Chris,

Yeah, this was a bad example. In fact, most of the actual <w> nesting 
situations in the KJV2003 are bad examples. I know just enough Greek to 
fix most (I'm rusty as it has been 25+ years since I studied Greek:) and 
maybe all.

The worst bug of this kind is in Luke 15.24.

Troy will probably have to confirm, but it appears that the src 
attribute is being used to indicate the position of the Greek word in a 
particular module. If so, it would be good to be able to construct an 
actual osisRef for it. While cp and s are defined as grain operators it 
would be good to also have wp (i.e. word position), too.

The <w> element lets one indicate that two or more words are being 
translated as in
<w lemma="s:1 s:2">word</w>.
Likewise, it is possible to have more than on morph as in
<w morph="r:9 r:10">word</w>
But in combining these,
<w lemma="s:1 s:2" morph="r:9 r:10">word</w>
it is not possible to know for certain, but only by convention that s:1 
maps to r:9 and s:2 maps to r:10.
Adding src to it makes the mapping even more difficult.
<w src="7,9" lemma="s:1 s:2" morph="r:9 r:10">word</w>
It appears that nesting is being used to indicate uniquely the origin.
<w src="7" lemma="s:2" morph="r:9"><w src="9" lemma="s:1" 
morph="r:10>word</w></w>

Is there any value in such a nesting construct? (Personally, I don't 
think so.)

DM

Chris Little wrote:[...][...]

Chris Little wrote:[...][...][...]

I think it is correct and also valid. In many languages, gender is 
communicated by context, but in translating to English it may be 
rendered explicitly. In the KJV these instances are indicated by italics.

Earlier today I saw "give *him* drink" where "give drink" was one word 
in Greek and *him* was added for clarity.

So I suggest that for OSIS 2.5 that transChange be allowed to be a child 
of <w>.[...]
Unbelievable as it may be, the Strong's entry is:
4155 pnigo pnee'-go  strengthened from 4154; to wheeze, i.e. (causative, 
by implication) to throttle or strangle (drown):--choke, take by the 
throat. see GREEK for 4154
[...]

I don't think I like splitting the word into a before and an after. This 
might communicate to a casual user of the document that 4155 can be 
translated as either "took" or "by the throat".
[...]

I like having explicit markup as it is easier to transform later.

DM Smith wrote:[...]

Yeah, src on <w> in CrossWire's KJV refers to the nth word in the same 
verse within the Byz module. I don't think this what what this attribute 
was intended for, but I don't have a better suggestion of an attribute 
on which to hang such a value. Technically, you'd want a token 
attribute, which would require either extension or modification of OSIS.

We discussed a word grain, but decided against it. I recall the 
ambiguity of "word" as one of the issues. More useful would probably be 
a document that indicates osisIDs down to the word level (a word-level 
tagged Byz, in other words--which we may already have).
[...]

No. OSIS is already being used in CrossWire's KJV in ways other than 
what was intended. The lemma and morph attributes indicate the lemma and 
morphology of their content, but the CrossWire KJV uses them to indicate 
the lemma and morphology of words in a completely different document. 
(We have an English language document with lemmatization and morphology 
information for a Greek language document.)

If the English text does not distinguish two Greek language tokens, such 
that they must be translated using a single English token (or vice 
versa), then there is no reason to disambiguate which lemma is 
associated with which morphological information in the translation text. 
I think it is simply not relevant information, from the perspective of a 
KJV-reader.

--Chris

Chris Little wrote:[...]

Can you recommend a better way? I would like to create the best OSIS for 
the KJV and use xslt to make it acceptable to the SWORD api.

I was thinking that POS is a better place for robinson's part of speech 
codes.

Or is this a "future" for OSIS?

DM Smith wrote:[...][...]
>>> The following construct if fairly frequent:
>>> <w src="19" lemma="strong:G4155"
morph="robinson:V-IAI-3S">and took 
>>> <transChange type="added">him</transChange> by the
throat</w>
>>>
>>> The problem is that <w> does not allow for
<transChange>.
>>> Should this be changed to <seg type="x-transChange" 
>>> subType="added">him</seg>?
>>> (This is the form that is used within study notes.)[...][...]

Based on the morphology code, it's person and number that are specified 
by morphology on the verb (which causes the rendering with "him" in the 
English). It wasn't the "him" I had a problem with. Greek is just a 
pro-drop language (one of the approximately 10 things I can name about 
Greek :). What I had a problem with was a verb meaning "to take by the 
throat", but it looks like (based on the Strong's definition you gave) 
that's just a metaphorical extension to the meaning.
[...]

I'm inclined to agree, unless anyone can come up with a good reason not to.
[...][...]

The casual user should (and probably did) turn Strong's numbers off.

--Chris

DM Smith wrote:[...][...][...]

I can't. It's just isn't a normal thing to want to markup (morphology 
from some other text) and I can't see it ever being a common enough 
issue to warrant special handing in OSIS (like a new attribute).
[...]

Robinson's codes identify more than just part of speech. POS shouldn't 
indicate case/number/person agreement or tense/aspect/mood/modality. 
Those are the domain of the morph attribute. (NB: Again, I don't know 
much about Greek, so if Greek doesn't have some portion of those 
attributes, obviously Robinson's doesn't either, but I know it presents 
a fairly complete description each word's morphology.)

--Chris