I'm fond of Vale, where you can write your own rules or grab others off github. I wish we could crowd source a grammarly clone using it. Some efforts have been made to do so, but didn't get far.
LanguageTool is based on a pretty powerful open-source tool, that's running in the cloud but can also be run on-prem. So, no need to reinvent the wheel.
Basically, whenever you catch an error in your writing you add a rule to ensure that it is caught the next time. After a few years your rule set will catch a large portion of your writing errors. It's fun too.
Sure, here are two, but I have hundreds. The second one triggers quite often for me.
The art is trying to writing a rule without too much false positives. False positives are myriad. Probably 60% of warnings are false positives, so you don't want to add too much more to that. On a typical document, I can get ten pages of errors, fortunately, going through it goes quite fast.
<rule id="" name="">
<pattern>
<token>comprise</token>
<token>be</token>
</pattern>
<message>comprise be</message>
<example correction=''>The thingy may <marker>comprise be</marker> configured to do things.</example>
<example >The thingy may be configured to do things.</example>
</rule>
<rule id="" name="">
<pattern>
<token regexp='yes'>arranged|configured</token>
<token>to</token>
<token postag='VBG'></token>
</pattern>
<message>You followed <suggestion>arranged to</suggestion> or <suggestion>configured to </suggestion> with a gerund.</message>
<example correction=''>The thing is <marker>arranged to replacing</marker> the other thing.</example>
<example>The thing is arranged to replace the other thing.</example>
</rule>
I have been a user of the Language Tool API as we're building (experimenting with?) a language learning app [1] and grammar correction is a key component. I really want to love it, and I'm so appreciative of the open source version that has allowed us to get going, but it's just so... limited. I haven't really dug into the code, but I gather that it is very rules based and that seems destined to fail because so much of language is context sensitive.
For example, it really can't determine in German if you're using the correct case or not for a particular verb. Or it can't really tell if you're using the right prepositions.
Ultimately, we ended up building a grammar engine just for language learners using transformers [2] + a version of errant [3] we ported to be multilingual. It's not perfect, but after a couple weeks of development was more helpful than LT.
Making too good open-source solution would ruin the business model of these guys.
I bet that most of the meaningful features are not included there. (I haven't looked into the code, just pricing).
Apologies - this wasn't clear. I didn't mean to say we generalize to all cases. I should have said it was more helpful for some of the common cases that plague language learners.
The Jetbrains Grazie plugin [1] for IntelliJ IDEA uses Language Tool under the hood. I've been mostly using it for checking my blog posts, but it can also inspect code comments and commit messages.
The premium version has a semi-hidden AI suggestion feature. If you double click a word it will select the whole sentence and suggest rewrites that are more formal or shorter. I've found the AI feature great for inspiration.
I use both Grammarly and LT. I wish LT had an option to select the desired tone like Grammarly does. The formal one is occasionally useful but is waaay too formal for regular work-related communications (at least in the US).
I've been using LanguageTool for a long time, and I'm also happily paying their premium subscription. 100% worth it and highly recommended. It's also excellent at non-English languages.
There is a big difference in terms of privacy: LanguageTool is open source and you can run it on your own computer. You can install it as browwer plugin, a LibreOffice plugin, or a standalone server that you can access from Emacs for example (the latter is no longer promoted on their website, but you can download JAR files from https://languagetool.org/download/).
Same here, very very smart. It even gets typographic quotes right when nesting them (such as „The sign said ‚yes‘, clearly.“), keeps an eye on not jumping between tenses or different forms of addressing people („Du“ vs. „Sie“ in German).
No Android or Linux app. Subscription is high (like others have said). Appears no offline version (might be wrong) and the text is stored on their servers be default unless you manually delete it.
No subscription is required for the base version no account, either. Of course, the base version doesn't have all the features as premium, but it is still excellent, in my opinion. I have used the free version for years. There is an offline version, as it is open source.
The full text isn't stored on the servers, but you can always just deploy it yourself for maximum privacy:
Sorta, you can always run the server locally on your own machine. It's a bit wasteful, but not terribly so.
> No Android or Linux app
I mean you can just use the plugins for browsers, email, and document writing. It won't work in the terminal (would it on MacOS regardless?) but I wouldn't consider that a downside. I'll be waiting for the Emacs package myself, but I doubt I'll have to hold my breath.
There is surely a version for Linux. I have been using it in my computer with Archlinux [1] for a couple of years with vim (there are at least two plugins [2,3]). It works pretty good. The only thing is that when running it locally it kind of uses a lot of resources. So not so good when you are with your laptop running on battery.
Readit News
https://github.com/errata-ai/vale
https://github.com/testthedocs/Openly
A really nice feature of LT is that you can add your own rules. A rule editor is provided here:
https://community.languagetool.org/ruleEditor2/
Basically, whenever you catch an error in your writing you add a rule to ensure that it is caught the next time. After a few years your rule set will catch a large portion of your writing errors. It's fun too.
The art is trying to writing a rule without too much false positives. False positives are myriad. Probably 60% of warnings are false positives, so you don't want to add too much more to that. On a typical document, I can get ten pages of errors, fortunately, going through it goes quite fast.
<rule id="" name=""> <pattern> <token>comprise</token> <token>be</token> </pattern> <message>comprise be</message> <example correction=''>The thingy may <marker>comprise be</marker> configured to do things.</example> <example >The thingy may be configured to do things.</example> </rule>
<rule id="" name=""> <pattern> <token regexp='yes'>arranged|configured</token> <token>to</token> <token postag='VBG'></token> </pattern> <message>You followed <suggestion>arranged to</suggestion> or <suggestion>configured to </suggestion> with a gerund.</message> <example correction=''>The thing is <marker>arranged to replacing</marker> the other thing.</example> <example>The thing is arranged to replace the other thing.</example> </rule>
For example, it really can't determine in German if you're using the correct case or not for a particular verb. Or it can't really tell if you're using the right prepositions.
Ultimately, we ended up building a grammar engine just for language learners using transformers [2] + a version of errant [3] we ported to be multilingual. It's not perfect, but after a couple weeks of development was more helpful than LT.
1. https://squidgies.app
2. e.g. this paper from Grammarly https://arxiv.org/abs/2005.12592, but there are a bazillion others
3. https://github.com/chrisjbryant/errant
Wow! If your results generalize to all cases why shouldn’t they scrap everything they have and start over?
[1] https://plugins.jetbrains.com/plugin/12175-grazie
I mean, sending all data to a Grammarly service brings the same privacy issues as sending it to Language Tool.
This is not true of Grammarly last I checked, you have to hack something together... (e.g. https://medium.com/@tardijkhof/how-i-made-grammarly-seamless... )
https://github.com/kitten/prosemd-lsp
I'll not be using it but good effort though.
The full text isn't stored on the servers, but you can always just deploy it yourself for maximum privacy:
https://dev.languagetool.org/http-server
Furthermore, there are options for docker: https://github.com/languagetool-org/languagetool#docker
Sorta, you can always run the server locally on your own machine. It's a bit wasteful, but not terribly so.
> No Android or Linux app
I mean you can just use the plugins for browsers, email, and document writing. It won't work in the terminal (would it on MacOS regardless?) but I wouldn't consider that a downside. I'll be waiting for the Emacs package myself, but I doubt I'll have to hold my breath.
https://github.com/mhayashi1120/Emacs-langtool
Now that I checked, looks like there’s another package, but I didn’t t try it.
https://github.com/emacs-languagetool
[1] https://archlinux.org/packages/community/any/languagetool/ [2] https://github.com/dpelle/vim-LanguageTool [3] https://github.com/rhysd/vim-grammarous