relative Clauses and Phrases
Relative clauses and phrases
We often add relative clauses -- or closely related phrases – to sumti. These have basically two purposes. On the one hand, the sumti may not of itself adequately characterize the object referred to and so the clause adds further characterization, narrowing the search for a referent still further. With description sumti it is in theory possible to do that within the predicate of the description, but doing so is often syntactically too complex or results in a structure too ambiguous to be what is useful – or reachable in a conversation. On the other hand, sometimes we have the object adequately identified (until someone objects) and know what we want mainly to say about it but also want to throw in some other information off the mainstream. These two moves are respectively restrictive and non-restrictive relative clauses (and similar distinctions for phrases).
The two differ in their logics (always a Lojbanic concern). A restrictive relative clause becomes a part of the way of referring to the object; it has the same status as the bridi in a description. Thus, it is unaffected by negations and other outside complexities: “It is not the case that the man that came to dinner at much” {naku le ninmu poi zukte le nu vitke le nu citka lo sanmi ku’o citka piso’o lo sanmi} goes smoothly over to {le ninmu poi zukte le nu vitke le nu citka lo sanmi ku’o naku citka piso’o lo sanmi}, the connection undisturbed by what happens in the rest of the sentence. The restrictive clause makes no claim, it merely tells us to look at things that satisfy its predicate to find the object intended. A non-restrictive clause on the other hand does make a claim about its object; it conjoins another claim onto the main one. The conjunction is a somewhat irregular one, however, since it ties more tightly to the sumti to which it is attached than to the rest of the sentence. Thus, when a negation boundary moves past it, it is neither negated nor does it “and” turn into an “or”: {naku le mi patfu noi kalte lo pavyseljirna ku’o te pixra lo xrula} goes over to {le mi patfu noi kalte lo pavyseljirna ku’o naku te pixra lo xrula}. The first expands to {naku le mi patfu cu te pixra lo xrula ije py kalte lo pavyseljirna} and the second to {le mi patfu naku te pixra lo xrula ije py kalte lo pavyseljirna}.
In Lojban a restrictive relative clause is introduced by {poi} or {voi} of selmaho NOI, the first being {lo}-like in pointing to properties the intended referent actually has, the second being {le}-like in only using properties I use to describe a thing, whether it has them or not. In the predication that follows this introductory word, the place where the sumti to which it is attached would occur is occupied by {ke’a}. When this is in first place, effectively right after the introductory word, the {ke’a} can be dropped. Lojban non-restrictive relative clauses are just the restrictive ones except that the introductory word is {noi}. Since the following clause is asserted, it claim that the object really has the property – unlike {voi} clause which are not asserted and allow using properties the thing does not have if they help to find the referent and also unlike the {poi} clause in being asserted where the {poi} is just to help find the referent, not make a claim. Sometimes we want to use more than one relative clause and then the two (or more) are joined using {zi’e} (of ZIhE), always a sort of conjunction. As the examples show, relative clauses of whichever type close with {ku’o} (of KuhO), which is usually elidable, though, if the clause is complex, good style recommends using it. And of course it is needed if two predicates – one from the clause and one from the higher structure – would come together.
Some relative clauses are so common that they get marks of their own, becoming mere phrases by dropping the oft-used predicate and leaving only a term. The most common of these is, of course, identity clauses, which reduce to just the term for the other half of the equation, introduced by {po’u} or {no’u} (of GOI) depending upon whether the identity is used restrictively or not. A special case of identity is not to report an existing identity between the referents of two terms, but rather to assign the referent of one term to another term which has no established meaning, the members of the {ko’a} and {fo’a} series in KOhA. This is done using {goi}, also in GOI, with the requirement that the one term be an assignable, usually a pronoun (in the recommendations section, the issue of which term that pronoun is will be dealt with). Using {ko’a} etc. saves repeating longer terms without relying on anaphoric pronouns.
Another relation almost as common as identity is possession. This attaches a reference of the possessor to a term for the possessed. “Possession” here is used in a variety of senses, of which three have distinctive introductory forms. {pe} is used for the loosest of relations that might be called possession (or use a genitive form – ending in ‘s in English). It amounts to little more than “is related to” or “is associated with” somehow: no legal claim of ownership is made; even “my favorite topic” would qualify for a {pe mi}. {pe} is restrictive; the corresponding non-restrictive is {ne} (not too surprisingly). The next level of possession is the normal one, marked by {po} in the restrictive usage. There is no shortening for the non-restrictive sense; one has to fall back on (roughly) {noi se ponse}. The final level of possession is “inalienable possession” (marked by {po’e}) the kind that cannot be unpossessed (in the given culture): body parts and family members spring to mind, though these – and most other cases – have predicates with places for this relation already, so that {le bersa po’e mi} is not clearly distinguished from {le bersa be mi}. This step is also only shortened to a phrase for restrictive cases. The only problem is that there is not a good predicate (at the moment) for “possesses inalienably.”
{pe} and {ne} are not restricted to terms as followers but may take other phrases – mostly modal – that we may want to have modify a term only, not the main predication. {mi viska le cripu vi le rirxe} means, roughly, “I saw the bridge when I was near the river” with {vi le rirxe} modifying the whole rest of the sentence – where the event of seeing took place. On the other hand, {mi viska le cripu pe vi le rirxe} means “I saw the bridge which is by the river,” the {vi le cripu} being used only to say which bridge is meant, and not saying where the seeing took place (I might be looking through a telescope several miles away for all this says).
Relative phrases end with GEhU, whose only member is {ge’u}. This is usually elidable.
When a term gets complicated by conjoining two or more terms or by complexing the predicates involved, and we want a relative clause or phrase to modify the whole of the complex, the complex is closed with VOhU (i.e., {vo’u}) and the clause or phrase is attached thereafter (otherwise it would apply only to the last part of the complex).
By and large, these systems have not caused much controversy. All that has arisen is one obscurity, one addition (or set of additions) and one extension.
The obscurity involves {goi}, which assigns the identity of one sumti to another. The question is, in {ko’a goi ko’e} or any other case where the two expressions are of the same sort (two descriptions, say) which identity is assigned to which. Is it that {ko’e} get the identity of {ko’a} or conversely. Of course, when one term is a description or a name and the other an assignable pronoun ({ko’V/fo’V}) the direction is clear; but if both are of that latter sort or both descriptors (“the lender hereinafter the party of the first part”) it is less clear. The usage pattern is pretty clearly that the term before the {goi} gives its identity to the one after. Of course, there is the practical test that the one whose referent is already established gives to the one that is not and this may conflict. In that case, the identity transfer is carried out by {no’u} {ko’a no’u le mamta be mi} – assignment in the afterthought mode. This last is a proposal, but affects very little usage except some backward {goi}s can be corrected by changing the mark rather than by reversing the order.
The addition is for clauses that connect to the quantifiers on the terms being modified. As of now, a relative clause or phrase attaches to the whole term, {poi broda} to {ci brode} in {ci brode poi broda cu brodi}, selecting from the brode three that also broda – partially answering the question, which three. There are several quantifiers, however, that call for a clause that does not modify the term as a whole but rather precise the quantifier itself: {du’e} “too many”, {mo’a} “too few” and {rau} “enough.” All of these raise the question “For what is the number too many, too few or enough?” The answer then is to give a purpose in the form of either a event description attached with, maybe , {pe} or a sentence attached with a new a new connective in NOI. There is also the possibility that we might want modifications that spoke of the results rather than the intention: “so many that…,” “so few that…” or “just right for…” where the expression here give what the number did – or even could have – achieved, whether or not it was intended (or even actually occurred). This would take, clearly, either a different set of quantifiers or different GOI or NOI or both. There are not yet concrete proposals for lexing this proposal, just the need to fill this gap.
Finally, an extension of the use of {noi} has occurred but not been much discussed. It has always so far been connected with an explained use of {zo’e} (see KOhA 7 8). This latter is apparently taken as referring to an unspecified object to which {noi} gives a specification (though not an incidental one). From the use made of the pattern {zo’e noi broda} it appears that the object referred to by {zo’e} is not ever simple concrete object and that {zo’e broda} is to be taken in some (not easily determinable) idiomatic sense, not simply that zo’e is in the set of brodas. Until this whole complex is better defined, this change (whatever it is) is not proposed or recommended.
That leaves only the proposed rule about {goi}: that the identity of the object named on the left be assigned to the name on the right (and that the reverse procedure be dealt with using {no’u}).
And the proposal to pursue lexing for a cmavo to attach to terms introduced by quantifiers that relativize size to some purpose or outcome an expression for that purpose or outcome.
The latter proposal will not affect past usage, unless someone has tried to do this with existing material and there is no evidence that this has happened. The first may require changing a few backward assignments, but again those are not numerous – if they exist at all.