- cross-posted to:
- technology@lemmy.world
- cross-posted to:
- technology@lemmy.world
Source: https://front-end.social/@fox/110846484782705013
Text in the screenshot from Grammarly says:
We develop data sets to train our algorithms so that we can improve the services we provide to customers like you. We have devoted significant time and resources to developing methods to ensure that these data sets are anonymized and de-identified.
To develop these data sets, we sample snippets of text at random, disassociate them from a user’s account, and then use a variety of different methods to strip the text of identifying information (such as identifiers, contact details, addresses, etc.). Only then do we use the snippets to train our algorithms-and the original text is deleted. In other words, we don’t store any text in a manner that can be associated with your account or used to identify you or anyone else.
We currently offer a feature that permits customers to opt out of this use for Grammarly Business teams of 500 users or more. Please let me know if you might be interested in a license of this size, and I’II forward your request to the corresponding team.
It still isn’t clear why anyone uses a product developed by non-native speakers to check their writing. For anyone who knows grammar, Grammarly sometimes makes… interesting… suggestions.
As a non-native speaker I’m surprised to the amount of grammar mistakes native speakers make. Being a native speaker is not a testament to how much of the language you know. And even that being true, it’s not like a real human corrects your text, so the creators being native or not is pretty much irrelevant.
at the amount of grammar mistakes
should’ve used grammarly
They’d’ve gotten it wrong too. Prepositions and postpositions are their own category of linguistic hell, especially in idioms and phrasal verbs.
They’dn’t’ve necessarily gotten it wrong. With a big enough dataset, an ML tool should be pretty accurate, at least in that it will make the same choices as most people have made in their writing.
Apostrophe mistakes aside, no native speaker would stack contractions like this. There’s an upper limit of three words in a single contracted form. It would be “They wouldn’t’ve gotten” or “They’d not’ve gotten.”
ML tools don’t write grammatically correct complex sentences precisely because their training sets contain too many discrepancies. They may learn how to apply prescriptive rules consistently one day, perhaps even one day soon, but this is not that day.
Who says there’s an upper limit? You might not be one of those people, but I’m.
Also, that’ll teach me to try to write tricky comments while also doing other things. Fixed.
Muphry’s Law strikes again.
Well, linguists say it. But you do you, friend.
LOL! Right there with you. If I had a dollar for every time this happens to me…. 😄
Native speakers don’t usually make major grammar mistakes. They may not follow prescriptive rules, but they’re generally understandable by other native speakers because grammar is so deeply embedded in their subconscious that they can’t help handling the language correctly. You do the same in your native language. Everyone does.
The problem with non-natives, and I include myself as a non-native speaker of a few languages, is that we don’t usually have the same instincts. It would be pretty arrogant to tell a native that they don’t know how to use their own language when we, almost by definition, cannot possibly understand it in the same way that they do.
well said/written
it’s not only that “we don’t usually have the same instincts”, we have a burden of confusing loans, imports, translations, false friends &c.
When you start dealing with gendered languages, it’s even worse. There’s no logic to it. A hand is a she in one language, a he in another and neutral in third.
also, this pronoun question of culture wars is ridiculous for someone who can speak non-gendered languages 🤷
(Love your handle)
I get what you’re saying about gendered languages. But if you speak one long enough, even as a non-native, you’ll start to develop a feel for genders and be able to predict them to some degree. So far as I know, the mechanism that determines gender is so deeply subconscious that no one has been able to find and articulate its rules, but it seems to exist.
Re: culture wars - The pronoun question is probably moot point in truly genderless languages. English, unfortunately, is not completely genderless, so it’s a bone of contention in the current climate.
all of which is irrelevant to how grammarly works
It is?
Grammarly seems optimized to make the user sound smart, not necessarily to make them type flawless English. Only spam bots, AI, and nitpickers write out perfect English all the time.
Email spam usually has heavily flawed English.
I’ve heard that this is intentional. It would be a waste of the spammer’s time to be contacted by people who are smart enough to not be fooled. Those smart people won’t bother contacting the spammer and wasting the spammer’s time if they see grammatical errors in a message that purports to be from a reputable organization, so the spammer throws in some errors to make the smart people filter themselves out. Or so the theory goes.
It’s a common theory.
That’s not the only spam out there, though. The real dangerous email is the phishing spam that’s written to convince the CEO or CFO to send money to the wrong account. You can’t go all Nigerian Prince if you want to steal the big bucks from companies.
Marketing spam is also often grammatically correct.
I’ve seen this filtering hypothesis, and it seems plausible. OTOH, it also gives James Veitch some fantastic material for his comedy routine.
*nitpicker (but I prefer pedant in polite circles, and grammar nazi on the Internet, or at least I did until actual nazis started showing up again)
Nice job, very appropriate this time :)
Certain uni composition students had better learn to write flawless English if they expect to earn their desired grade in my courses.
For sure, but they won’t practice their perfect paper English outside of your courses. The PDFs they submit will be (mostly) correct, but I doubt they’ll put as much effort into their emails, let alone emails to anyone else. It’s not like customer support will correct you on your they’re/their/there/deer mistakes.
Maybe customer support should take a stronger stance on understanding and being understood using standard dialect. At least the CSRs that I usually seem to talk with could use a good basic communication course.
Students will use what they learn from me more than you think if they want a degree. If they don’t want one… well, we have several excellent nearby trade schools where they can learn a skill that won’t require formal standard English and will make them a whole lot more money in the long run (I’m honestly saying this respectfully).