This has got to be old news to everyone else (I’m pretty late to the whole smartphone game).
Got my first ever smartphone (an iPhone) seven months ago. Only over the past week or two has it figured out my texts are more likely to use the word “just” than the word “Judy”.
I don’t even know anyone named Judy.
Autocorrect “rules”!
Something. I’m not sure what. Personally, it makes me crazy.
iphone iknowwhatyouwant.
Obviously, my phone does not know that I use the word “too”. It never predicts it until I backspace to add the additional “o”. sigh
I keep trying to convince mine that I’m approximately 12000 times more likely to be using the word “some” than “sine” and “like” than “Luke.” So far, my phone is smarter than I am.
I have recently started using an app on my phone to study a bit of Spanish. Since I’ve never texted much, the majority of my typing on the device has been in these lessons, alternating between English and Spanish. Worse, some of the sentences in the lessons are rather odd, if not downright wacky.
The phone’s autocomplete is now hopelessly confused and offers more or less random suggestions.
Mine is convinced that I use the word “puerile” more frequently than “people” people people people people (just trying to offset that one “other p word” with five “people” in case it is keeping count).
For some reason almost every time I type ‘just’ - a word which I appear to use with startling frequency, to my surprise - I do ‘jst’ and it corrects that to ‘jet’. I yearn for the day when it learns that one!
A better designed program would switch the language mode of the keyboard when you are supposed to type in another language. Or just disable autocomplete altogether.
Anyways, does anyone know why this seems so hard? Use common language, sorted by how common it is, and then learn from use . It seems so simple.
The keyboard/autocorrect app I use has multiple language support: you can choose different keyboard displays, different languages to grab dictionaries for, and switch autocorrect off completely.
But when it’s on, well, it’s got to deal with someone that can switch between three languages in the middle of a conversation, or of a sentence. My conversations with one of my brothers are in English (he wants to improve his level), with occasional bits of “how do you say…”; I speak with my aunt in Catalan (both of us want to improve our spelling), with either part occasionally having trouble remembering a word. That can be confusing to overhearing humans, I realized a while back that a machine can’t, won’t and possibly shouldn’t do better.
My last phone always used “bee” instead of “be.” How many people are text about bees!
I really dislike the autocorrect function on my phone and have turned it off. It would correct a word that I’d spelled right into a word that wasn’t even close, and it would do it to every word I’d type.
The phone I had before this one had a really nice feature, it would fix conjunctions and make ‘i’ uppercase. I loved it as I could type ‘ive’ and it would fix it to ‘I’ve’. I really wish my new phone would do that, but I’ve gotten used to adding the ’ when needed.
The worst autocorrect I’ve seen was in NeoOffice. I typed “Friday, 7 November 2014”, and it autocorrected it to “11/7/14”. It wouldn’t even let me undo the “correction”: Every time I undid it, it would just immediately re-“correct” it. I had to go search out that option (with help from Google) and turn it off before I could proceed. Why on Earth would I even be typing it out the long way, if I wanted it short? If anything, it should be correcting the other direction.
Actually, you do, you just don’t know her by that name. 
What I want to know is why my phone often autocorrects “so” to “do.”
Not on my phone, but when I type “eth”, THE is never given to me as a correction option when composing an email in Outlook. I get eh, et, Beth Seth, etch. Never the most likely word: THE!!!
And can we just forget about ducking?
Is it simple to pick up a book in a foreign language and teach yourself to read it? Language is hard. It’s big, complicated, and full of exceptions. Computers, in particular, find that last bit difficult.
Sophisticated autocorrects and autocompletes don’t just keep word counts. They track relationships between words as well. It’s how they offer suggestions before you even start typing the next word. A lot of that information is pre-populated by common usage statistics across lots of people. So, if you type “i”, it probably knows enough to offer “I” as an option, and once you select it, it offers a list of words it thinks are often associated with “I”–“am”, “have”, “will”, and so forth. That’s well and good until you have some reason to text about I, Claudius. The general usage stats are weighted enough that you would have to type that title in an awful lot before the phone starts assuming that “I” should always be followed with a comma, fortunately, but that’s also why it takes a lot of use for it to adjust to things you personally type often.
Things that don’t get used much in common texting don’t have as much statistical weight behind their associations, so they tend to be picked up faster by the software. Which is why my phone is now convinced that spiders eat bread.
“S” and “D” are adjacent on the keyboard, making it a common typo to type one in place of the other. The stats presumably indicate that “do” is more commonly used in texting, so when it sees “so”, it sometimes guesses that you made a typo and wants to substitute the more common word. It doesn’t do it every time because other words you’ve already typed are also factored in.
I’ve always hated spellcheck for that same reason. If I type “comples” (which I often do, from muscle memory), it never offers “complex” as one of the five alternative choices, and I have to manually correct…