: I don’t mean to imply that this would be convincing if only the data were better and all three comparisons went the right way. There’s no statistical analysis, not even a whiff of it, and there’s no way you could convince me of any conclusion from this experiment as currently devised. But at least if the comparisons went the right way, I could understand jumping the gun and saying you’ve found evidence. As it is, it’s imagining a gun just to try to jump it.
Remember when Lynne Truss’s Eats, Shoots and Leaves was the big thing? Surely you remember the heady rush when our society realized it was alright to publicly shame someone for their grammatical, punctuative, or spelling errors because a humourously mean British woman said it was, right? I sure do, because it was this blossoming of societal unpleasantness that definitively kicked me off the rolls of peevers and into my current role as a shamer of the shamers.
If there was anything new about Truss’s book, it was the philosophical stance of Zero Tolerance toward errors. Sure, previous writers had been intolerant; reading through Bierce’s Write it Right or Vizetelly’s Handbook or Partridge’s Usage and Abusage will provide ample examples of small errors treated as signs of complete illiteracy. But Truss’s Zero Tolerance policy took off among non-professionals in a way that these previous books hadn’t.
It’s been eight years since Truss’s book hit the scene, and while it’s no longer as prominent as it once was, the Zero Tolerance philosophy remains influential. Witness Kyle Wiens’s post on the Harvard Business Review’s blog from earlier this week. Wiens has started his own company, where he demands that any potential employee pass a grammar test before being hired, regardless of whether the position involves any substantial writing component.
His argument isn’t without merit. Basically, Wiens argues that attention to grammar is an indication of attention to detail in general. Of course, it’s a noisy indicator — especially when he’s hiring programmers, I imagine — but is it any noisier than the fashion-based or etiquette-based decisions that we already expect employers to use in their hiring decisions? If we tolerate employers using whether our shoes are shined or whether we hold the handshake appropriately long as indicators of future job performance, then surely there’s nothing strange about them using our grammatical competence. At least grammar shows up in every interaction, face-to-face or electronic. So if I may damn with faint praise, a grammar test probably isn’t worse than most of the other assessment methods employers use.
Thus I’m not going to condemn his use of a grammar test, but rather his method of using it: he’s an adherent to Truss’s Zero Tolerance approach.
Zero Tolerance might be a valid enforcement approach to matters like murder, where the delineation between “okiedokie” and “not okiedokie” is obvious.* But grammar simply isn’t one of those things, or at least it isn’t when you’re talking about what most people mean when they refer to “grammar”. I think we can all agree that The CEO are mistaken is wrong, but no native speaker is going to say that’s okay. Instead, what Wiens appears to be concerned with is pretty much just spelling, as Geoff Pullum notes. That’s fairly settled if you assume that all test-takers use Standard American English spellings (so no favourite, cancelled, etc.).**
But Wiens undermines his own intolerance in his post, where he uses some “debatable” constructions and includes links on each of them to justify their use. They’re things that any reasonable person ought to know are fine, like starting a sentence with a conjunction or ending a clause with a preposition.
Good for him, I say, but Zero Tolerance doesn’t accept explanations for deviations from its norm. That’s kind of the definition of Zero Tolerance: when confronting a possible error, don’t seek out explanations or rationales, just mark it wrong. There is no excuse that can justify deviation from the norm. Anything less than that is playing fast and loose with the term “Zero Tolerance”. And if we’re doing that, then I’m Zero Tolerance, too, in that I only accept usages that are standard or that can be reasonably justified as a dialectal difference or a reasonable/useful extension of current norms.
A real Zero Tolerancer wouldn’t be interested in the facts that Wiens marshals in favor of his choice; everything is black-and-white. If questions about split infinitives or final-prepositions were on such a test, Wiens would fail. It doesn’t matter that he’s right, he’s justified, and he’s seeking out relevant information to explain his decision-making. These all sound like good qualities for an employee, yet Wiens would be, to the Zero Tolerancer, inattentive and unemployable. Quite simply, Wiens is aware of a grey area even as he’s arguing for a black-and-white view.
Lastly, though it’s downright hackneyed to point out when a Zero Tolerancer makes a mistake, it is at the same time essential. As Dan of Our Bold Hero notes, Wiens failed to put a hyphen in the compound verb grammar test, and he falls into the same unhyphenated trap as Truss did by not hyphenating zero tolerance as a prenominal adjective. Were we Zero Tolerancers, his post would already be in the dustbin.
*: Even this isn’t clear enough, as evidenced by the distinction between murder and manslaughter and the various levels of each. As you might have guessed, I don’t believe in Zero Tolerance for anything.
**: Though in both of these cases, typos and thinkos still happen, and Zero Tolerance is unwilling to forgive this. As a result, employees of a philosophically-committed-to-ZT company will have to waste a lot of time proofreading even the quickest correspondence to make sure that not a single mistake makes it through.
It looks like CNN credulously spit out another story from Global Language Monitor (GLM). Basically, GLM did their usual thing of running a speech (in this case, Obama’s oil spill speech from mid-June) through some mindless statistics, getting out the Flesch-Kincaid Grade Level, and then reporting it as though it was actually meaningful analysis. Language Log and Johnson already explained why the GLM analysis is nonsense, and as a result, CNN actually substantially re-wrote the story.
I discussed the meaninglessness of grade level analysis a year and a half ago in more depth, but this time let me just offer an illustration of why grade-level analysis is not at all appropriate for political analysis. Here’s a bit from the early part of Obama’s address. It has a Flesch-Kincaid Reading Level of 10.2, a level that GLM said reflected Obama’s “elite ethos”
“Already, this oil spill is the worst environmental disaster America has ever faced. And unlike an earthquake or a hurricane, it’s not a single event that does its damage in a matter of minutes or days. The millions of gallons of oil that have spilled into the Gulf of Mexico are more like an epidemic, one that we will be fighting for months and even years.”
Okay, but let me show you another passage that I’ve chosen to exactly match the above passage in Flesch-Kincaid Grade Level. It ought to be equally reflective of an elite ethos:
“one Gulf Already, America unlike millions even has oil does of spill ever be its minutes of not for disaster the And the single matter event earthquake we this epidemic, are a damage spilled The worst into environmental months it’s that or of a that of faced. will oil an is or like a hurricane, fighting Mexico more days. in an gallons that have and years.”
If the extent of your analysis is to look at grade levels, you’re going to say that these two passages are equivalent. That’s because the Flesch-Kincaid Grade Level formula is merely a weighted linear combination of number of words per sentence and number of letters per word. Since these two paragraphs contain the same words, letters, spaces, and periods, the statistics are the same for each, and therefore any conclusion drawn about the first paragraph solely from these statistics necessarily must be drawn about the second paragraph as well.
That’s the problem. These statistics and readability tests don’t look into word frequency, semantics, pragmatics, fluidity, rhetoric, style, or anything that actual humans do to assess the readability and meaningfulness of a text. The tests, after all, are intended as an approximation for when an informed analyst is not available, not as a data source in lieu of informed analysis.
To be fair, GLM’s analysis doesn’t stop at grade levels. They also offer the proportion of passive sentences in the address, which they report as “the highest level measured in any major presidential address this century”. And that’s something, except for Mark Liberman’s discovery that it’s not nearly true. Bush’s similar post-Katrina address had 17% passives; Obama’s post-oil-spill address lagged behind with a mere 11%. (GLM’s president, by the way, considers “There will be setbacks” to be a passive sentence, so it’s not terribly surprising that their passive statistics aren’t great.) But even if the count were right, the passive proportion is not an inherently meaningful statistic either, because passives are employed by good writers for reasons other than evasion, which seems to be the only use GLM can come up with for them.
I hesitate to say that there is no useful information to be found by calculating simple statistics on major presidential addresses. But readability scores are dependent on the choice of punctuation for a speech, overlook rhetorical devices and structure, ignore frequency and semantics, and haven’t been shown to correlate very well with listener comprehension. It is unlikely that useful information will come such simplistic analyses. And though it is not impossible that one day someone will find it, I have not yet seen a single informative result from grade level or other simple statistical analysis on political speech.
There is nothing wrong in English with splitting an infinitive. There never was anything wrong with it, either. You probably all knew that already. Unfortunately, the loudest grammar snobs are the ones who’ve put the least research into their opinions, and so, for every ten people quietly aware that infinitives can and sometimes should be split, there’s one vocal grammaticaster shouting over them that split infinitives are an abomination in the eyes of Pope. That means that there’re still a substantial number of people out there either objecting to or grinding their teeth over Star Trek’s to boldly go.
These people are mistaken. But the fact that they are mistaken will not stop them from complaining and possibly thinking less of you. And you may very well be in a position where the opinion of the misinformed matters to you; you might be an author, editor, or even a job applicant whose cover letter will be read by a lunkhead whose personal grammatical prejudices may blind him to your outstanding qualities. This leads a large number of people aware that there is no linguistic reason to avoid split infinitives (or singular they, or sentential hopefully, etc.) to still avoid using them for fear that someone of some importance will judge them harshly. It’s an unfortunate state of affairs, best summarized by Ann Daingerfield & Arnold Zwicky’s line: “Crazies win“.
Now, in many cases, it’s not so bad. It’s unfortunate that reasonable people have to bow to the whims of the mad, but that’s life, innit? After all, would you really notice if someone changed (1a) to (1b)?
(1a) I’m going to angrily split infinitives.
(1b) I’m going to split infinitives angrily.
And sometimes it even sounds better to not split an infinitive:
(2a) Alfonso Ribeiro taught me to gracefully dance.
(2b) Alfonso Ribeiro taught me to dance gracefully.
But these bad-to-split situations are not as pervasive as some people seem to think. That’s because prescriptivists have a bad habit of not actually looking at the language that they’re claiming domain over. For example, the normally reasonable folks at AskOxford write that “Split infinitives are frequently poor style, but they are not strictly bad grammar,” and illustrate this claim with exactly zero examples. In so doing, they completely ignore the fact that sometimes the split infinitive is the only right way of doing it. For example, consider
(3a) She decided to gradually get rid of the teddy bears she had collected.
(3b) She decided gradually to get rid of the teddy bears she had collected.
(3c) She decided to get rid of the teddy bears she had collected gradually.
(3d) She decided to get gradually rid of the teddy bears she had collected.
(3e) She decided to get rid gradually of the teddy bears she had collected.
(3f) She decided to get rid of gradually the teddy bears she had collected.
This is an example from R. L. Trask. (3b) and (3c) unsplit the infinitive, but make it unclear where gradually is attached; is she gradually getting rid of the bears, gradually deciding to get rid of them, or getting rid of bears collected gradually? And (3d), (3e), and (3f) are just plain awkward, so if someone thinks a split infinitive is poor style, surely they’d think these ones still worse. A reasonable person might avoid split infinitives in other situations, so as not to incur the wrath of idiots, but in cases like this, no one would intentionally ruin their sentence in order to placate the misinformed.
Or so I’d figured. But then Amy McDaniel posted a worksheet from a class taught by David Foster Wallace, who very well may have been a talented writer, but also held some severely backward prescriptivist views, as discussed/destroyed at Language Hat. The worksheet is a list of sentences, each of which Wallace claims contains an error. One of the sentences is:
8. She didn’t seem to ever stop talking.
Now, in light of all this discussion about split infinitives, it’s clear that Wallace’s objection will be to the phrase to ever stop. But how do you fix it? The answer, given by McDaniel in the comments on the post, may surprise you:
the easy, unawkward fix, according to Wallace, is “She didn’t seem ever to stop talking.”
I don’t often use interrobangs, but: WHAT?! I could see “She didn’t ever seem to stop talking.” I could see “She didn’t seem to stop talking, ever.” Heck, I think I might even prefer one of those to the original. If you’re willing to change the words, you could also use: “She never seemed to stop talking.”; “It seemed she never stopped talking.”; “She seemingly never stopped talking.” Any of those would be reasonable, unawkward replacements.
But “She didn’t seem ever to stop talking”? Does anyone find to be that a good sentence, or even an “unawkward” one? It sounds awful to me, but then, being from Pittsburgh, I’m not entirely standard in my usage of negative polarity items like ever or anymore. If you like this sentence, please say so.
This is the weird thing with this worksheet: Wallace was a well-renowned writer, as well as a native speaker of English. How can someone so close to the language be so blind to what does and doesn’t sound like English? Because his re-phrased sentence most certainly does not.
Please, dear reader, I beg of you. Don’t let fear of what other people will say about your writing cause you to write something obviously awkward. And if you should disregard my plea, at least don’t pull a David Foster Wallace and convince people who respect you to fly in the face of all that sounds right in English.
[Hat tip to bradshaw of the future for pushing me to finish this post.]
Wallace’s other sentence revisions have already been intelligently discussed (attacked) in many other blogs, among them Arnold Zwicky’s discussion of each other and one another, Chris Potts’s succinct dismissal at Language Log, the truly stunning point-by-point gutting of the test delivered at Mackerel Economics, and another equally stunning point-by-point evisceration from Starlingford Chronicles. I highly recommend you check these out.