scope of tenses and NA

In a recent message on the list (16095), John Cowan has made clear the book's official interpretation of NA and tenses as follows:

  1. NA tagging the selbri moves to the left end of the prenex.
  2. Tenses have scope in the order that they occur in the bridi, whether they are free-floating or tag the selbri, and always to the right of any na which directly tags the selbri.

Thus, na roroi is equivalent to roroi na and means not always, etc. This goes against the long-established tradition* of those who tend to debate such things, as well as the common-sense notion that roroi na means always not, or never. The scope of NA was hard enough to use correctly as it was, now it's even more broken. Therefore I think it is inevitable at this point that the complicated rules given by the book will be ignored, and both tenses and and NA will be given scope in the order in which they occur in the bridi, whether they are free-floating or not, and the sooner the better. — Adam

The following was originally from bi'ai, but is more appropriate here:

It should be noted, that in lojban (as defined by cll),
ka'e na means the exact same thing as na ka'e. The na is converted
to naku and interpreted before anything else (by moving it to the
left of the prenex), according to chapter 16. Additionally,
ka'enai is bad grammar. In lojban, na ka'e na broda is
exactly the same as naku naku zo'u ka'e broda which is the same
as ka'e broda. I imagine xorxes was already aware of this,
but it is misleading to claim the above as fact when it is contrary
to the offical language, so I feel obligated to comment. A possible
way I can think of to actually say that the negative is possible
is ka'e na'e broda. --mi'e .djorden.

  • I was not aware of that at the time I wrote it. In fact, the rule that makes na ka'e and ka'e na mean the same while naku ka'eku and ka'eku naku are clearly different is so crazy that I still can't believe it can be defended. But to clear up any ambiguity bi'ai should be defined as bi'aiku = naku ka'eku naku. In the negation chapter, cll leaves open a little window for sanity:
    • You can elide that ku btw: "ka'e naku" is the same as "ka'eku naku". The rule which allows this is term -> (tag | FA #) (sumti | /KU#/). --mi'e .djorden.

8.5) la djan. [cu] na ba klama la paris. .e la rom.
John [false] later-will-go-to [both] Paris and Rome.
It is false that John will go to Paris and Rome.

8.7) la djan. [cu] ba na klama la paris. .e la rom.
John later-will [false] go-to [both] Paris and Rome.

We stated in Section 3 that sentences like Example 8.5 and Example 8.7 appear
to be semantically identical, but that subtle semantic distinctions may
eventually be found.

Unfortunately, no example with clearly conflicting scope is given, but at least the possibility that na <tag> and <tag> na are different is not excluded. --xorxes

  • The possibility of semantic difference is allowed, but it is clearly stated that they both mean "It is false that ...". The semantic differences which I imagine were in mind are things such as which part is focused on; for example the ba being more important than the na (stressing that la djan. pu isn't neccesarily false) could be indicated by putting the ba before the na. --mi'e .djorden.
    • There is no clearly stated "It is false that..." gloss for 8.7! The subtle difference I imagine was in mind was between "It is false that...will..." and "It will be false". There is no apparent difference between the two, but the text allows that one may eventually be found. If the <tag> used had been something less transparent to scope, the example would have been more informative. --xorxes

* See for example message 15193, in which pc and xorxes seem to agree without too much discussion:

    • This example hardly justifies calling it "long-estabilished" tradition (It's from two months ago!). -mi'e .djorden.
      • No, it's is merely an example. I don't believe that the rule that tenses and NA have scope in the order they occur has ever been disputed, before the above message from Cowan, and it's been used numerous times.

xorxes: Where would you put the ta'e relative to {da poi nu}/{lo nu}? [In the sentence da poi nu mi citka lo cakla zo'u mi ta'e nelci da]

pc: When prefixing is implicit, tenses have to be the outermost item except for negation, thus the quantifier must be inside them.

xorxes: The way I understand it: to make a fully prenexed expression,
you start with already explicit prenex terms, then selbri tcita,
then non-prenexed terms.

pc: Yes, but {na} and tense if they occur ARE already prenex and at the far left end.

Here with the problems. For most normal uses, the whole pre-selbri pack (tense in thr broad sense, modals, negation) need to be at the far left and in order. But it is more than occasionally needed that quantifiers in particular get outside some or all of those items. We cannot do for tense as we do for negation: deMorgan the quantifier (or something therelike). How then to manage this. The current system allows at most one solution ("at most" because I am not sure even this will actually work) of {ku}ing the items at the appropriate places around the problematic quantifiers. It is unclear whether {ku}ing one would restrict all the following ones or whether they would hop over it to the far left again, thus destroying what little is left of a connection between visible order and logical. On the other hand, just about any other convention — that everything goes far left in order, for example — will make the usual situation the most complex one.

I am inclined to think the whole notion of scope and order needs a new look, rather than a variety of piecemeal solutions for individual cases. I would suggest starting with the underlying logical forms and see how to insert them into the surfaces structure. This may come up with principles that will be as bad as the present, but at least modifications would be motivated and clear, unlike the present situation.

Created by admin. Last Modification: Monday 22 of September, 2003 21:50:06 GMT by pycyn.