A way to add words with accented letters to my personal English dictionary?

[ngoomie]

I've been using TT9 for a few days now and I like it a lot. I make a lot less typos than I do with a normal QWERTY keyboard and it's more convenient for some things than outright using a handwriting keyboard. One small point of pain for me though is that I can't seem to add any words with accents to my personal English dictionary (I like to type, for example, Québec instead of Quebec when I can).

I did see in the in-app help documentation that CSV import doesn't allow use of accented characters, so instead what I tried was typing them in either using another keyboard entirely or using the French mode in TT9, and then adding them to the English dictionary with the plus button, but this doesn't seem to actually add them properly as they never show up in my suggestions after that point.

Is there some other way around this, is it something you'd be interested in looking into fixing, or should I just keep switching back and forth between English and French? It's a bit of a pain since I actually have 3 language options enabled (Japanese also) so switching back and forth between English and French for one word isn't as smooth as it would be if I had only those two languages enabled, but I also understand if this is too niche a problem for you to consider fixing. Thanks in advance either way!

[ngoomie]

Poking around at the issues again just in case I missed anything relevant, I did discover the Google Play version has the 'grouped characters' option, so for now I'm going to try adding the French letters I need there and see how that works for me (conveniently, this also solves another issue I was having: no em/en dashes!)

I'll still keep this issue open just in case though :P

[sspanak]

It is not possible to add words with letters that are not included in a given language. The thing is, in T9 keyboards, you always type numbers, and the keyboard tries to interpret the digit sequence into one or more words. If there is a letter that does not match any digit, it will never appear when you type.

Nevertheless, "é" in particular, is indeed common enough in all English variants. For example, "née", "fiancée", and "café" are used often enough even outside Canada. Probably, I can make an exception and add "é" to English.

[rwded]

I'm in a similar situation. I've also been using TT9 for a few days now and I really believe it's a great app. However, I'm facing a similar usability issue due to words with accented letters not being supported in German. Both of my daughters’ names include accented letters such as é or ë, and when I'm writing text messages I often need to switch from German to French to type their names and then switch back to German. On older Nokia phones, accented letters were available across multiple languages (including German, English, French, and Italian), which made this much more seamless. I was wondering whether it might be possible to include accented letters in the German dictionary as well, or to allow a shared set of accented characters across languages. I believe this would be a valuable improvement for multilingual users. Thank you very much for your time and for your great work on TT9.

[sspanak]

I was wondering whether it might be possible to include accented letters in the German dictionary as well
...
I believe this would be a valuable improvement for multilingual users.

Well, I think almost everyone in Europe and Asia uses at least 2, if not 3 languages. Or if they don't, they still need to write words in an another language every now and then. I understand the problem very well.

But now we get to the reason why I don't want to add extra letters. I, myself, use 4 languages, Bulgarian, English, French, Japanese. I have friends and have had coworkers from Serbia, Romania, Ukraine, Poland, Sweden, Finland, France, Portugal, Spain, USA, Australia, India... I can't even remember all the countries anymore. In other words, I understand the problem very well.

Despite the above, I definitely do not want the special letters from all these languages in English (that I primarily use in TT9), or Ukrainian letters in Bulgarian. In my opinion this will reduce usability, not improve it. And I am sure many people will agree, especially those that prefer ABC mode. They constantly need to browse the list of letters, so adding 10-20 more per each key will turn typing into nightmare.

... or to allow a shared set of accented characters across languages.

I think that's the solution. How about an extra configuration screen? It can be similar to the Punctuation Order screen, but it will allow adding letters for the keys from 2 to 9, instead of punctuation marks for 0 and 1. After you add the new letters, you will also be able to add new words, then type them as usual.

There is only one complication - you will not be able to delete letters, because this will leave garbage words in the dictionary. The only way of deleting them will be by deleting the respective language dictionary too.

[rwded]

Thanks for the thoughtful explanation - the concern about adding extra letters and the potential impact on usability is completely understood. Especially for users who rely heavily on ABC mode, having to cycle through many additional characters per key would indeed make typing more cumbersome rather than better.

The idea of an extra configuration screen that allows users to add letters to keys 2–9 sounds like an excellent solution. It would give each user the flexibility to tailor the app to their own language mix and typing habits, instead of enforcing a one-size-fits-all character set. That kind of individual customization seems like a very user-friendly approach.

But I also see the difficulty you mention when it comes to deleting letters and the risk of leaving “garbage” words in the dictionary.
Might it be possible, however, to filter the suggested words and automatically exclude those that contain undefined characters? Alternatively (or additionally), perhaps a dedicated function could be added to clean the dictionary by removing all words that include letters no longer defined in the current configuration. I believe that could add the flexibility to "experiment" with custom letters without requiring users to delete an entire language dictionary.

[sspanak]

The idea of an extra configuration screen that allows users to add letters to keys 2–9 sounds like an excellent solution.

Awesome! Let's go with that.

But I also see the difficulty you mention when it comes to deleting letters and the risk of leaving “garbage” words in the dictionary.
Might it be possible, however, to filter the suggested words and automatically exclude those that contain undefined characters? Alternatively (or additionally), perhaps a dedicated function could be added to clean the dictionary by removing all words that include letters no longer defined in the current configuration. I believe that could add the flexibility to "experiment" with custom letters without requiring users to delete an entire language dictionary.

Filtering would be too resource-intensive and may cause typing lag. But maybe I can add an extra cleanup function that executes upon saving the letter list. This way, saving will take longer when also deleting letters. Unfortunately, I have no idea how much longer, but hopefully, it will be in a matter of seconds. Since the words to delete will be custom words, it shouldn't take that long locate them.

Thanks for the ideas!

[rwded]

Great!
Looking forward to seeing it implemented.
Just bought you a couple of beers to help you stay focused. 😊

[ngoomie]

My other, less elegant thought of how to solve this was to allow adding words with accents to the dictionary without adding extra accents to the keyboard, and search/filter for them as if they were spelt without the accent, if that makes sense? Then I could bang out 783232 and it would match Québec as though it were written Quebec, or the same for (i.e.) 5242483 matching laïcité as though it were laicite. I imagine this could potentially be computationally expensive though, or maybe even a nightmare to program depending on how manually you have to define which accented letter maps to which unaccented letter (I think this is why notetaking software Obsidian doesn't have accent-stripped filename search as a base feature) and if you can even map one digraph to two plain ASCII letters (like œ/oe), though that part could maybe be alleviated by making those mappings user-defined. It also definitely means you'd have to use another language in ABC mode for the initial word add, but then any subsequent typings would be fine, maybe?

[sspanak]

My other, less elegant thought of how to solve this was to allow adding words with accents to the dictionary without adding extra accents to the keyboard, and search/filter for them as if they were spelt without the accent, if that makes sense? Then I could bang out 783232 and it would match Québec as though it were written Quebec, or the same for (i.e.) 5242483 matching laïcité as though it were laicite. I imagine this could potentially be computationally expensive though, or maybe even a nightmare to program depending on how manually you have to define which accented letter maps to which unaccented letter (I think this is why notetaking software Obsidian doesn't have accent-stripped filename search as a base feature) and if you can even map one digraph to two plain ASCII letters (like œ/oe), though that part could maybe be alleviated by making those mappings user-defined. It also definitely means you'd have to use another language in ABC mode for the initial word add, but then any subsequent typings would be fine, maybe?

This is more complicated. Let's stick with the extra letters option.

[ngoomie]

Yeah lol I didn't think you'd go for that one anyways because of that, I just figured it could be worth a mention :p