Complex searching

[Ken Manson]

Hi

I am wanting to search my interlinearised texts for a sequence in the text
line that occurs after a certain part-of-speech (e.g. la-blan 'one-time'
occurring after a verb). The language I am working on is highly isolating
and has numerous homophones, making just a search for the particular text
sequence not very helpful.

Can this be done in Toolbox, or do users have other suggested
programs/procedures?

Many thanks
Ken

[ToolboxSupport]

Dear Ken,

I was hoping someone would come up with a brilliant suggestion... I'm
afraidI don't have one either. I will talk with the programmer about
this situation; I'm sure you aren't the only one who would like to do
something like this.

One question: Do you have your data with a whole text per record (as
we've recommended) or with one sentence/clause/whatever per record?

Toolbox Support

[D. Rowe]

Ken:

> I am wanting to search my interlinearised texts for a sequence in the text
> line that occurs after a certain part-of-speech (e.g. la-blan 'one-time'
> occurring after a verb).

Could you post a few examples? For example, two examples that have the
sequence where it occurs in the context you want and two that have the
sequence in another context.

Perhaps there is a way to define a filter that matches the text (in one
field) and the meaning (in the gloss field).

David Rowe

[Susanna]

Hi Ken,
In FLEx (FieldWorks Language Explorer), it is possible to do a
concordance on a specific homophone. In interlinear views, right-click
on the word in focus to see a menu which includes various concordance
options. There are different options depending which interlinear line
you right click on. These options provide jumps to a specific
concordance of the item chosen, rather than just a concordance on the
text sequence. These jump commands have also been provided in the
Lexicon, e.g. providing a concordance of a specific entry or morph not
just the text sequence.
FLEx doesn't yet have a way for the user to state a sequence to the
concordance, e.g. a certain text sequence coming after any verb.
However, it could be that the specific concordance options FLEx
provides would meet your current need.
FLEx doesn't yet have a way to import interlinear texts, which would
make it very difficult for you to make use of these features at this
time, I suspect.
See http://groups.google.com/group/flex-list for more questions/
answers/discussions of FLEx.
-Susanna

[Hiram Ring]

Hi Ken,

  I was recently wondering if ToolBox has the functionality of being able to search multiple fields at the same time, and came across your post.

  I am aware of the concordance feature and use it quite a bit, but like you wanted to be able to find, say every morpheme in the 'tx' field with the form ka that is followed by any morpheme with the part of speech ('ps') label n.

  I can currently do this kind of search using the ToolBox Search library for R, which is also extremely powerful for statistical counts and such (seeing as it uses the R framework). You might look into it - if you don't mind learning a bit of R coding, this ToolBox Search library can turn your existing ToolBox text files into an annotated corpus that is highly searchable. This tool is really improving my analysis of the language I'm working on, though I am having to re-interlinearize my corpus to ensure consistency. But you have to do that when you change your analysis of a morpheme too, unless you can find a way to find/replace items in the text file.

  I'm attaching here a short R script with annotation, which I developed from reading through the PDF manual that T. Zakharko has made available on the downloads page via the link above. Make sure to email him if you have problems with installation - the first packages I downloaded wouldn't work with the 3.x version of R, but the current ones I have do.

  What would be great is if somehow we could incorporate this into Toolbox or FLEX. Or maybe incorporate something like the Redet tool...

Hiram

[ToolboxSupport]

Complex search of this nature has been wanted for a long time, and we are sorry it has

not been implemented in Toolbox. It may happen at some point, but is not currently scheduled.

Some people use external tools. We appreciate your reference to one such tool.

One thing that can help with complex search is that XML export puts morphemes in groups

with their annotations. Save As XML does not do that.

--Toolbox support