Easy way to detect undefined phonemes in FLEx project?

[Kevin Warfel]

Now that there seem to be at least a few FLEx users who have experimented with the Hermit Crab parser, maybe one of you can answer this question for me.  Is there an easy way to know whether the Phoneme inventory in a FLEx project is exhaustive – that is, it includes all of the symbols (letters, hyphen, apostrophe, etc.) that are used in the orthography?  Phrased another way, is there an easy way to know if there is a symbol that occurs somewhere in your lexicon, texts, etc. that is not in your list of Phonemes?  I thought this would show up in the Phoneme section of the Grammar Sketch, but it doesn’t seem to, at least not in the project I’m trying to help someone with.

 

Thanks for any ideas or suggestions you might have!

Kevin

 

[Snofriacus]

One thing that could help is to set up Phonology Assistant (PA) to read your FLEx data. It won't be able to see your texts and wordform inventory, but will read all your lexeme forms and make a list of all the characters. Note that this is PA's own list - no automatic connection to FLEx's list of phonemes. But maybe it wouldn't be too hard to then manually check that everything in PA's list has been accounted for in FLEx's phoneme list.

This brings to mind several nice things that probably could be done, and which I'd like to eventually see take shape in FLEx.

• Bring PA or its equivalent into FLEx so they can be better integrated. It would make good sense then for FLEx's PA function to have the task of facilitating the creation of FLEx's phoneme list.
• Implement phonological rules, and let the rules be tested by using them to map from phonetic to phonemic forms. This is a nice piece of functionality we had in LinguaLinks which hasn't yet been reproduced in FLEx.
  
• Allow the PA function to take its data from the wordform inventory. It doesn't need to look directly at the texts. FLEx already automatically manages the wordform inventory so that whenever something is added or changed in a text, those things are reflected in the wordform inventory. So in essence, by looking at the wordform inventory, PA -is- looking at the texts. In my thinking, to have PA looking at lexeme forms is a bit less useful, because at least for the language type I work with, the forms of lexemes are sometimes abstractions that are subject to an analyst's opinion. The wordform inventory better represents the pure expressions of the language as they come from the speakers' mouths.
• Similar sentiments regarding ability to record wordforms and transcribe their pronunciation forms. These capabilities need to exist on the wordforms of the wordform inventory, since this is the data which most consistently corresponds to spoken expressions.

Allan

[Todd Hoatson]

On 5/9/2012 4:26 AM, Kevin Warfel wrote:

 

... Is there an easy way to know whether the Phoneme inventory in a FLEx project is exhaustive – that is, it includes all of the symbols (letters, hyphen, apostrophe, etc.) that are used in the orthography?  Phrased another way, is there an easy way to know if there is a symbol that occurs somewhere in your lexicon, texts, etc. that is not in your list of Phonemes?  I thought this would show up in the Phoneme section of the Grammar Sketch...

 

This would not only be helpful, but important (speaking more for the teams I oversee than for myself, since I don't usually get to do much first-hand linguistic work).

On 5/9/2012 8:53 AM, Allan (Snofriacus) wrote:

 

...

This brings to mind several nice things that probably could be done, and which I'd like to eventually see take shape in FLEx.

• Bring PA or its equivalent into FLEx so they can be better integrated. It would make good sense then for FLEx's PA function to have the task of facilitating the creation of FLEx's phoneme list.
• Implement phonological rules, and let the rules be tested by using them to map from phonetic to phonemic forms. This is a nice piece of functionality we had in LinguaLinks which hasn't yet been reproduced in FLEx.
• Allow the PA function to take its data from the wordform inventory. It doesn't need to look directly at the texts. FLEx already automatically manages the wordform inventory so that whenever something is added or changed in a text, those things are reflected in the wordform inventory. So in essence, by looking at the wordform inventory, PA -is- looking at the texts. In my thinking, to have PA looking at lexeme forms is a bit less useful, because at least for the language type I work with, the forms of lexemes are sometimes abstractions that are subject to an analyst's opinion. The wordform inventory better represents the pure expressions of the language as they come from the speakers' mouths.
• Similar sentiments regarding ability to record wordforms and transcribe their pronunciation forms. These capabilities need to exist on the wordforms of the wordform inventory, since this is the data which most consistently corresponds to spoken expressions.

I whole-heartedly agree with all of this.  This would be way-cool, but also would greatly simplify & enhance the jobs of both field teams & trainers.  I give it all my votes!

Todd

[Kevin Warfel]

On Tuesday, May 08, 2012 9:53 PM, Snofriacus wrote:

• Implement phonological rules, and let the rules be tested by using them to map from phonetic to phonemic forms. This is a nice piece of functionality we had in LinguaLinks which hasn't yet been reproduced in FLEx.

Thanks for your response, Allan, but I want to make sure that one thing is clear, both in your mind and in the minds of all others who are participating passively in this conversation.  The term “phonological rules” in FLEx is a bit misleading IMO (I prefer to think of them as “morphophonological rules”), as they do not map from phonetic to phonemic forms, but from phonemic to orthographic forms.  They take as input the lexeme forms (as entered in that field in the lexicon) and produce as output the form used in the orthography.  The rules are applied sequentially, so that the first rule takes its input directly from the lexicon, while each subsequent rule takes as its input the output of the previous rule.  The HC parser uses these rules, when parsing a given word, to find candidate combinations of lexemes, then applies the rules to each of those candidate combinations to verify (or not) that the application of the phonological rules to that particular set of lexemes is a) legitimate from a grammatical perspective, and b) correct, in terms of the output of the application of the rules being identical to the orthographic form being parsed.

Maybe you already understood this, but some other readers of what you wrote may not, so I thought it worth clarifying.

Blessings,
Kevin

[Craig Farrow]

Hi Kevin,

You asked this quite a while ago.... I've put up some information on how
to use the stand-alone HC parser to find mismatches between the Phoneme
inventory and what is in your lexicon. This doesn't help with invalid
phonemes in Texts, though.

http://craigstips.wikispaces.com/HermitCrab+Parser

Craig.

9/05/2012 4:26 a.m. dï, Kevin Warfel pišdimiš:

> --
> You received this message because you are subscribed to the discussion
> group "FLEx list". This group is hosted by Google Groups and is open
> for anyone to browse.
> To post to this group, send email to flex...@googlegroups.com
> To unsubscribe from this group, send email to
> flex-list+...@googlegroups.com
> For more options, visit this group at
> http://groups.google.com/group/flex-list

[Kevin Warfel]

Thanks, Craig. I think that this tool will be helpful to me in the future.

When I encounter a word in my texts that the HC parser is unable to analyze, the first thing I check is whether or not all the morphemes that make up the word are actually in my lexicon. If they're not, I add them in. Once all of the morphemes are in the lexicon, if the word still can't be successfully (and correctly) parsed, I could then use your tool to verify that the problem is not with undefined/undeclared phonemes.

Thanks again,
Kevin