The following cmavo are discussed in this section:

The syntax of descriptions is fairly complex, and not all of it can be explained within the confines of this chapter: relative clauses, in particular, are discussed in Chapter 8. However, most descriptions have just two components: a descriptor belonging to selma'o LE or LA, and a selbri. (The difference between selma'o LE and selma'o LA is not important until Section 6.12.) Furthermore, the selbri is often just a single brivla. Here is an elementary example:

The long gloss for le is of course far too long to use most of the time, and in fact le is quite close in meaning to English “the”. It has particular implications, however, which “the” does not have.

The general purpose of all descriptors is to create a sumti which might occur in the x₁ place of the selbri belonging to the description. Thus le zarci conveys something which might be found in the x₁ place of zarci , namely a market.

The specific purpose of le is twofold. First, it indicates that the speaker has one or more specific markets in mind (whether or not the listener knows which ones they are). Second, it also indicates that the speaker is merely describing the things he or she has in mind as markets, without being committed to the truth of that description.

Note that English-speakers must state whether a reference to markets is to just one ( “the market”) or to more than one ( “the markets”). Lojban requires no such forced choice, so both colloquial translations of Example 6.6 are valid. Only the context can specify which is meant. (This rule does not mean that Lojban has no way of specifying the number of markets in such a case: that mechanism is explained in Section 6.7.)

Now consider the following strange-looking example:

Example 6.7 is not self-contradictory in Lojban, because le mlatu merely means something or other which, for my present purposes, I choose to describe as a cat, whether or not it really is a cat. A plausible instance would be: some animal we had assumed to be a cat at a distance turned out to be actually a dog on closer observation. Example 6.7 is what I would say to point out my observation to you.

In all descriptions with le , the listener is presumed to either know what I have in mind or else not to be concerned at present (perhaps I will give more identifying details later). In particular, I might be pointing at the supposed cat or cats: Example 6.7 would then be perfectly intelligible, since le mlatu merely clarifies that I am pointing at the supposed cat, not at a landscape, or a nose, which happens to lie in the same direction.

The second descriptor dealt with in this section is lo. Unlike le , lo is nonspecific:

Again, there are two colloquial English translations. The effect of using lo in Example 6.8 is to refer generally to one or more markets, without being specific about which. Unlike le zarci , lo zarci must refer to something which actually is a market (that is, which can appear in the x₁ place of a truthful bridi whose selbri is zarci). Thus

must be false in Lojban, given that there are no objects in the real world which are both cats and dogs. Pointing at some specific cats or dogs would not make Example 6.9 true, because those specific individuals are no more both-cat-and-dog than any others. In general, lo refers to whatever individuals meet its description.

The last descriptor of this section is la , which indicates that the selbri which follows it has been dissociated from its normal meaning and is being used as a name. Like le descriptions, la descriptions are implicitly restricted to those I have in mind. (Do not confuse this use of la with its use before regular Lojbanized names, which is discussed in Section 6.12.) For example:

In Example 6.10 , la cribe refers to someone whose naming predicate is cribe , i.e. “Bear”. In English, most names don't mean anything, or at least not anything obvious. The name “Frank” coincides with the English word “frank” , meaning “honest” , and so one way of translating “Frank ate some cheese” into Lojban would be:

English-speakers typically would not do this, as we tend to be more attached to the sound of our names than their meaning, even if the meaning (etymological or current) is known. Speakers of other languages may feel differently. (In point of fact, “Frank” originally meant “the free one” rather than “the honest one”.)

It is important to note the differences between Example 6.10 and the following:

Example 6.12 is about a specific bear or bearlike thing(s), or thing(s) which the speaker (perhaps whimsically or metaphorically) describes as a bear (or more than one); Example 6.13 is about one or more of the really existing, objectively defined bears. In either case, though, each of them must have contributed to the writing of the story, if more than one bear (or “bear”) is meant.

(The notion of a “really existing, objectively defined bear” raises certain difficulties. Is a panda bear a “real bear” ? How about a teddy bear? In general, the answer is “yes”. Lojban gismu are defined as broadly as possible, allowing tanru and lujvo to narrow down the definition. There probably are no necessary and sufficient conditions for defining what is and what is not a bear that can be pinned down with complete precision: the real world is fuzzy. In borderline cases, le may communicate better than lo.)

So while Example 6.10 could easily be true (there is a real writer named “Greg Bear”), and Example 6.12 could be true if the speaker is sufficiently peculiar in what he or she describes as a bear, Example 6.13 is certainly false.

Similarly, compare the following two examples, which are analogous to Example 6.12 and Example 6.13 respectively:

Example 6.14 says who the author of the story is: one or more particular human beings that the speaker has in mind. If the topic of conversation is the story, then Example 6.14 identifies the author as someone who can be pointed out or who has been previously mentioned; whereas if the topic is a person, then le remna is in effect a shorthand reference to that person. Example 6.15 merely says that the author is human.

The elidable terminator for all descriptions is ku. It can almost always be omitted with no danger of ambiguity. The main exceptions are in certain uses of relative clauses, which are discussed in Section 8.6 , and in the case of a description immediately preceding the selbri. In this latter case, using an explicit cu before the selbri makes the ku unnecessary. There are also a few other uses of ku : in the compound negator naku (discussed in Chapter 16) and to terminate place-structure, tense, and modal tags that do not have associated sumti (discussed in Chapter 9 and Chapter 10).