Typo likelihoods?

[Lord Antares]

I am trying to get information (statistical preferably) on how typo likelihoods work. Do you know what would be the most likely typos of a random word? Do you know how sequence and the position of the previous character would affect these? What about typos on mobiles? I have done a search on google but I can't find anything relevant.

Does anyone here know any info? Has anyone researched it a bit? Thanks.

[OldChemE]

Check out letters that are on opposite sides of a standard keyboard.  when a touch-typist has to type one letter with the left hand and the next with the right (or vice versa) typos arise if the fingers of the two hands are not equal in speed of movement.  For example, I have a tendency with words that end in 'tion' of typing 'iton' because my right hand seems to be a little faster (I am right handed).  Similarly, I tend to type 'hte' instead of 'the'  again because I seem to reach the letter h with my right hand a split-second faster than I can hit the 't' key with my left.  I haven't done any research, so I don't know if this is just because I am right handed, or if it is just an age thing-- my left slowing down a little, or some other factor.

[Lord Antares]

Thank you. I did not even consider this. I was only looking into keyboard key proximity, direction and previous key position etc. Also, I imagine the answer might different if one types on the phone. Anyone else got any knowledge to share?

[Strange]

There is some discussion of this here: http://languagelog.ldc.upenn.edu/nll/?p=37059

It seems there is a shortage of research (or even data) in this area. 

[Sensei]

Why not make scientific experiment.. ?

Make keylogger which will store the all data entered by people who will receive it (friends/family?).

Then after months of using, they will send you back results, stored words with errors and fixes.

Then make parser which will find places where they made mistake (use database of English words)..

 

But you would have to give them some way to filter out passwords and logins they use.

[Lord Antares]

2 hours ago, Strange said:

There is some discussion of this here: http://languagelog.ldc.upenn.edu/nll/?p=37059

It seems there is a shortage of research (or even data) in this area. 

Thanks. Yes, there really is a shortage. I've had zero luck with google searches and I mean zero.

2 hours ago, Sensei said:

Why not make scientific experiment.. ?

Make keylogger which will store the all data entered by people who will receive it (friends/family?).

Then after months of using, they will send you back results, stored words with errors and fixes.

Then make parser which will find places where they made mistake (use database of English words)..

 

But you would have to give them some way to filter out passwords and logins they use.

 

I would if I knew how to. 

On the other hand, asking a bunch of people (say, on a reddit) to type words fast would only get me so far and would in no way be representative of an actual statistic.
Also, a lot of people use mobile phones for browsing and typing. This would need to be considered as well.

[Sensei]

31 minutes ago, Lord Antares said:

I would if I knew how to. 

You have opportunity to learn programming C/C++..

To install keylogger routine you need SetWindowsHookEx()

https://msdn.microsoft.com/en-us/library/windows/desktop/ms644990(v=vs.85).aspx

and it must be put in separate DLL (Windows XP+ requirement, in Win95 it could work without separate DLL).

The most primitive rough version can be done within 10 minutes or so.

 

Edited March 31, 2018 by Sensei

[Lord Antares]

No thanks. Although programming is something that sounds interesting, one would need to dedicate a LOT of time to it. I can only dedicate so much to things I like, and there are many, many things worth dedicating to. Programming is not on top of my list.