I understand SuStel's distinction between sintactic and semantic roles, and I find it a really exciting topic, but since in the thread I'm referring to "direct object" and "indirect object" are used, I will also use these terms here.

And that's fine, so long as we understand that direct object and indirect object are semantic, not syntactic, roles in Klingon.

2. However, I don't understand why *Holmey law' lujatlh* couldn't be an instance of the "prefix trick" (I'm referring to De'vID's answer here: http://lists.kli.org/pipermail/tlhingan-hol-kli.org/2022-June/063074.html). There is no other interpretation of the verb prefix which makes sense. It is an usual error to forget *lu-* in informal speech, but not to use it when it isn't required, so its "unexpected" presence would point out to the *prefix trick* in the same way the presence of *qa-* point to it in *paq qanob*. Or am I missing something?

Understand that no one in that thread was speaking in terms of absolute rules. I was talking about ambiguity and uncertain interpretations. De'vID was a little more certain that Holmey law' lujatlh would just be considered an error.

Ok! I suppose *Holmey law' lujatlh* is more likely to be understood as an error because it's a third-person object with a prefix referring to a third-person object, whereas in *paq qanob* we have a third-person object with a prefix referring to a second-person object.

I think it's more likely to be considered an error because there's a Hol as the object of jatlh, which is a very common construction. Without some kind of super-obvious prompt that you're employing the prefix trick, most people will probably just think it's a prefix error. In paq qanob, it's much more obvious, even without context, that the prefix trick is being used because it's used a lot with nob.

The prefix trick is only used when the indirect object is not said explicitly. It's either SoHvaD paq vInob or paq qanob.

But if I'm not mistaken, in the thread I'm referring to *loDnI'Daj vavDaj je ja' qeylIS* or *SuStel vIjang* are interpreted as instances of the "prefix trick" with third-person prefixes.

Neither of those are uses of the prefix trick. In both sentences, the verb's prefix agrees with both the object and the subject.

<third-person plural> ja' <third-person singular>

<third-person singular> vIjang [first-person singular]

So the prefix trick is not only used when the indirect object is not explicit, or am I missing something?

The prefix trick is used when you want the prefix to agree with an otherwise unstated indirect object.

Anyway, re-reading the thread I found an explanation of De'vID in which he says that Okrand confirmed him that if Kahless were telling *a story* to his father and brother, these would be marked with *-vaD*, so no possibility of "prefix trick" with both explicit objects.

Yes, if he wanted to mention a lut, the direct object, it would be the object of the verb, and the indirect objects, the father and brother, would become the beneficiaries.

vavDajvaD loDnI'DajvaD je lut ja' qeylIS.

4. *toDuj lutraj quv lutraj je / QoymeH tlhIngan tuqmey / tIja'* (*paq'raD* 6, 4-6): Actually, *tI-* could here refer to *toDuj lutraj quv lutraj je*, making it impossible to interpret it as an instance of the "prefix trick", but in 7-9 (*DaH peHarghchoH / DaH molor yISuvchoH / tIja'*) it is clear that *tI-* refers to *tlhIngan tuqmey* ("prefix trick" in a direct quotation), so the context suggests that the first *tI-* is also an instance of the "prefix trick", doesn't it? And if so, is it right to use *ja'* only with an indirect object when it isn't in a direct quotation? ("Tell them so that the Klingon tribes hear your story of courage and your story of honor").

I don't think you're interpreting those lines correctly.
DaH peHarghchoH
    DaH molor yISuvchoH
    tIja'
Tell them, "Start a major battle now! Start Fighting Molor now!"
This is an instance of quotation. The quotation is not any object of any kind; it is just another sentence pushed against the sentence that refers to speaking. The prefix trick may or may not be in use here (its use in ja' quotations has never been fully solved, in my opinion), but this stanza follows the pattern of ja' completely.

I'm afraid I'm not understanding your point. I know that quotations are not the object of the sentence with the verb of speech. But in the thread I've mentioned *lujang* ("They answer him") is considered an example of "prefix trick" with third-person prefixes. What is the difference with *tIja'* ("Tell them")? In both cases there is no direct object. We also have the example *loDnI'Daj vavDaj je ja' qeylIS*, which is followed by a direct quotation, and in that thread seems to be considered an instance of the "prefix trick".

It is not an instance of the prefix trick. It is an instance of an indirect object (semantic role) appearing as the object (syntactic role) of the verb. Normally, the verb prefix agrees with the object (whether direct or indirect). When there is a direct object as the verb's object, or no possible direct object, the prefix trick lets you agree with an unstated indirect object instead of the explicit object.

You can see why I want to make sure the terms "object," "direct object," and "indirect object" are clearly defined.

Or are you using the concept "prefix trick" just for situations in which we have an explicit direct object and a prefix pointing to an elided indirect object,

...or no direct object, as in qajatlh...

 and you consider sentences like *lujang*, *SuStel vIjang* or *loDnI'Daj vavDaj je ja' qeylIS* just examples of verbs that can take an object with the semantic role of the indirect object (but you don't consider them instances of the "prefix trick")?

Yes, that's it exactly. Neither of those two sentences employ the prefix trick. They are examples of verbs taking objects with the semantic role of indirect object. I'm certain that's the case with ja'; I'm a little less sure with jang, since I'm not sure whether the direct object of the verb is the person or thing answered or something else. Examples in paq'batlh of lujang aren't totally revealing, just as the examples in TKD of qaja'pu' weren't revealing of what the natural direct object of ja' is (we didn't find this out until paq'batlh).

Ultimately, I think we give the prefix trick too much importance, analyzing it in far more a fine-grained way than Klingon linguists ever would. The verb prefix agrees with an object of some kind. It might not always be the object explicitly mentioned. It should always be clear what the agreement means. The distinction between direct and indirect object isn't carefully observed, though it does exist. Whether the object is "acted upon" or "receives the action" only matters when you've got more than one object, in which case only one can be explicit, the "acted upon" object. The "receives the action" object might become a beneficiary or might just be agreed with implicitly with the prefix trick.

-- 
SuStel
http://trimboli.name