Are there 2 different parsing engines?

[Zach Wellstood]

Dear FLExperts, 

I'm curious about the parser used in the Parser > Try a Word menu and the parser which parses texts in the Texts & Words tab. 

We're working on a Romance language with gender and typical verb morphology (thematic infinitives, agreement, etc.) In our project, we have made use of Inflection Features to mark nouns & affixes of different genders (fem. and masc.). 

For instance, take the following noun: 

per -- masculine 'hair; pear'  [nagr:[gen:m]]

And the following homophonous suffixes: 

-i -- infinitive [inf:i]

-i -- 2nd person present tense [sbj:[pers:2]]

-i --  masculine plural [nagr:[gen:m]]

Ostensibly, the parser should only match the 3rd -i when presented with the form peri.

In fact, that is what happens when I use the parser under the Parser > Try a Word menu: 

However, there's a different outcome if I put the word peri into the baseline of a text and then click on the Analyze tab: 

As you can see at the bottom, peri remains unanalyzed.

If I segment -i manually (which is not ideal because this require a LOT of manual segmentation), then it still gives me the wrong options: 

1. It is suggesting the -INF suffix first, which makes no sense because the stem is marked with the masculine agreement feature and the -INF suffix is marked with the verbal infinitive feature. There's no featural match, so why's it being parsed this way? 

2. When I click the arrow which lets you pick from the potential analyses, I am given all -i affixes to choose from, even the ones *not* marked with the masculine agreement feature. Shouldn't I only get the affixes marked with the agreement features that match the stem? 

3. Even though the parser under Parser > Try a Word got the word peri completely correct, the parser in the text does not even give a 'best guess' based on the inflection features. 

Is the parser under Try a Word the same parser that's being used in the texts? Is there a way to automate the parsing in the texts so that words like peri would be analyzed (not manually) and a best guess analysis is given based on matching inflection features? If not, then is the only  possible solution to manual segment each noun? 

Thanks! 

Zach 

[Kevin Warfel]

Hi Zach,

 

There are two different parsers, yes, but that is not the source of what you are experiencing. You can choose the parser you want to use by going to Parser, Choose Parser.

 

What you’re seeing happening in the data in the Analyze tab is a result of FLEx giving you information from the lexicon, without any parser smarts applied to it. It will only show you entire morphemes that match your word, or if you break it up manually, it will give you all matches to those pieces, without any application of logic as to what can or cannot co-occur with what else.

 

To get the parser to operate on the word in the Analyze tab, highlight the word and then go to Parser, Parse Current Word or to Parser, Parse Words in Text (if you want to do all the words in the whole text—which could take a while if parsing is complex in your language). I think Parser, Parse Current Word will give you the same results as in Try A Word, but it will add information to the data in Word Analyses, which the use of Try A Word does not.

 

Others who know more about this area of FLEx can correct me if I’ve misrepresented anything.

 

Blessings,

Kevin

--
You are subscribed to the publicly accessible group "FLEx list".
Only members can post but anyone can view messages on the website.
To change your status, please write to flex_d...@sil.org.
---
You received this message because you are subscribed to the Google Groups "FLEx list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flex-list+...@googlegroups.com.
To post to this group, send email to flex...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/CAMvzDNJ-a4-DnobwVGGPobjU_FQf6kTd04Pj-DFbFV2MrRMgMw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

[Zach Wellstood]

Hi Kevin, 

Thanks for your response! I did not know about the Parser > Parse Words in Text function, but now it's working more like the "Try a Word" parser. Great, thank you so much for pointing this out! 

Zach

portfolio · 身先朝露 - 画龙点睛 · 

To unsubscribe from this group and stop receiving emails from it, send an email to flex-list+unsubscribe@googlegroups.com.

To post to this group, send email to flex...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/CAMvzDNJ-a4-DnobwVGGPobjU_FQf6kTd04Pj-DFbFV2MrRMgMw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

You are subscribed to the publicly accessible group "FLEx list".
Only members can post but anyone can view messages on the website.
To change your status, please write to flex_d...@sil.org.
---
You received this message because you are subscribed to the Google Groups "FLEx list" group.

To unsubscribe from this group and stop receiving emails from it, send an email to flex-list+unsubscribe@googlegroups.com.

To post to this group, send email to flex...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/flex-list/9505df3ef51002fb0420534cc631a610%40mail.gmail.com.