Possible LN.org Application: Grammar Annotations of Texts

[wm.annis]

I have been playing around with a text annotation tool with the terrible name BRAT (Brat Rapid Annotation Tool).  It is easy to install, so I put a copy on my server and have been configuring it to know about Na'vi grammar.

I have just started in on annotation some sentences from the last blog post: http://lingweenie.org/brat/index.xhtml#/Navi/Naviteri-2012-06-19 — take a moment to put your mouse over  the little boxes above the words.

I still have a little tinkering to do.  But the main question is, does anyone think they'd actually use this tool?  Right now, it seems like it would be of most benefit to beginners, to explain how things relate.  If we think people would find it generally useful, I will refine the configurations for Na'vi grammar, and we can install it on the LN.org servers.  If not, there's no reason to install a tool no one will look at.

[Sìkat]

As a beginner, I'd love to give this a spin

[Irtaviš Ačankif]

So it automagically understands Na'vi sentences and analyzes them?

[wm.annis]

So it automagically understands Na'vi sentences and analyzes them?

No.  Someone has to do that.  The process of annotation is surprisingly straightforward for a web application.  It just takes a little care to set up the configuration with the correct sorts of annotations.

[Plumps]

Not a beginner, but I think this can definitely help beginners in understanding how words are related in Na'vi

[Ngawng]

Very impressive and nifty showing how the words are related to each other. Just wow

[Seze Mune]

I'm not too lucid about how these things work, computermagically speaking, but would there be a way for us to parse our own sentences in this format, or would this be for a static set of sentences?

Regardless, yes, I would use it, and be very grateful for it too.

[Meuiama Tsamsiyu (Toruk Makto)]

I would most definitely use something like this. I am stuck at a hurdle where I use words learned and repeated but cannot form actual sentences on my own as I forget words or infixes. Thank you for trying new things for us beginners.

[wm.annis]

Very impressive and nifty showing how the words are related to each other. Just wow

Well, it is very nifty, and it looks cool.  The real question is, does it actually help people's understanding?  For a complete beginner it's probably overwhelming, and advanced Na'vi speakers probably don't need it.

I'm not too lucid about how these things work, computermagically speaking, but would there be a way for us to parse our own sentences in this format, or would this be for a static set of sentences?

Regardless, yes, I would use it, and be very grateful for it too.

The application has a very poor security model.  Other people could use it once they got an account (which happens by editing a configuration file).   Everyone with an account as full access to the system, which is a little blunt.  But, texts can be added either directly on the server, or by uploading through the application itself if you have an account.

[wm.annis]

Another text to look at: http://lingweenie.org/brat/index.xhtml#/Navi/A_Message_From_Paul

[Yawne Zize’ite]

I didn't think this would help me much, since I'm a beginner overall but my skills are wildly skewed towards morphology and syntax (pragmatics, not so much). However, looking over this latest sample, I realize that I was relying heavily on Naʼvi's slight bias towards right-branching clauses and had some trouble parsing a relative clause before its referent!

One request: is it possible within this framework to mark stress in some way? I'm using the tool for vocabulary more than anything else, and the stress isn't written. The languages I've studied before have predictable or weak stress, so I don't have an easy time remembering which syllables are stressed in Naʼvi.

[wm.annis]

One request: is it possible within this framework to mark stress in some way?

I'm afraid not.  For starting input, it just wants text with no other markup.

[Blue Elf]

Well, this idea is at least very interesting. Although I'm not beginner (or at least I don't feel to be one ), I think it can be useful. The best usage IMO is to create web application where you can enter your own sentence and get its analysis. I think Tirea Aean prepares something like this in new version of his great vrrtepcli tool.

[wm.annis]

The best usage IMO is to create web application where you can enter your own sentence and get its analysis.

That would indeed be very useful, but so far no computer is capable of performing that task reliably.  There are tools that manage a first approximation, but that's the last thing to give to beginners.

[Irtaviš Ačankif]

Except the fact that Na'vi is a really regular constructed language 

[wm.annis]

Except the fact that Na'vi is a really regular constructed language 

It's not that regular.  If nothing else, the nonconfigurational word order will cause mayhem for software that doesn't understand a great deal of semantics.

[Irtaviš Ačankif]

But the case markings and the small vocabulary...

I actually wrote a C# program once that reordered any Na'vi simple sentence into SVO word order

[`Eylan Ayfalulukanä]

Except the fact that Na'vi is a really regular constructed language 

I think Yawne Zize'ite would disagree with you on that!

I think this is a wonderful tool, and for a really important reason: Na'vi makes extensive use of forms that differ on verb transitivity. With free word ourder, it experiences left- and right- branching attribution. For complex sentences, a tool like this would be wonderful for teaching these confucing concepts.

Has Prrton seen this?

Txantsan, ma oeyä 'eylan!

[wm.annis]

Has Prrton seen this?

He has.

I'm still not 100% convinced of its value as a teaching tool.  Maybe for a few people, but probably not widely.

[Sìkat]

wm.annis, could you post the configuration you use for brat (minus anything obviously sensitive)?  I'd like to install brat on my own server and play around with it some.