Feature Request: True multi-lingual spell check.

[Mart°n]

The language style will also have settings for word boundaries. This includes whitespace and/or punctuation to let Mellel differentiate between one word and another of that language.

That’s the point! Why should the application decide for me whether the question mark belongs to the Hebrew or to the English part? Based on which rules? (see also the objections with regard to Chinese and Japanese)

I don’t think that Mellel should decide to which language a punctuation mark belongs but Mellel could detect the beginning and the end of a word based on a whitespace and punctuation marks. In case of spell checking and hyphenation, I don’t think that it matters to which language a comma belongs.
The boundary detection could be based on the selected language. It could be enable for latin languages and disabled for languages where it doesn’t fit (cjk).

In the case that there are two or more language styles within the same "word", then the rules for the language that's set as the paragraph default take precedence.

Why should anyone want to assign only half a word to a foreign language? Do we really need those ludicrous “security measures”?

I don’t think that anyone wants to assign a language to only some letters of a word, but if you use the visually same style with two language settings, this could happen, if you do a lot of copy&paste or delete some words across two language settings. Once it has happened, it’s difficult to find those “mutant” words and correct them. I (blind as I am) couldn’t see the drawback of an “only one language per word” option, and think it is a helpful one.
Ther’s also a one color per character limit built into Mellel that prevents you from creating rainbow characters which is useful, I think (drawing applications don’t have this limit, for example)
Silly example, but I don’t have another at hand.

Maybe we should follow the less-is-more principle and let the Redlers do their job. I’m sure they’ll come up with a usable, user-friendly solution.

Maybe. But disscussions about a feature could save both, the users and the Redlers much time and workload. Who benefits, if the Redlers spent some weeks to realize a option that nobody could use because it doesn’t fit the user’s needs?
Discussing a feature also is the main reason for the Nitty and the Gritty forum. Otherwise it may have been named “Wait and see!”

[verma]

I'll add a bit of mine. I understand the problems with language as character style, and perhaps also the added complexity of a level in between paragraph and character. I have no clue about how this is coded now in Mellel, but it seems to me that one could perhaps try another strategy, instead of using another (new) "level".

1. Spelling. For me, spelling is just a tag, in HTML speak <language="English">Well, you see?</language="English">. In monolingual settings, you tag the whole document with a specific language. In multilingual settings, the language-tag can be applied to larger portions of text, or to individual words. Combining parts of words in different languages (Mart°n's previous example of combined tags as in <english>atta<german>ch</german>ment</english>, would be technically fine iff <german>ch</german> had been marked as such; and in <english>atta<german>ch</english> we would have a tag conflict, and hence would be marked as incorrect. Means: user intervention required. Seems fair to me. Or if one copy/pastes a word from a larger portion of text (where the whole paragraph or section is marked for a specific language), that copied word will not be marked for language until it's inserted into another language-defined area, or explicitly marked for a specific language.

2. Hyphenation. Hyphenation, the principle, is a paragraph thing. No doubt about it. But this is just the "when" part. The "where" part is a word thing: languages use sometimes language-particular syllabification rules, and that's where the language-tag (and hence the specific hyphenation rules for language X) serves a purpose: in the general scheme of hyphenation things, the language tag of a word in language Y signals that this particular word obeys to hyphenation rules of language Y; even if it's used inside a paragraph that has a language tag "X". Hyphenation, then, if it involves that specific word (that's determined on a paragraph basis), will have to follow the hyphenation rules of that particular language (assuming the word is not an exception) - checking the hyphenation rules for that specific language.

3. Typographer's quotes. For typesetting purposes, one is best served with a general convention that if you use language X, you follow the typographer's quotes for language X. However, not everyone likes that - so I'd suggest the option to turn the connection between language & typo quotes on/off at will: one has to be able to write in French and use typoquotes from English. Typographer's quotes, in my view, are in theory blind to the linguistic specification (which language) of the elements that they delimit, but not to the linguistic setting of the text they are part of, so you could very well use English typoquotes in an English text, and still quote French words (tagged for language=French) without seeing the typoquotes change - that is, if multilingual settings are an option. Again, if you have larger portions of multilingual text, you could also set the typoquote option to reflect your changing language choice consistently.

In a sense, the flexibility we seem to want with linguistic settings, is not completely different from character variations like boldface or italics although it's possibly unfit to call it "character variation" because a French word in bold has to remain French when it's turned into italics. It should however be possible to apply a language setting to whatever portion of a text/word one would like, much in the same sense as we can do that with boldface or italics, and to attach the Language Tag to specific paragraph styles as well.

Hmm ... well with more time and ideas, I'd have to flesh this out in a more decent way, but what can you reasonably expect at 10:00 am on a Saturday morning?

[rpcameron]

The language style will also have settings for word boundaries. This includes whitespace and/or punctuation to let Mellel differentiate between one word and another of that language.

That’s the point! Why should the application decide for me whether the question mark belongs to the Hebrew or to the English part? Based on which rules? (see also the objections with regard to Chinese and Japanese)

I don’t think that Mellel should decide to which language a punctuation mark belongs but Mellel could detect the beginning and the end of a word based on a whitespace and punctuation marks. In case of spell checking and hyphenation, I don’t think that it matters to which language a comma belongs.
The boundary detection could be based on the selected language. It could be enable for latin languages and disabled for languages where it doesn’t fit (cjk).

Mart°n is closer to what I was thinking with this. The whitespace and/or punctuation would not belong to a specific language, but serve as markers to define what a "word" is. Punctuation rules, such as the possibility of allowing punctuation to extend beyond the margin of a paragraph (quite frequently used in CJK–justified environments) would be a language setting as well, but would be part of the paragraph's settings, and not the "word" level. The boundaries of whitespace and/or punctuation are the only means I could think of to determine what ought to be spell–checked. If you can think of something else that makes more sense, then I'm all for it.

In the case that there are two or more language styles within the same "word", then the rules for the language that's set as the paragraph default take precedence.

Why should anyone want to assign only half a word to a foreign language? Do we really need those ludicrous “security measures”?

I don’t think that anyone wants to assign a language to only some letters of a word, but if you use the visually same style with two language settings, this could happen, if you do a lot of copy&paste or delete some words across two language settings. Once it has happened, it’s difficult to find those “mutant” words and correct them. I (blind as I am) couldn’t see the drawback of an “only one language per word” option, and think it is a helpful one.

No, this is not to allow users to randomly assign parts of a word to one language. The intention of this was to provide a fall–back method for when a user accidentally does this. We also need to think about these situations, because while you may never mistakenly allow more than one language to be assigned to a portion of the word, other users may not be so careful. Also, just because you use Mellel for one type of work, it's very possible that other users will find it useful to break their words down into their constituent morphemes, and mark those morphemes for their language of origin. (I personally would not be opposed to this use. And, I'm pedantic enough that if I am meta–marking my document for language, then each portion ought to be correctly assigned, even if only for completeness' sake.)

Maybe we should follow the less-is-more principle and let the Redlers do their job. I’m sure they’ll come up with a usable, user-friendly solution.

Maybe. But disscussions about a feature could save both, the users and the Redlers much time and workload. Who benefits, if the Redlers spent some weeks to realize a option that nobody could use because it doesn’t fit the user’s needs?
Discussing a feature also is the main reason for the Nitty and the Gritty forum. Otherwise it may have been named “Wait and see!” :mrgreen:

Again, I'm completely with Mart°n on this point. The reason this forum exists is for users to discuss features, both implemented and those not–yet–implemented. In addition, if we don't fully discuss what our needs are for particular features, when they are implemented we may have many users bemoaning the fact that, while have a language style implemented in a future version of Mellel is great, it does do exactly what they need. This is our chance to let the Redlers know what we need from certain features. If we don't tell them, then they'll never know.

[Mart°n]

I have no clue about how this is coded now in Mellel, but it seems to me that one could perhaps try another strategy, instead of using another (new) "level".

Sounds o.k, I only haven’t found the hint to those strategy in your post.

1. Spelling. For me, spelling is just a tag, in HTML speak <language="English">Well, you see?</language="English">. […] Combining parts of words in different languages (Mart°n's previous example of combined tags as in <english>atta<german>ch</german>ment</english>, would be technically fine iff <german>ch</german> had been marked as such; and in <english>atta<german>ch</english> we would have a tag conflict, and hence would be marked as incorrect. Means: user intervention required. Seems fair to me.

My first thoughts were the same, but I think this doesn’t work in a word processor like Mellel. If you edit your text with an HTML (text) editor or work with LATEX the language markup is an easy thing and would work perfectly the way you’ve described. Working in an WYSIWYG editor is quite different as you couldn’t see the tags at all and the word processor is the one that needs to set them correctly. One (not very likely) example:

You write (or copy&paste) the sentence:

John has bought a new multimedia device, he likes to use for a specific purpose.

Later you decide to move the word “device” to the end (selection in blue):

John has bought a new multimedia device, he likes to use for a specific purpose.

which leads to:

John has bought a new multimedia, he likes to use for a specific purpose device.

Now you like to delete some words you don’t need any longer (selection in blue):

John has bought a new multimedia, he likes to use for a specific_ purpose device.

which leads to:

John has bought a new multipurpose device.

If you try this in Mellel, you’ll notice that the bold tags <bold></bold> used in this example were automatically corrected so that a case like
<english>atta<german>ch</english> couldn’t happen in Mellel. If you create a example as the one above only with two languages and replace the <bold> tag by a language tag, the result would be a bilingual word. As the user couldn’t instantly see this word (as only the language but not the appearance is different on the first and the second half of this word), it’s likely that he has created such a monster without knowing or noticing it. When writing technical or computer related stuff, it could be easy to end up with such words, as for example with German and English, some of those words could be spelled in both languages (as they are adopted from the English version) or parts of a word could be spelled in both languages.

If a language is attached to a character style, I guess it would behave exactly the same and this behavior doesn’t prevent examples like the above one. A case where a closing tag would be missed and user interaction is required to solve the problem would not occur here. That’s why I’ve requested the “word” limit.

Or if one copy/pastes a word from a larger portion of text (where the whole paragraph or section is marked for a specific language), that copied word will not be marked for language until it's inserted into another language-defined area, or explicitly marked for a specific language.

Why should a marked word lose it’s language because you’ve copied/pasted it? If you copy/paste a styled word (bold/font/size) the style also would be preserved. The “auto-clean” behavior is one of the things I hated most when I had to deal with PowerPoint some time ago.
If you use Edit › Copy Special › Plain Text it would be o.k. to delete the language tag too, but otherwise I would like to preserve it.

It should however be possible to apply a language setting to whatever portion of a text/word one would like, much in the same sense as we can do that with boldface or italics, and to attach the Language Tag to specific paragraph styles as well.

Usually I would agree but as the main goal of this feature request is multilingual spell checking and the added wagon could be called multilingual hypenation the option to apply a language to whatever portion of text would knock out (or, even worse, lead to wrong results) both features as multilingual words couldn’t be spell checked nor hyphenated (let alone correctly).
The tricky thing about this is, that you couldn’t see the language (only if you select some crazy colors for your different language styles) as a language tag is a invisible thing in opposite to the mentioned boldface or italic One could workaround this by either using the color example from above or by a search/replace action, that would find multilingual words, so that you could correct them.

But the question I’ve asked myself is why should one use a workaround for something that could be prevented from the beginning. Considering that there is no use of creating multilingual words regarding spell checking and hyphenation, my personal answer were “word” limits that would prevent bilingual words.

One possible use of this was brought up by rpcameron:

Also, just because you use Mellel for one type of work, it's very possible that other users will find it useful to break their words down into their constituent morphemes, and mark those morphemes for their language of origin. (I personally would not be opposed to this use. And, I'm pedantic enough that if I am meta–marking my document for language, then each portion ought to be correctly assigned, even if only for completeness' sake.)

but while it may be o.k. if you like to tag pieces of words with different languages, the option to spell check or hyphenate those pieces of text would be knocked out by this (for the sake of an invisible tag). If a teacher likes to create some papers where different portions of words where marked up differently, I would use a visible markup (as colors/styles…) as the students could comprehend and see the differencies, but that’s only me.
A invisble language tag is hard to see (which may be only true for my eyes ) on a printout.

[joewiz]

Punctuation rules, such as the possibility of allowing punctuation to extend beyond the margin of a paragraph (quite frequently used in CJK–justified environments) would be a language setting as well, but would be part of the paragraph's settings, and not the "word" level.

Great point about the need for Mellel to "allow punctuation to extend beyond the margin of a paragraph (quite frequently used in CJK–justified environments)" -- as language & word are formalized in Mellel.

[verma]

Hi Mart°n, thanks for your time & reply.
I was referring to the tagging-strategy à la HTML or LaTeX.

Working in an WYSIWYG editor is quite different as you couldn’t see the tags at all and the word processor is the one that needs to set them correctly. One (not very likely) example:
You write (or copy&paste) the sentence:
John has bought a new multimedia device, he likes to use for a specific purpose.

If Mellel does automatic tag correction, then tagging is the way to go. After all, if you insert the same language item into a bigger portion of text that has the same language tag, the item will fit in without problems. In such a case, the copied word gets the default language setting for that paragraph and, if not spelled correctly for that language, will be marked appropriately. And if your item is from another language-style, if you respect word boundaries, I think hyphenation will simply call in the hyphenation rules for different portions of the text. Suppose I start a paragraph with <english> and end it with </english>. It means all text between those tags will be considered English for spelling and hyphenation purposes. Then, if I copy a part of a word from another paragraph, which I tagged <german>... erolsteiner ...</german>, the general tag from the copy/paste operation could be transferred to the copied portion, and then it means I'm copying <german>erolsteiner</german> into an English setting Gerry<german>erolsteiner</german resulting in a word that has partly English and partly German spelling. At that point, since spelling rules will go nuts (there's no word boundary) ... they'll mark it as incorrect.

Usually I would agree but as the main goal of this feature request is multilingual spell checking and the added wagon could be called multilingual hypenation the option to apply a language to whatever portion of text would knock out (or, even worse, lead to wrong results) both features as multilingual words couldn’t be spell checked nor hyphenated (let alone correctly).

Well, if you're using tags, I don't see the problem. You would have a specific word tagged as <english>frankly</english> embedded in a larger paragraph tagged <german>... </german>. The <english> tag will be hyphenated and spell-checked according to the english rules.

The tricky thing about this is, that you couldn’t see the language (only if you select some crazy colors for your different language styles) as a language tag is a invisible thing in opposite to the mentioned boldface or italic One could workaround this by either using the color example from above or by a search/replace action, that would find multilingual words, so that you could correct them.

The other option is to use a small flag-icon somewhere - as you browse over your text, when it encounters changes, it verts into a new flag etc.

A invisble language tag is hard to see (which may be only true for my eyes ) on a printout.

Right, but I don't need to see a tag on a printout. I have to be able to see its effects on screen only, though. And if visbility is an issue, I wouldn't mind having a small icon inside the text, as an "invisible" item - which you could turn off or not. I believe that's a minor problem?

[Mart°n]

Then, if I copy a part of a word from another paragraph, which I tagged <german>... erolsteiner ...</german>, the general tag from the copy/paste operation could be transferred to the copied portion, and then it means I'm copying <german>erolsteiner</german> into an English setting Gerry<german>erolsteiner</german resulting in a word that has partly English and partly German spelling. At that point, since spelling rules will go nuts (there's no word boundary) ... they'll mark it as incorrect.

That’s o.k. so far, but I think it would be helpful, if bilingual words could be highlighted in a different way so that you could see them even if you’ve disabled spell checking.

Usually I would agree but as the main goal of this feature request is multilingual spell checking and the added wagon could be called multilingual hypenation the option to apply a language to whatever portion of text would knock out (or, even worse, lead to wrong results) both features as multilingual words couldn’t be spell checked nor hyphenated (let alone correctly).

Well, if you're using tags, I don't see the problem.

I’ve interpreted the phrase “portion of text” to include partial words. Without those multilingual words, there won’t be any problem but with them, hyphenation could produce wrong results as not only complete words are hyphenated but there are also rules that hyphenate specific letter combinations.

The tricky thing about this is, that you couldn’t see the language…

The other option is to use a small flag-icon somewhere - as you browse over your text, when it encounters changes, it verts into a new flag etc.

Could be a option but also could lead to a screen full of flags, if you change your language quite often.

A invisble language tag is hard to see […]on a printout.

Right, but I don't need to see a tag on a printout. […] I believe that's a minor problem?

I also don’t need to see the tag on a printout. My concern was about rpcameron’s idea to tag parts of words (morphemes) in different languages “even if only for completeness' sake”. One could do this if there’s a need but with regards to spell checking and hyphenation, it would lead to unusable results. As the tagging of parts couldn’t be seen on the printout but the non working hyphenation and spell checking does influence the printout, I’ve asked myself if this action (tagging morphemes) justifies the option to tag parts of words in different languages.

To conclude my personal needs:

• No multilingual words within my text. If a warning (maybe a blue highlight color as we have a red one for misspelled words now), a limitation to one language per word (word boundaries) or another option will prevent this, doesn’t matter to me.

• Keyboard Shortcuts. I would like to see the option to switch languages via a keyboard shortcut. The Language Style described above would do the trick as different characters styles could solve the problem, but:

• No Character Style mess. I use fairly well planed character styles for some of my documents that not only cover the different visual styles but also different content markup styles. That means, that I assign different styles to normal text or to a citation (even if they look the same). This leads to some more style than I would need to format my document the way I like. This also leads to my variations all being already used. The same is true for my Character Styles keyboard shortcuts. There are none left and I would give some away. If the language setting will be squashed into character styles, I have to multiply not only one style to be able to work with different languages. This leads to a huge number of styles that I couldn’t chose via keyboard shortcuts anymore. So I would prefer either a separate language style on a “word” basis or a style that lays above the character style (hierarchy wise the same as a word style but still on a character basis) which I could select independently from the visual Character Style.

I think that’s it.

[gke]

We have been discussing these issues in February of this year, and this is what I posted at that time:

From the point of view of people using languages with different script systems the ideal solution would be to have text automatically tagged as belonging to a language based on the input mode used to type the text. Switching to Russian input mode, for example, tags the text you type as Russian, and this opens the possibility of having this text automatically spell-checked and hyphenated using the appropriate dictionaries, no matter whether this is being done while writing or afterwards. This is the way Word 2004 has gone, and it works pretty well, but of course supports only the languages supported by the programme, which are just a few. Therefore, the challenge would be to come to a similar system which combines invoking both internal application resources (hyphenation) and systemwide resources (spellcheck).


Basically, my position has not changed much since then. What I would like to see is a way of having spelling dictionary, hyphenation, typographer quotes and whatever language-specific sections automatically follow the input-method for the text, with an option to turn this off for those who do not want it, or, even better, to tweak its settings.

This is what I would call a true multi-lingual application - one that allows you to continually change languages and have the benefits of word-processing (automatic spell-check, hyphenation, typographer quotes etc.) automatically follow course, possibly after an initial set-up which allows the user to customize the feature.

[Hans-Reinhard Koch]

Dear All,

I have just reread the posts concerned with multi-lingual spell-checking. The discussion was started in 2004 and ended over half a year ago. The panellists got into the subject rather deeply and there was no consensus on which level the language style(s) should be applied.

To summarize: there were some supporters of a character style language feature, but then one character alone (or a group of characters within a word) is not identifiable as belonging to a specific language.

Others, including Ori, were preferring paragraph-style-based language attributes. This would not allow spell-checking in multilingual paragraphs.

The natural boundary between languages can only be an end-of-word (white space, punctuation mark, CR, etc.). So, indeed, in my understanding language style is something that applies to a level between character and paragraph styles (also see above).

I still think that transalpin's language palette (see earlier this topic) would be a very nice way to allow the different aspects of language selection, although I would not want to attach the language features to the character style.

gke‘s suggestion (see earlier this topic) of coupling the language features to the script will only solve the problem for multi-lingual writers that use languagues with different scripts, e. g. German, Russian and Arabic, but not for different languages of the same (e.g. Roman) script, like in my case English, French and German.

So my suggestion would be to set default language features at the document level (for the dominant language, in which the manuscript is written). These language features can be overridden for selected text. If the selection starts or ends in the middle of a word, it should be expanded to the next word boundary when applying language features. (Because spelling and hyphenation only make sense for whole words, see above.)

There should be an option to apply an „ignore“ language feature to selected text, e. g. when quoting historic texts with an unusual historical orthography. It makes no sense to press ignore again and again in the „spelling...“ dialog during every run of the spell checker. And it would even be worse to let the spell-checker learn these forms of spelling that are definitely wrong in current usage ot the language and thus make it unusable for modern texts.

With my post I want to revive this topic in the forum, and maybe stimulate the Redlers to re-concentrate on spelling. I find it odd that this true (?) multi-lingual word processor should still not offer multi-lingual spell-checking at all.

Hans-Reinhard

[nicka]

Good to see renewed interest in this. It certainly is a key feature for a multi-lingual word processor to get right.

In response to Hans-Reinhard's point I just wanted to contribute a gentle reminder that many writing systems do not mark word boundaries with spaces, punctuation or anything else. The standard orthographies for East Asian languages are mostly this way and old ways of writing some European languages are similar in this respect.
Of course the languages written in this way still have words (despite what native speakers sometimes say) so there's no conceptual bar to having an intermediate 'word' level. But autodetection of word boundaries will not generally be practical for these orthographies, so the user will have to mark up the text manually if different languages are used together.

I also want to second the point that the idea of using a script change to indicate language change is good, but limited. As well as roman letters being used for many languages -- English and other European languages, of course, but also many, many others; other scripts are shared: Japanese, Chinese and Korean share hanzi/kanji, Russian and central Asian languages share Cyrillic. Even very specialised scripts are often used to write more than one language: for example the syllabary used in Taiwan to gloss Mandarin pronunciation can be used to write Taiwanese phonetically. And so on. So any automatic system will need to be overridable, and the overrides must not be lost when changing paragraph or character styling.

[gke]

Let me chime in on this as the author who originally suggested linking language to script - regardless of the script almost any language you would want to write in has its own keyboard layout, even if these are often almost identical (Dutch and US being a case in point). Therefore, one could couple language to selected keyboard layout. This is what Word does, in fact, and it works pretty well, except that it would be much better to reverse the order; selecting a certain language should activate the spelling and keyboard layout for the selected language. This cold be achieved by having a flag menu in Mellel similar to the keyboard layout with the option to assign keyboard shortcuts to the different languages. If this is technically possible, the only group of users left out in the cold would be those typing "foreign" languages using non-standard keyboard layouts, but since this is basically a hack I am not sure how far application developers should go to accommodate for this.

[transalpin]

almost any language you would want to write in has its own keyboard layout, even if these are often almost identical (Dutch and US being a case in point).

Why should I change my keyboard setting for writing in French or English, if I can use my Austrian keyboard layout?

This cold be achieved by having a flag menu in Mellel similar to the keyboard layout with the option to assign keyboard shortcuts to the different languages.

Unfortunately, a lot of people confuse “nationality” and language. The English language is spoken all over the world, and if some web sites attach a UK/US flag to it, it is just bad practice.

[nicka]

selecting a certain language should activate the spelling and keyboard layout for the selected language

This is an excellent way of doing things. It should set the hyphenation language too, in an ideal world.

the only group of users left out in the cold would be those typing "foreign" languages using non-standard keyboard layouts,

Not necessarily. If there is the option in Mellel to define a language style and associate with it whatever keyboard layout you want, then this would not be a problem.
For example, I usually write in English (faute de mieux) but I mostly use either the Dutch keyboard layout or a custom one based on the Dutch one but with a bunch of logical symbols on Option keys. In the ideal world I'm imagining I would simply create a language style in Mellel (or edit the existing one), calling it UK English, giving it the British English dictionary, the English hyphenation dictionary and the Dutch keyboard layout. What could be simpler or more versatile?

Edit: and my suggestion would deal with transalpin's objections, I think. You would just set up different language styles in Mellel, each with the correct dictionary for spelling and for hyphenation, but all with the same keyboard layout. I would do the same: I know how to find an acute accent in the keyboard layout I use all the time, so if I write in French I don't want the French keyboard layout imposed on me -- it would just make life more difficult.
I quite agree about the stupidity of associating a national flag with English, by the way.

[gke]

So it appears this idea of mine is actually more versatile than I thought. Now, of course, my original idea had been to make switching languages as little of an effort as possible. Therefore I had thought of using the standard keyboard shortcuts for switching between keyboard layouts, since this would not constitute an extra efort for those of us using two languages with different scripts and having to change between scripts with Command-Space. For those of us using two languages within the same script range it would of course require an extra keyboard shortcut (Command+Option+Space), but then any other option enabling one to switch from one language style to another would require an extra action, so here nothing would be lost as well.
To wind this message up: if the Redlers brothers woudl think of adopting this solution, I would greatly appreciate it if they could minimize the number of steps necessary to switch between language styles; I often have to switch between Russian and English several times in one sentence when writing and having to go thorugh a lot of actions to get the language style right would impede the writing process too much.

[transalpin]

For those of you who still believe that assigning flags and keyboards to languages “is an excellent way of doing things”:

http://www.w3.org/TR/i18n-html-tech-lan ... .173208643

Do not use flag icons to indicate languages. Use text.
Discussion: Flags represent countries, not languages. Numerous countries use the same language as another country, and numerous countries have more than one official language. Flags don't map onto these permutations.