scope of UIs

From Lojban
Jump to navigation Jump to search

Simple bridi

In simple bridi, UIs have scope over the whole bridi and mark the focus on the preceding selbri or sumti. For example:

la djan la'a pu dunda lo xrula la meris

It was probably John who gave flowers to Mary

la djan pu dunda lo xrula ju'a la meris

It was flowers John gave to Mary

la djan pu dunda lo xrula la meris xu

Was it Mary John gave flowers to?

la djan e'a dunda lo xrula la meris

Let it be John who gives flowers to Mary

la djan dunda lo xrula ei la meris

It should be flowers John gives to Mary

la djan pu dunda lo xrula la meris ja'o

So it was Mary John gave flowers to.

The great majority of (if not all) UIs work like that. When the UI is at the beginning of the bridi, there is no special focus (or the focus is on the entire bridi).

Subordinate bridi

ta vasru lo xrula poi la'a la djan pu dunda ke'a la meris

That contains flowers which probably John gave to Mary.

ta vasru lo xrula poi la djan pu dunda xu ke'a la meris

That contains flowers which John gave to Mary?

ta vasru lo xrula poi ei la djan dunda ke'a la meris

That contains flowers which John should give to Mary

It would seem then that for simple subordination the rule is that the UI has scope over the immediate bridi. We don't want {la'a}/{xu}/{ei} to apply to vasru in the above examples.

Abstractions

la alis do tugni le du'u ei la djan dunda lo xrula la meris

Alice agrees with you that John should give flowers to Mary.

It would seem that we don't want that to mean "Alice should agree with you...", so the rule holds so far that the UI has scope over the immediate bridi.

This is like indirect discourse:

la alis cusku le se du'u la djan pu dunda lo xrula ju'a la meris

Alice said that it was flowers John gave to Mary.

la alis cusku le se du'u la djan e'a dunda lo xrula la meris

Alice said to let it be John who gives flowers to Mary

Unfortunately, Lojban lore has tended to be that attitudinals, no matter where they appear, must show the attitudes of the speaker, and also that questions within du'us are direct questions, so this reading:

la alis cusku le se du'u xu la djan pu dunda lo xrula la meris

Alice said words to the effect of asking whether John gave flowers to Mary.

is not canonical.


The subordinate {xu} case seems different. I would take

ta vasru lo xrula poi la djan pu dunda xu ke'a la meris

That contains flowers which John gave to Mary?

to be a question, "Does the vase contain the flowers John gave to Mary?" with, to be sure, the core of the question being whether John gave them to her (as opposed to what?). Maybe "Is it giving to Mary that John did with the flowers that vase contains?" (trl. from Erse?). I think English cannot say this very well at all. But it easier to see in explicit contrast cases "Are the flowers in the vase the one John gave to Mary -- or the ones he made her pay for?"

  • So you're saying that the {ta vasru lo xrula} bit is presupposed, not claimed.
    • I think even that the flowers are somehow connected with John and Mary is presupposed, if we are to take placement as being indicative of the central contrast allowed. "I request a selection from claims of the form X(Lx(the vase contains the flowers John x Mary)gave), where "X" ranges over affirmation, denial and all the waffles in between.
      • Yes, that sounds right. So putting a UI within subordinate poi clauses seems to be just a matter of focus and presuppositions. What would happen with noi clauses? If a noi clause can be removed without changing the assertive status of the clause, does that include any attitudinal within it?

ta vasru lo xrula noi la djan pu dunda ke'a la meris

That contains flowers, which by the way, were given by John to Mary

ta vasru lo xrula noi xu la djan pu dunda ke'a la meris

That contains flowers, which btw, were they given by John to Mary?

      • Is that a claim plus a side question, or is it just a question with some odd presupposition/focus?

The indirect question markers (and question markers and the like are separate from the others I think) need, as you well know, {kau}. This seems to read all right as a top level:"Was it whether John gave the flowers to Mary that Alice said (or "asked")?"

  • But to say whether John did it is not the same as to ask whether John did it. See my response to And below.
    • Quite true, and I would take the use of {cutse} here to favor "said" over "asked."

It may be useful to do a componential analysis: what is given in each case, what is focused on, and what is to be done about it.


And Rosta:

  • Reasoning by analogy with {xu}, the readings of the subordinate bridi cases should be obtained by UIkau, where kau restricts scope to the local bridi.
    • I did experiment with that for a while some time ago, but I was not very satisfied with the results. I think there is a difference in saying whether John gave the flowers or not: la alis cusku le se du'u xu la djan pu dunda lo xrula la meris, she said either that he did or that he didn't, and saying words to the effect of asking a question. Hmmm, maybe something like:

la alis cusku le se du'u pau xukau la djan pu dunda lo xrula la meris

Alice said (question) whether John gave flowers to Mary.

la alis cusku le se du'u ju'a xukau la djan pu dunda lo xrula la meris

Alice said (assertion) whether John gave flowers to Mary.

Of course, each attitudinal can have a specific selbri to report it, so {pe'i} is reported with {jinvi}, {ju'a} with {xusra}, {a'o} with {pacna}, {e'a} with {curmi}, questions with "asked", and so on, even when there is no gismu we can make an appropriate lujvo, so the {cusku le se du'u UI} construction may not be really necessary. But in any case, trying to give a scope beyond the du'u to the attitudinals within it normally results in weird stuff, so it seems like a waste to require the kau in those cases.

      • Weird, but sometimes just what we want to say. I take {cusku}to be more about the form of words than the content (though the wordlist is not perfectly happy with that) and so that with a {xu} it comes out as asking a question rather than makign an assertion. On the other hand, "express" sound like an assertion -- or some more emotional utterance. In any case, I would take {ju'a xukau} as odd and then go looking for what it might be idiomatically. I am not sure I would end up where you do, but it is possible. I would find the {pau} redundant.
  • We ought to have a proper classification of UI, based on their logical/semantic properties. For those that are truly interjections (like 'wow', 'ouch' etc.), they by their very nature should not have scope, since they are mere vocalizations of what could equally well be expressed by gesture, facial expression, or whatever. For those that aren't interjections, do contribute to logical form and hence do have scope, then ideally they will follow the basic scope rule, whereby scope follows linear order.
    • That should be a part of the BPFK program. I'm not sure a strict line can be drawn though. It seems to me that lexicalized interjections will inevitably end up acquiring logical/semantic properties. For example, take {ua}, "discovery". It sounds like a true interjection, but I wouldn't mind being able to say something like:

la alis cu morji le du'u ua le ckiku cu cnita le rulpatxu

Alice remembers that (discovery!) the key is under the flower-pot.

similarly to:

la alis cu morji le du'u ti'e le ckiku cu cnita le rulpatxu

Alice remembers that (hearsay) the key is under the flower-pot.

la alis cu morji le du'u la'a le ckiku cu cnita le rulpatxu

Alice remembers that (probably) the key is under the flower-pot.

      • Moving in logical operators like {la'a} might be proper, but I think I would shift to {lakne} to be safe; we want to able to focus on what she probably remembers by picking out its head here too. The same applies. even more strongly,

for evidentials.

  • In general, I dislike the way that in the above examples word order is used to do focus. For example, I should be able to say {la djan pu dunda lo xrula la meris ja'o} and mean "John gave flowers to Mary, I conclude" rather than "It was Mary that John gave flowers to, I conclude". IMO the order of elements should be governed by uncodified pragmatic considerations, so as to allow greater flexibility in usage. (The now unavoidable exception is when order signals logical scope.) It follows that Lojban needs some overt way of marking focus, either a cmavo that precedes the focused element, or else something like {la meris poi'i la djan pu dunda lo xrula ke'a} or {la meris du da poi la djan pu dunda lo xrula ke'a}.
    • But the poi'i/du versions are even more dependant on word order. I think it would be good to have a focus marker so that focus can be marked independently of the attitudinal position. Should it have the grammar of BAhE?

pc:

Something And said above, combined with a middle-of-the-night flash, convinces me that I have fallen for one of the oldest Loglan/Lojban traps: thinking UI have predicative force, that {a'o mu klama} is true or false at least partly depending on what my hopes actually are. Not so; {a'o} is not part of the content of the claim but part of its manner of presentation. It is the surface residue of a variation in deep structure top level performative, in this case what would come up as "I say hopefully" in English. And this is true for most UI, whether the change is adverbial to the basic "I say/assert/..." or marks a different performance altogether. It is for this reason that the scope of these critters is sentential -- and always the speaker's emotion (loosely speaking). Moving to third person reporting, means making the appropriate raising of that buried "verb"; repeating the UI does not work, since it gives the new speaker's emotion, not the one reported. That is, all of the examples above of displaced emotions fail, coming back to the speakers emotions -- regardless of where they are placed (except inside direct quotes, of course).

  • It would be a shame if {a'o} were limited to expressing hope, and not be usable to express a hope. I don't think that agrees with usage, and it seems it would be hard to impose such a restriction.
    • I am not sure I get what you are saying. {a'o} does express a hope; it just isn't true or false depending upon whether I actually hope what I have expressed. (I think it is neither true nor false at all, but that is another matter.)
      • How do you know which hope a'o expresses if it is merely an adverbial to the basic "I assert"? You are saying that {a'o la djan pu klama} makes the assertion that John came, and the speaker makes the assertion with hope. That's not the same as expressing the hope that John came. With your reading, if you say {a'o la djan pu klama}, I'd have to report that as {la pycyn cu xusra le du'u la djan pu klama ije la pycyn pacna}, not as {la pycyn pacna le du'u la djan cu klama}. How do you say something to express the hope that John will come?
        • I think we are into terminology again. If I say {a'o la djan pu klama) (a slightly odd hope, but intelligible), the the hope I express is that John came/has come. Hoping is not a performance and so I don't have a performative to cover it, the most I can do is say hopefully. But I can do that even if I have no such hope. Still, we tend to infer from hopeful utterances to hopes. I can report the utterance even as {mi pu paxna cutse le du'u la djan pu klama}, from which you might very well in fer {la pycyn pu paxna le nu la djan pu klama} (I think I hope for events not propositions). Note, please, I did NOT say that I thought that {a'o la djan pu klama} made an assertion; in fact I said that I did not think it did: note I stressed that it "hopefully" modifies "say," not "assert." I think that "O that John will have come" is a perfectly good translation of the sample (but not everyone agrees -- I am glad that you do).
          • I agree that "O that John will have come" is a good translation. (And "O that John would have come" for {au la djan pu klama}). I would report them as "he expressed the hope that John has come" and "he expressed the wish that John had come". I want {le du'u a'u la djan pu klama} to be "the hope that John has come" and {le du'u au la djan pu klama} to be "the wish that John had come", and {le du'u e'u la djan cu klama} "the suggestion that John come", {le du'u xu la djan klama} "the question whether John has come", pe'i for the opinion, ju'a for the assertion, etc. To use (say) the UI-bridi is to express the hope, the wish, the opinion, to make the suggestion, the question, the assertion. And {le du'u UI-bridi} can be used to refer to the hope, the wish, the opinion, the suggestion, the question, the assertion.
            • Well, it is a system, but I think it is in the wrong place. Starting with {le du'u} suggests that these are all proposition of some sort, but that is exactly what they are not -- nor anything directly related to propositions, except in the case of questions. Do we use {le du'u ko klama} for "the command to come"? For the most part, these critters are about events not propositions at all (even {ko}, but not {xu} etc.). That said, I don't know a general way to talk about the hope/wish/assertion/request/suggestion that John come. The instant case, in the context "he expressed....," seems naturally to be handled by the appropriate variations on "express." But there are probably contexts where that does not work. Note that hope and desire differ from most of the others in that they are not performances nor evidentials. The others are still a mixed bag tha need separate discussions.
              • These are not propositions, but they do have propositional content. The bare bridi without the attitudinal provides the propositional content. The attitudinal adds to the propositional content whatever else is needed to make it a hope/wish/assertion/request/suggestion. I would certainly include {lo du'u ko klama} for the command. Indeed I think {ko} is just {do} plus an attitudinal that encompasses e'a/e'e/e'o/e'u and probably some more that don't have a cmavo assigned, there's no difference that I can see between {e'u do} and {e'u ko}, {e'o do} and {e'o ko}, etc. I don't have a problem with discussing things separately, but it is also interesting to study the aspects that they do have in common.
                • Surely. They all require modified/replaced speech performatives. As for the proposition/event dichotomy, English pretty much has both with not difference I can see. Lojban is, of course, all over the place, but I think there is a preference for events (marked in {pacna}, for example). And, after all, it is the event tht counts, for it makes the proposition true or false. On the other hand, if we are just mdoiffying or replacing speech performances, we are pretty much stuck with propositional. I think your analysis of {ko} is correct, though what combination of attitudinals -- or maybe a sui generis one -- I am not sure.

There are some exceptions, of course. {la'a}, for example, is a logical operator, on a par with {ka'e} (well, not exactly, but...) and not an emotion operator or relativized to the speaker (even though it may be his/er estimate). That is, the truth of the claim really does depend upon what the probabilities are, in contrast to items in the {ju'o} setm which merely indicate confidence levels. (This is not the only interpretation of {la'a} of course, but it is the one that fits best with the "logical language" idea.)

  • Do you consider the e-series (ei, e'a, e'e, e'i, e'o, e'u) to be "emotions"? Is there no parallel between these:

la alis do tugni le du'u la'a la djan pu dunda lo xrula

Alice agrees with you that probably John gave flowers.

la alis do tugni le du'u ei la djan cu dunda lo xrula

Alice agrees with you that John should give flowers.

    • I would be wary of attributing anything to a whole class of Lojban expressions based on their form. In this case, {e'i} seems to be different from the others in the set. Unlike them, it does not seem to point to a different performance (judging, permitting, requesting, suggesting), but only an adverbial addition to the usual expressing. In other words, {e'i} sentences seem to me to be true or false, the rest not. I would translate your latest English as {la alis do tugni le du'u la djan bilga le nu dunda lo xrola} And the Lojban as "Is it that John give the flowers about which Alice ought to agree with you" (or something like that). As I said, {la'a} looks like a pure logical operator "it is probable that;" I think it can remain in abstractions, where {ei} cannot (in the same meaning). {e'e}, which I admit to not understanding, seems, in its most plausible reading, to be more like {e'i} or even over into evidentials.
      • I'll pass on {e'i}, I don't really know what to make of it. {e'e} I use for exhortations, given that there is no other UI for that, that it fits in the e-series and that I couldn't make much sense of the keyword either. So I use {e'e ko zukte} for "go for it!". We'll just have to disagree on the scope {ei} within du'u. --xorxes
        • Hey, {e'e} as "attaboy!" makes the most sense of it so far: speakers judgment about competence of an agent in embedded sentence. Exactly what happens there logically is less clear, but putting it with the other non-assertions makes the most sense so far. I take {e'i} to be an admission that what I say is under someone else's control -- not necessarily with the implication that I would say otherwise. Someday, we will have to settle the issue of the scope of {ei}, as you (or was it &) have pointed out on other occasions. I don't have much trouble with it your way (provided you then shift all the others as well), but it does seem to limit the possibilities -- for weird stuff admittedly -- and the verbing of many of the other cases seem more natural than continuing the UI.
          • I can still get the weird meaning with my method by saying: {la alis do tugni le du'u la djan cu dunda lo xrula kei ei}, so that is not a limitation.
            • Neat. In fact neater than the original idea. But aesthetics need not be decisive here (or anywhere).

There are also genuine interjections, which have no deep structure source, being inserted at the level of utterance. these obviously cannot carry over to reported form, but can be represented only predicatively (thereby loosing some of the effect, to be sure). That is, although they are different in origin from the other UI, they are treated the same as we move from immeidate first person expression.

I am less clear about how to place the focus except in the case of indirect questions (which, yes, do require -- see above -- {te preti} rather than {cutse}). The {kau} suggestion has its charms, but I think that at this level the emotion has become part of the content and so needs to be presented as such. This will usually give a complex which will automatically sort out the focus into first place or something like it. Otherwise, {ba'e} itself is a pretty good focusing device. If focus has a deep source, it is in application of lambda to argument and that allows several different representations by its very nature -- the structure can be resolved or kept in bare form, which latter makes the argument a focus.