Paratext and Flex use the same parser

Question

It has been in the pipeline to further integrate Paratext and Fieldworks by having them use the same parser. I would like to see this happen.

docs.google.

FLEx-ParaTExt Text and Words sharing - Round 4

Project Name: FLEx-ParaTExt Text and Words sharing - Round 4 Submitted by: Beth Bryson Submitter’s email: [Email Removed] Champion: Colin Suggett, David Coward Champion’s email: Document Navigation Worksheet Elaboration of details Project...

Publicación antigua - se muestra en su idioma original

Answer 1

I assume that the implied information in this thread’s title is that not only the parser is the same, but also the dictionary, word list, spelling conventions etc. Them NOT being the same was the price we all paid for leaving the old SIL Scripture Editor behind, but that’s water over the dam. OK developers, here’s a shot in the dark from an out-of-practice programmer in his former life (think punched cards). Can you read the Flex data base, and pull all the bits and pieces of lexical data from Flex that it would take to create an exact imitation of the data structures which ParaText uses natively? If so, could you build such a routine into Paratext and then let users choose whether to a) use the native Paratext data structures, or b) read them from Flex and use the imitation structures? This would allow the rest of Paratext’s source code to be remain unchanged, and people with no interest in Flex (which is probably most translators) would never be any the wiser.

Publicación antigua - se muestra en su idioma original

Comments

mvk1000:

Can you read the Flex data base, and pull all the bits and pieces of lexical data from Flex that it would take to create an exact imitation of the data structures which ParaText uses natively?

Almost anything is technically possible.
It’s mostly a question of the cost to benefit ratio. Reading in the data from FLEx to produce data in Paratext would not buy you much because Paratext couldn’t use most of the data that FLEx would provide. Paratext’s lexical model is very simple. I would imagine that people using FLEx to do lexical analysis really want Paratext to use all the data that FLEx has available to produce better results in Paratext (e.g. better spell checking or better interlinearization). Without being able to handle that extra data, Paratext wouldn’t do any better than what it does now, so the added complication of getting data from FLEx wouldn’t be worth it from a cost-benefit perspective without implementing the other pieces needed to handle the data.

Publicación antigua - se muestra en su idioma original

---

Actually it would buy a LOT and the cost benefit ration would be huge because I wouldn’t have to approve the parsing and spelling of the ten-thousand word forms I have TWICE, never mind the almost impossible task of subsequently keeping the two sets of data consistent over a couple decades (one down one to go) as spelling conventions evolve. Truth be told, you are right, I don’t need most of what Flex can provide, but it just kills me to have to do all the same lexical work in Flex for the non-Biblical corpus that I have already done once in Paratext for the Biblical corpus. I might mention that this conversation got started on another thread of this forum under the rubric of “Extra ‘Extra’ books”. As I mentioned there, I have for all intents and purposed abandoned Flex and am now trying to cram the entire non-biblical corpus into the XX books of Paratext. I have a suspicion that is not what the XX books were designed for.

Publicación antigua - se muestra en su idioma original

Answer 2

I think this is a wonderful idea!!! I really hope to see this become a reality.

Publicación antigua - se muestra en su idioma original

Answer 3

I was speaking with a translation consultant that works in Angola and other parts of Southern Africa, and he was very interested in a unified parser for various reasons. One is that he wants as many teams as are able to do discourse analysis using Fieldworks.

Publicación antigua - se muestra en su idioma original

Answer 4

@mvk1000 I think you are hinting that you like how easy it is in Paratext to mark spelling status and to divide words into morphemes. I am praying that if the Paratext and Flex ever do have a unified parser, that there would be a mode that is as easy to use as Paratext is today. Not everyone wants to describe all allomorphs and their environments nor define word classes and their morphology. There could be an “linguistics” mode that would enable all that power of Flex for those who want to take their analysis to that level.

Publicación antigua - se muestra en su idioma original