FLEx connection to Paratext

Question

This was posted on the FLEx list today. I have a comment I will add below this quoted question:

On Oct 12, 2021, at 3:29 PM, jeffh [Email Removed] wrote:

Can anyone explain the current connection between Paratext andd FLEx? I know I can associate a FLEx project to a Paratext project, and look up words, and I believe I can add words that are not found. There used to be a way to import Scripture texts into the interlinear (in the BTE version of FLEx, which doesn’t exists anymore, I believe). Is that still possible? Thanks for clarifying.

Hi, jeffh.

On one hand, it is still possible to “import” Scripture into the Texts area. (I don’t think of it as an import; I think of it more as a view. But I may be wrong.) It does matter what versions you are using. I think “FLEx 8 works with Paratext 8” and “FLEx 9.x works with Paratext 9.x”. It is not quite that simple–it also matters if you are using 32-bit or 64-bit versions of both.

However, as anon653292 noted on the FLEx list, there are bugs:

On Oct 12, 2021, at 5:13 PM, anon653292 [Email Removed] wrote:

Hi jeffh,

Beside the dictionary look-up feature, Paratext will add glosses to your dictionary when you use the the interlinearizer: whenever you add a new gloss to a morpheme in the PT interlinearizer it will add add it to your FLEx database.

In principle this is a great feature, however, it is flawed, so much that I don’t add words to the interlinearizer when I have a FLEx project associated. FLEx will add glosses as a new sense often to the wrong homonym. It does not ask for confirmation. Sometimes it adds a new sense even if a homonym with that sense already exists. The worst part is that Paratext leaves no trace, it does not update the date stamp, so you can’t search for entries that have been changed while you were working in Paratext.

I have submitted this issue to the Paratext developers and am eagerly looking forward to the day it is fixed.

I think you can still (manually) import your Bible translation into the text area of FLEx and interlinearizer it there. Haven’t done it in a while.

Someone else I know is on a team where others do the Paratext interlinear, and he is the one who wants to work in FLEx. They found that FLEx can also write to Paratext, and mess that side up as well. But he found a workaround.

His approach is:

• Normally the FLEx and Paratext are not associated.
• When he wants to get the latest Paratext data into FLEx, he turns on the association.
• He does his work in FLEx.
• Then he turns off the association, before he does a S/R.

His experience is that if he (a) doesn’t use the Paratext interlinear and (b) doesn’t do a S/R while the two are associated, then he can get away without messing up either his own FLEx database or the Paratext interlinear work of the others on his team.

I am interested to hear from anyone else:

• Is anyone using this functionality as is, and you would “lose data” if they changed how it works (that is, make it so that FLEx and Paratext no longer write to each other’s interlinear databases).
• How many want to use this feature, but they are avoiding using it because of this bug?

Answer 1

I was actively using the Flex interlinear for Scr text a few years ago. I found a lot of bugs, but most/all of them got fixed. I also use the dictionary-lookup feature from PT all the time. I tried the PT interlineariser earlier this year, and it seems that there is some good potential with the FLEx integration, but there are several issues that make it unusable for my project. Here are some points that I wrote up a few months ago:

1. PT Interlinear
   I’ve discovered that the PT IL now reads FLEx entries and shows their glosses as suggestions, and it also writes new senses/entries into Flex, but there are a bunch of issues:

   • a bug: for homographs it only shows glosses for the first lexical entry.
   • in the word parses line, adding a new gloss or editing one creates a new entry. Two problems: firstly, it might need to be a new sense in an existing entry (but how does the user indicate/control that?; and for homographs, which entry it should be added to?); secondly, I wanted to edit the morpheme gloss sometimes, but that would create a new entry (or sense if it was working properly), but I don’t want that–just an adjusted gloss in this context. As a result of these (and #2 below) the morpheme line was unusable for me, so I turned it off.

2. Flex & morphology
   My language has complex morphology with many allomorphs, vowel-harmony and other rules. I have most of that modelled in Flex, but PT uses its own parser and it doesn’t work with my FLEx entries. For example I have suffix allomorphs, which means that the form in the PT text doesn’t match the Flex entry. If I add a gloss, then PT creates a new FLEx entry for the surface form, which again isn’t what I want.

I’m not sure if there is any workable solution to this set of problems; basically I need to use either the PT IL without word parses, or stick with FLEx. I guess it needs to be made clear in the Help and other places (e.g. tutorial videos) what range of situations/languages this tool is suited to.

I haven’t seen the Interlinear-FLEx integration promoted, and the issues above make it seem like it is still experimental/in-development. What is the status?

Overall I see some good potential in this tool, especially for languages without complex morphology (And where the word-order matches; the export of English glosses to the BT project is pretty useless for our SOV language.) If the hope is for it to be used by a wider user-base then there are some significant things that need to be improved in the design.

Answer 2

Thank you, anon835088.
I do hope and pray that this issue will be fixed in the next major update of PT.

Answer 3

I haven’t used the interlinearizer because of this issue, but would probably try it out again if the issues could be resolved.

Comments

I use the interlinearizer in PT primarily because it is the only way to indicate a word has more than one valid morphological analysis (AFAIK). I liked having PT and Flex linked just to be able to look up words in Flex from PT but I have removed the link and kept it that way since this issue came up.

JG

Answer 4

Thanks @anon835088 for posting this question in this forum. I should have checked here before posting on the FLEx list, as I see that there are already a couple of other threads:

How do FLEx 9 and PT 9 work together for interlinearization? Questions

We have done a lot of PT interlinearization in our project, done expressly with the goal of helping advisors’ and consultants’ understand the language and to reduce the automatic parsing load for PT. We have done much more precise and linguistically oriented interlinearization on non-Scripture data in FLEx. We have had our FLEx database associated with the PT project for several years, but we have not attempted to interlinearize Scripture in FLEx. I have seen evidence that PT9 interlinearizer i…

How does the Paratext Interlinearizer and Flex parser work together Questions

Reposted from @KimB Thank you, jeffh. We have been using Match based on stems since before PT9. My problem is not how to do interlinearizing in PT; it is that PT now seems to be sharing its existing morphology and gloss data with FLEx—bidirectionally. Which is definitely not what we want, because we have been deliberately (and systematically) underspecifying morpheme breaks in PT to avoid bogging PT down, make it faster to do, and provide glossing that is more understandable by translation c…

I was asking for someone else anyway, so I will point them to these responses. Thanks.

Answer 5

One person replied to me directly with this helpful information:

To pull in texts from Paratext, when you’re in the “Texts & Words” pane in FLEx, there’s a small icon at the top that says “Choose Texts” when you hover over it. You just click that, and it opens up a window with a tree of texts to choose from, including the Old Testament and the New Testament, and you can select which books you want to pull in from Paratext. Just click OK, and it will pull in your texts from Paratext. It’s that easy–you don’t have to do anything special to make the texts from Paratext available to FLEx, somehow FLEx already knows about them!

The Choose Texts icon looks like this:

Note that you have to make the association to the FLEx project from the Paratext project properties (Associations tab) for the Paratext texts to show up there.

Comments

The ‘Choose texts’ button in FLEX works for me, but selecting large texts can bog the system down. Don’t choose the entire NT–like I did the first time! A chapter at a time seems like enough to work through anyway. Also, be aware that once you deselect the Bible text from the ‘Choose texts…’ list, the scriptures disappear from the list of interlinear texts in FLEX.

Here are some issues that I have encountered:

1. We use some backslash codes in our translation for formatting and conversion of proper nouns. The import process has problems when it encounters one of these markers–the name (and sometimes the rest of the verse) doesn’t appear in FLEX.
2. The Back Translation line also runs into issues, so most of the free translation lines in FLEX don’t contain the full BT.

Answer 6

I’m late in responding to this question, anon835088, but it is a very important question!

With our specific project work flow, we have used very different strategies when interlinearizing drafted Scripture in PT, vs texts for linguistic analysis in FLEx. Our PT interlinear is mainly done to help non-MT advisors and consultants to understand words and basic morphology in the language for purposes of consulting with the translation team. Thus it has been much less technical than our FLEx interlinear morphology and glossing: We have conflated some verbal suffix strings into single units in PT, for instance, to decrease multiple glosses for short homophonous morphemes.

In addition, different users make different decisions when interlinearizing; we have tried to keep the FLEx database as consistent as possible but we haven’t worried as much about PT’s because it is more temporarily needed. So we have been avoiding interlinearizing in PT since we discovered that PT was messing up our FLEx wordforms.

But we use the PT word lookup in dictionary every day since a Rapid Word Collection workshop gave us 10,000 lexicon entries. We would be very grateful and relieved if the Dictionary word lookup functions in PT were split from the Interlinearizer function, with ability to choose between FLEx-based and PT-only interlinearizing.

Thank you!