by Michael S. Kaplan, published on 2007/09/01 17:01 -04:00, original URI: http://blogs.msdn.com/b/michkap/archive/2007/09/01/4693025.aspx
It was with great interest that I read Mike Plumpe's blog post on speech @ microsoft entitled Windows Speech Recognition language support in Windows Vista.
Although I admit the content managed to inspire dismay when I read this bit:
In Windows Vista, Windows Speech Recognition works in the current language of the OS. That means that in order to use another language for speech recognition, you have to have the appropriate language pack installed. Language packs are available as free downloads through Windows Update for the Ultimate and Enterprise versions of Vista. Once you have the language installed, you’ll need to change the display language of the OS to the language you want to use. Both of these are options on the “Regional and Language Options” control panel. You can look in help for “Install a display language” or “Change the display language”.
By tying language specific speech recognition to user interface language display packs, the Speech folks essentially confuse internationalization with localization and localizability, distinctions which I have talked about previously (in posts like Its not localization, really and internationalization vs. localizability) pointed out can be crucial.
Consider the following:
In what way is speech recognition different from any of the above, especially that last point?
I know, I know, I am not completely naive -- this is another attempt at SKU differentiation, something I talked about in Additional personal speculation on the Vista MUI SKU story.
But in my own personal opinion (which I'll keep stating until the legal and PR folks tell me stop) is that this is not the correct way to do the split.
Consider all of the people out there whose lists of languages they can read, languages they can write, and languages they can speak really can vary. Why on earth would one kind of input method (keyboards) be "universal" for everyone while another (speech) not be available except to those who buy the Ultimate version of Vista?
And according to Mike's post, it is not just an installation issue; they are truly tying the functionality to the user interface language -- which makes the bold claim to customers that they should not dare to speak a language before they plan to be able to read it.
As a feature, it should not be based on either user interface language or user locale -- this is entirely an input language thing, and tying it to anything else violates not just the common sense issues I am talking about but the whole architecture of input selection inherent in the Text Services Framework!
And that is ignoring the rather bitter pill that due to that Error 2 bug I mentioned which keeps me from being able to install even the language packs that have been offered. I have currently downloaded over 10gb of language pack content in repeated attempt to address this bug using various steps people have put online, and the fact that I miss out on yet another cool feature that I should not be required to depend on is really annoying....
Of course, the battle for common sense and appropriate SKU definition is hard enough to be involved with when it is related to features in your building, let alone when it is about features that are across the street or even farther away.
If you want differentiate SKUs and features, fine.
Hell, if you truly want to limit it Vista Ultimate then that's not so fine but I'll live with it and just gripe a bit about the cases where people might be willing to shell out a bit of money for a feature where MS wants to say no thank you.
But the language recognition feature of speech and the MUI feature of allowing one to change one's user interface language are completely separate and lumping them together violates the important distinction between internationalization and localization....
This post brought to you by ᓔ (U+14d4, a.k.a. CANADIAN SYLLABICS LAAI)
# MichaelGiagnocavo on 1 Sep 2007 5:52 PM:
Is it my poor memory, or could you install speech recognition in other languages in XP? I remember changing IMEs and doing handwriting recognition... I think I remember something where speech was enabled for other languages...
# Rosyna on 1 Sep 2007 7:14 PM:
Why's it matter? Everyone in the world should just speak and write in broken, fractured English like all Americans.
# Michael S. Kaplan on 1 Sep 2007 11:13 PM:
Hello Michael,
That is my recollection too. And as far as I can tell it is not the TSF architecture that has changed at all, it is just where they are packaging things on the shared speech side in Vista, and how they are making some of the switching functionality to work....
# Michael S. Kaplan on 1 Sep 2007 11:14 PM:
Hey Rosyna,
Well, just in case anyone agrees with me that such a plan would be bad, I'll keep trying to tilt windmills here. :-)
# Johannes Roessel on 2 Sep 2007 7:59 AM:
I also stumbled over the point that speech recognition only works for the installed user interface language. Working here with Vista Business kinda sticks me to English which I don't mind having as an interface language but most content I write is naturally in German. But changing interface languages (and therefore become accustomed to new keyboard shortcuts again) to use speech recognition in my own language kinda eludes me. I don't see a point in doing so and I'm sure folks out there being bilingual and using both languages quite often are very happy about changing interface languages every time ...
Granted, the speech recognition might pay some more attention for controlling windows and the like. When I want to switch to a window with a German title I have to say "switch to" and then thy to pronounce the window title in English (which may be easy for native English speakers but I fail miserably and so does speech recognition :). Being able to use any language you want because menus, window titles and the like are not in the language speech recognition is recognizing right now is surely not an easy problem to solve but even without tying speech recognition to MUI this problem is there.
# Mike on 2 Sep 2007 10:07 AM:
For years Microsoft.com has labelled most of ithe Office proofing tool products as either being for US English, or tying them to the interface language. Of course the point is the same as yours: that all of these tools are available to be used regardless of the UI language.
During the Vista beta, one of the triage PMs seriously told me that I should only expect UK language tools to work on the "localized" UK-English version of Vista (which of course does not exist). I despair.
# Adrian Posor on 18 May 2008 6:42 AM:
From my point of view I understand why the language for speech recognition and the UI have to match. It is surely not relevant for dictation, but if you control the computer via your voice, the speech recognition cannot understand some of your commands if the text in the menu and so on does not match the language the recognition sofware is currently adjusted to understand. Nevertheless, it would be nice to be able to dictate in another language than the language used to control the computer. This problem could be solved with a greater seperation of control and dictation mode.
referenced by