Chapter 8. Relative Clauses, Which Make sumti Even More Complicated

The following cmavo are discussed in this section:

Let us think about the problem of communicating what it is that we are pointing at when we are pointing at something. In Lojban, we can refer to what we are pointing at by using the pro-sumti ti if it is nearby, or ta if it is somewhat further away, or tu if it is distant. (Pro-sumti are explained in full in Chapter 7.)

However, even with the assistance of a pointing finger, or pointing lips, or whatever may be appropriate in the local culture, it is often hard for a listener to tell just what is being pointed at. Suppose one is pointing at a person (in particular, in the direction of his or her face), and says:

What is the referent of ti? Is it the person? Or perhaps it is the person's nose? Or even (for ti can be plural as well as singular, and mean “these ones” as well as “this one”) the pores on the person's nose?

To help solve this problem, Lojban uses a construction called a “relative clause”. Relative clauses are usually attached to the end of sumti, but there are other places where they can go as well, as explained later in this chapter. A relative clause begins with a word of selma'o NOI, and ends with the elidable terminator ku'o (of selma'o KUhO). As you might suppose, noi is a cmavo of selma'o NOI; however, first we will discuss the cmavo poi, which also belongs to selma'o NOI.

In between the poi and the ku'o appears a full bridi, with the same syntax as any other bridi. Anywhere within the bridi of a relative clause, the pro-sumti ke'a (of selma'o KOhA) may be used, and it stands for the sumti to which the relative clause is attached (called the “relativized sumti”). Here are some examples before we go any further:

In the literal translations throughout this chapter, the word “IT”, capitalized, is used to represent the cmavo ke'a. In each case, it serves to represent the sumti (in Example 8.2 through Example 8.4, the cmavo ti) to which the relative clause is attached.

Of course, there is no reason why ke'a needs to appear in the x1 place of a relative clause bridi; it can appear in any place, or indeed even in a sub-bridi within the relative clause bridi. Here are two more examples:

In Example 8.6, ke'a appears in an abstraction clause (abstractions are explained in Chapter 11) within a relative clause.

Like any sumti, ke'a can be omitted. The usual presumption in that case is that it then falls into the x1 place:

almost certainly means the same thing as Example 8.3. However, ke'a can be omitted if it is clear to the listener that it belongs in some place other than x1:

is equivalent to Example 8.4.

As stated before, ku'o is an elidable terminator, and in fact it is almost always elidable. Throughout the rest of this chapter, ku'o will not be written in any of the examples unless it is absolutely required: thus, Example 8.2 can be written:

without any change in meaning. Note that poi is translated “which” rather than “such-that” when ke'a has been omitted from the x1 place of the relative clause bridi. The word “which” is used in English to introduce English relative clauses: other words that can be used are “who” and “that”, as in:

and

In Example 8.10 the relative clause is “who was going to the store”, and in Example 8.11 it is “that the school was located in”. Sometimes “who”, “which”, and “that” are used in literal translations in this chapter in order to make them read more smoothly.

The following cmavo is discussed in this section:

There are two basic kinds of relative clauses: restrictive relative clauses introduced by poi, and incidental (sometimes called simply “non-restrictive”) relative clauses introduced by noi. The difference between restrictive and incidental relative clauses is that restrictive clauses provide information that is essential to identifying the referent of the sumti to which they are attached, whereas incidental relative clauses provide additional information which is helpful to the listener but is not essential for identifying the referent of the sumti. All of the examples in Section 8.1 are restrictive relative clauses: the information in the relative clause is essential to identification. (The title of this chapter, though, uses an incidental relative clause.)

Consider the following examples:

In Example 8.12, the information conveyed by poi blanu is essential to identifying the dog in question: it restricts the possible referents from dogs in general to dogs that are blue. This is why poi relative clauses are called restrictive. In Example 8.13, on the other hand, the dog which is referred to has presumably already been identified clearly, and the relative clause noi blanu just provides additional information about it. (If in fact the dog hasn't been identified clearly, then the relative clause does not help identify it further.)

In English, the distinction between restrictive and incidental relative clauses is expressed in writing by surrounding incidental, but not restrictive, clauses with commas. These commas are functioning as parentheses, because incidental relative clauses are essentially parenthetical. This distinction in punctuation is represented in speech by a difference in tone of voice. In addition, English restrictive relative clauses can be introduced by “that” as well as “which” and “who”, whereas incidental relative clauses cannot begin with “that”. Lojban, however, always uses the cmavo poi and noi rather than punctuation or intonation to make the distinction.

Here are more examples of incidental relative clauses:

In this example, mi is already sufficiently restricted, and the additional information that I am a judge is being provided solely for the listener's edification.

In Example 8.15, the speaker is presumed to have only one car, and is providing incidental information that it is white. (Alternatively, he or she might have more than one car, since le karce can be plural, in which case the incidental information is that each of them is white.) Contrast Example 8.16 with a restrictive relative clause:

Here the speaker probably has several cars, and is restricting the referent of the sumti le mi karce (and thereby the listener's attention) to the white one only. Example 8.16 means much the same as Example 8.17, which does not use a relative clause:

So a restrictive relative clause attached to a description can often mean the same as a description involving a tanru. However, blabi karce, like all tanru, is somewhat vague: in principle, it might refer to a car which carries white things, or even express some more complicated concept involving whiteness and car-ness; the restrictive relative clause of Example 8.16 can only refer to a car which is white, not to any more complex or extended concept.

The following cmavo are discussed in this section:

There are types of relative clauses (those which have a certain selbri) which are frequently wanted in Lojban, and can be expressed using a shortcut called a relative phrase. Relative phrases are introduced by cmavo of selma'o GOI, and consist of a GOI cmavo followed by a single sumti.

Here is an example of pe, plus an equivalent sentence using a relative clause:

In Example 8.18 and Example 8.19, the link between the chair and the speaker is of the loosest kind.

Here is an example of po:

Example 8.20 and Example 8.21 contrast with Example 8.18 and Example 8.19: the chair is more permanently connected with the speaker. A plausible (though not the only possible) contrast between Example 8.18 and Example 8.20 is that pe mi would be appropriate for a chair the speaker is currently sitting on (whether or not the speaker owned that chair), and po mi for a chair owned by the speaker (whether or not he or she was currently occupying it).

As a result, the relationship expressed between two sumti by po is usually called “possession”, although it does not necessarily imply ownership, legal or otherwise. The central concept is that of specificity (steci in Lojban).

Here is an example of po'e, as well as another example of po:

Example 8.22 and Example 8.23 on the one hand, and Example 8.24 on the other, illustrate the contrast between two types of possession called “intrinsic” and “extrinsic”, or sometimes “inalienable” and “alienable”, respectively. Something is intrinsically (or inalienably) possessed by someone if the possession is part of the possessor, and cannot be changed without changing the possessor. In the case of Example 8.22, people are usually taken to intrinsically possess their arms: even if an arm is cut off, it remains the arm of that person. (If the arm is transplanted to another person, however, it becomes intrinsically possessed by the new user, though, so intrinsic possession is a matter of degree.)

By contrast, the bottle of Example 8.24 can be given away, or thrown away, or lost, or stolen, so it is possessed extrinsically (alienably). The exact line between intrinsic and extrinsic possession is culturally dependent. The U.S. Declaration of Independence speaks of the “inalienable rights” of men, but just what those rights are, and even whether the concept makes sense at all, varies from culture to culture.

Note that Example 8.22 can also be expressed without a relative clause:

reflecting the fact that the gismu birka has an x2 place representing the body to which the arm belongs. Many, but not all, cases of intrinsic possession can be thus covered without using po'e by placing the possessor into the appropriate place of the description selbri.

Here is an example of po'u:

The cmavo po'u does not represent possession at all, but rather identity. (Note that it means poi du and its form was chosen to suggest the relationship.)

In Example 8.26, the use of po'u tells us that le gerku and le mi pendo represent the same thing. Consider the contrast between Example 8.26 and:

The facts of the case are the same, but the listener's knowledge about the situation may not be. In Example 8.26, the listener is presumed not to understand which dog is meant by le gerku, so the speaker adds a relative phrase clarifying that it is the particular dog which is the speaker's friend.

Example 8.28, however, assumes that the listener does not know which of the speaker's friends is referred to, and specifies that it is the friend that is the dog (which dog is taken to be obvious). Here is another example of the same contrast:

The principle that the possessor and the possessed may change places applies to all the GOI cmavo, and allows for the possibility of odd effects:

Example 8.31 is useful in a context which is about my friend, and states that his or her cup is small, whereas Example 8.32 is useful in a context that is primarily about a certain cup, and makes a claim about “my friend of the cup”, as opposed to some other friend of mine. Here the cup appears to “possess” the person! English can't even express this relationship with a possessive – “the cup's friend of mine” looks like nonsense – but Lojban has no trouble doing so.

Finally, the cmavo ne and no'u stand to pe and po'u, respectively, as noi does to poi- they provide incidental information:

In Example 8.33, the white dog is already fully identified (after all, presumably the listener knows which dog bit him or her!). The fact that it is yours is merely incidental to the main bridi claim.

Distinguishing between po'u and no'u can be a little tricky. Consider a room with several men in it, one of whom is named Jim. If you don't know their names, I might say:

Here I am saying that one of the men is a poet, and incidentally telling you that he is Jim. But if you do know the names, then

is appropriate. Now I am using the fact that the man I am speaking of is Jim in order to pick out which man I mean.

It is worth mentioning that English sometimes over-specifies possession from the Lojban point of view (and the point of view of many other languages, including ones closely related to English). The idiomatic English sentence

seems strange to a French- or German-speaking person: whose pockets would he put his hands into? and even odder, whose hands would he put into his pockets? In Lojban, the sentence

is very natural. Of course, if the man is in fact putting his hands into another's pockets, or another's hands into his pockets, the fact can be specified.

Finally, the elidable terminator for GOI cmavo is ge'u of selma'o GEhU; it is almost never required. However, if a logical connective immediately follows a sumti modified by a relative phrase, then an explicit ge'u is needed to allow the connective to affect the relativized sumti rather than the sumti of the relative phrase. (What about the cmavo after which selma'o GOI is named? It is discussed in Section 7.5, as it is not semantically akin to the other kinds of relative phrases, although the syntax is the same.)

Sometimes it is necessary or useful to attach more than one relative clause to a sumti. This is made possible in Lojban by the cmavo zi'e (of selma'o ZIhE), which is used to join one or more relative clauses together into a single unit, thus making them apply to the same sumti. For example:

The most usual translation of zi'e in English is “and”, but zi'e is not really a logical connective: unlike most of the true logical connectives (which are explained in Chapter 14), it cannot be converted into a logical connection between sentences.

It is perfectly correct to use zi'e to connect relative clauses of different kinds:

In Example 8.39, the restrictive clause poi blabi specifies which dog is referred to, but the incidental clause noi le mi pendo cu ponse is mere incidental information: the listener is supposed to already have identified the dog from the poi blabi. Of course, the meaning (though not necessarily the emphasis) is the same if the incidental clause appears first.

It is also possible to connect relative phrases with zi'e, or a relative phrase with a relative clause:

Note that if the colloquial translation of Example 8.40 were “My bottle, which is blue, is broken”, then noi rather than poi would have been correct in the Lojban version, since that version of the English implies that you do not need to know the bottle is blue. As written, Example 8.40 suggests that I probably have more than one bottle, and the one in question needs to be picked out as the blue one.

Example 8.41 illustrates that more than two relative phrases or clauses can be connected with zi'e. It almost defies colloquial translation because of the very un-English contrast between pe mi, implying that the chair is temporarily connected with me, and po do, implying that the chair has a more permanent association with you. (Perhaps I am a guest in your house, in which case the chair would naturally be your property.)

Here is another example, mixing a relative phrase and two relative clauses, a restrictive one and a non-restrictive one:

There is another member of selma'o NOI which serves to introduce a third kind of relative clause: voi. Relative clauses introduced by voi are restrictive, like those introduced by poi. However, there is a fundamental difference between poi and voi relative clauses. A poi relative clause is said to be veridical, in the same sense that a description using lo or loi is: it is essential to the interpretation that the bridi actually be true. For example:

it must actually be true that the dog is white, or the sentence constitutes a miscommunication. If there is a white dog and a brown dog, and the speaker uses le gerku poi blabi to refer to the brown dog, then the listener will not understand correctly. However,

puts the listener on notice that the dog in question may not actually meet objective standards (whatever they are) for being white: only the speaker can say exactly what is meant by the term. In this way, voi is like le; the speaker's intention determines the meaning.

As a result, the following two sentences

mean essentially the same thing (except that Example 8.46 involves pointing thanks to the use of ti, whereas Example 8.45 doesn't), and neither one is self-contradictory: it is perfectly all right to describe something as a man (although perhaps confusing to the listener) even if it actually is a woman.

So far, this chapter has described the various kinds of relative clauses (including relative phrases). The list is now complete, and the rest of the chapter will be concerned with the syntax of sumti that include relative clauses. So far, all relative clauses have appeared directly after the sumti to which they are attached. This is the most common position (and originally the only one), but a variety of other placements are also possible which produce a variety of semantic effects.

There are actually three places where a relative clause can be attached to a description sumti: after the descriptor (le, lo, or whatever), after the embedded selbri but before the elidable terminator (which is ku), and after the ku. The relative clauses attached to descriptors that we have seen have occupied the second position. Thus Example 8.43, if written out with all elidable terminators, would appear as:

Here ku'o is the terminator paired with poi and ku with le, and vau is the terminator of the whole bridi.

When a simple descriptor using le, like le gerku, has a relative clause attached, it is purely a matter of style and emphasis where the relative clause should go. Therefore, the following examples are all equivalent in meaning to Example 8.47:

Example 8.47 will seem most natural to speakers of languages like English, which always puts relative clauses after the noun phrases they are attached to; Example 8.48, on the other hand, may seem more natural to Finnish or Chinese speakers, who put the relative clause first. Note that in Example 8.48, the elidable terminator ku'o must appear, or the selbri of the relative clause (blabi) will merge with the selbri of the description (gerku), resulting in an ungrammatical sentence. The purpose of the form appearing in Example 8.49 will be apparent shortly.

As is explained in detail in Section 6.7, two different numbers (known as the “inner quantifier” and the “outer quantifier”) can be attached to a description. The inner quantifier specifies how many things the descriptor refers to: it appears between the descriptor and the description selbri. The outer quantifier appears before the descriptor, and specifies how many of the things referred to by the descriptor are involved in this particular bridi. In the following example,

mu is the inner quantifier and re is the outer quantifier. Now what is meant by attaching a relative clause to the sumti re le mu prenu? Suppose the relative clause is poi ninmu (meaning “who are women”). Now the three possible attachment points discussed previously take on significance.

As the parentheses show, Example 8.52 means that all five of the persons are women, whereas Example 8.53 means that the two who are going to the market are women. How do we remember which is which? If the relative clause comes after the explicit ku, as in Example 8.53, then the sumti as a whole is qualified by the relative clause. If there is no ku, or if the relative clause comes before an explicit ku, then the relative clause is understood to apply to everything which the underlying selbri applies to.

What about Example 8.51? By convention, it means the same as Example 8.53, and it requires no ku, but it does typically require a ku'o instead. Note that the relative clause comes before the inner quantifier.

When le is the descriptor being used, and the sumti has no explicit outer quantifier, then the outer quantifier is understood to be ro (meaning “all”), as is explained in Section 6.7. Thus le gerku is taken to mean “all of the things I refer to as dogs”, possibly all one of them. In that case, there is no difference between a relative clause after the ku or before it. However, if the descriptor is lo, the difference is quite important:

Both Example 8.54 and Example 8.55 tell us that one or more persons are going to the market. However, they make very different incidental claims. Now, what does lo prenu noi blabi mean? Well, the default inner quantifier is ro (meaning “all”), and the default outer quantifier is su'o (meaning “at least one”). Therefore, we must first take all persons, then choose at least one of them. That one or more people will be going.

In Example 8.54, the relative clause described the sumti once the outer quantifier was applied: one or more people, who are white, are going. But in Example 8.55, the relative clause actually describes the sumti before the outer quantification is applied, so that it ends up meaning “First take all persons – by the way, they're all white”. But not all people are white, so the incidental claim being made here is false.

The safe strategy, therefore, is to always use ku when attaching a noi relative clause to a lo descriptor. Otherwise we may end up claiming far too much.

When the descriptor is la, indicating that what follows is a selbri used for naming, then the positioning of relative clauses has a different significance. A relative clause inside the ku, whether before or after the selbri, is reckoned part of the name; a relative clause outside the ku is not. Therefore,

says that the speaker sees a person with a particular name, who does not necessarily fear any horses, whereas

refers to one (or more) of those named “Man”, namely the one(s) who are afraid of their horses.

Finally, so-called indefinite sumti like re karce, which means almost the same as re lo karce (which in turn means the same as re lo ro karce), can have relative clauses attached; these are taken to be of the outside-the- ku variety. Here is an example:

The restrictive relative clause only affects the two cars being affected by the main bridi, not all cars that exist. It is ungrammatical to try to place a relative clause within an indefinite sumti (that is, before an explicitly expressed terminating ku.) Use an explicit lo instead.

In Example 8.15 through Example 8.17, the sumti le mi karce appears, glossed as “my car”. Although it might not seem so, this sumti actually contains a relative phrase. When a sumti appears between a descriptor and its description selbri, it is actually a pe relative phrase. So

and

mean exactly the same thing. Furthermore, since there are no special considerations of quantifiers here,

means the same thing as well. A sumti like the one in Example 8.59 is called a “possessive sumti”. Of course, it does not really indicate possession in the sense of ownership, but like pe relative phrases, indicates only weak association; you can say le mi karce even if you've only borrowed it for the night. (In English, “my car” usually means le karce po mi, but we do not have the same sense of possession in “my seat on the bus”; Lojban simply makes the weaker sense the standard one.) The inner sumti, mi in Example 8.59, is correspondingly called the “possessor sumti”.

Historically, possessive sumti existed before any other kind of relative phrase or clause, and were retained when the machinery of relative phrases and clauses as detailed in this chapter so far was slowly built up. When preposed relative clauses of the Example 8.60 type were devised, possessive sumti were most easily viewed as a special case of them.

Although any sumti, however complex, can appear in a full-fledged relative phrase, only simple sumti can appear as possessor sumti, without a pe. Roughly speaking, the legal possessor sumti are: pro-sumti, quotations, names and descriptions, and numbers. In addition, the possessor sumti may not be preceded by a quantifier, as such a form would be interpreted as the unusual “descriptor + quantifier + sumti” type of description. All these sumti forms are explained in full in Chapter 6.

Here is an example of a description used in a possessive sumti:

Note the explicit ku at the end of the possessor sumti, which prevents the selbri of the possessor sumti from merging with the selbri of the main description sumti. Because of the need for this ku, the most common kind of possessor sumti are pro-sumti, especially personal pro-sumti, which require no elidable terminator. Descriptions are more likely to be attached with relative phrases.

And here is a number used as a possessor sumti:

which is not quite the same as “the fifth juror”; it simply indicates a weak association between the particular juror and the number 5.

A possessive sumti may also have regular relative clauses attached to it. This would need no comment if it were not for the following special rule: a relative clause immediately following the possessor sumti is understood to affect the possessor sumti, not the possessive. For example:

means that my car isn't going; the incidental claim of noi sipna applies to me, not my car, however. If I wanted to say that the car is sleeping (whatever that might mean) I would need:

Note that Example 8.64 uses vau rather than ku'o at the end of the relative clause: this terminator ends every simple bridi and is almost always elidable; in this case, though, it is a syllable shorter than the equally valid alternative, ku'o.

The following cmavo is discussed in this section:

Normally, relative clauses attach only to simple sumti or parts of sumti: pro-sumti, names and descriptions, pure numbers, and quotations. An example of a relative clause attached to a pure number is:

And here is an incidental relative clause attached to a quotation:

which may serve to identify the author of the quotation or some other relevant, but subsidiary, fact about it. All such relative clauses appear only after the simple sumti, never before it.

In addition, sumti with attached sumti qualifiers of selma'o LAhE or NAhE+BO (which are explained in detail in Section 6.10) can have a relative clause appearing after the qualifier and before the qualified sumti, as in:

Example 8.68 is a bit complex, and may need some picking apart. The quotation lu le xunre cmaxirma li'u means the string of words “The Red Pony”. If the la'e at the beginning of the sentence were omitted, Example 8.68 would claim that a certain string of words is in a room distant from the speaker. But obviously a string of words can't be in a room! The effect of the la'e is to modify the sumti so that it refers not to the words themselves, but to the referent of those words, a novel by John Steinbeck (presumably in Lojban translation). The particular copy of “The Red Pony” is identified by the restrictive relative clause. Example 8.68 means exactly the same as:

and the two sentences can be considered stylistic variants. Note the required lu'u terminator, which prevents the relative clause from attaching to the quotation itself: we do not wish to refer to an old quotation!

Sometimes, however, it is important to make a relative clause apply to the whole of a more complex sumti, one which involves logical or non-logical connection (explained in Chapter 14). For example,

The incidental claim in Example 8.70 is not that Frank and George are men, but only that George is a man, because the incidental relative clause attaches only to la djordj, the immediately preceding simple sumti.

To make a relative clause attach to both parts of the logically connected sumti in Example 8.70, a new cmavo is needed, vu'o (of selma'o VUhO). It is placed between the sumti and the relative clause, and extends the sphere of influence of that relative clause to the entire preceding sumti, including however many logical or non-logical connectives there may be.

The presence of vu'o here means that the relative clause noi nanmu extends to the entire logically connected sumti la frank. .e la djordj.; in other words, both Frank and George are claimed to be men, as the colloquial translation shows.

English is able to resolve the distinction correctly in the case of Example 8.70 and Example 8.71 by making use of number: “who is” rather than “who are”. Lojban doesn't distinguish between singular and plural verbs: nanmu can mean “is a man” or “are men”, so another means is required. Furthermore, Lojban's mechanism works correctly in general: if nanmu (meaning “is-a-man”) were replaced with pu bajra (“ran”), English would have to make the distinction some other way:

In spoken English, tone of voice would serve; in written English, one or both sentences would need rewriting.

Vocative phrases are explained in more detail in Section 6.11. Briefly, they are a method of indicating who a sentence or discourse is addressed to: of identifying the intended listener. They take three general forms, all beginning with cmavo from selma'o COI or DOI (called “vocative words”; there can be one or many), followed by either a name, a selbri, or a sumti. Here are three examples:

Note that Example 8.75 says farewell to something which doesn't really have to be a horse, something that the speaker simply thinks of as being a horse, or even might be something (a person, for example) who is named “Horse”. In a sense, Example 8.75 is ambiguous between co'o le xirma and co'o la xirma, a relatively safe semantic ambiguity, since names are ambiguous in general: saying “George” doesn't distinguish between the possible Georges.

Similarly, Example 8.74 can be thought of as an abbreviation of:

Syntactically, vocative phrases are a kind of free modifier, and can appear in many places in Lojban text, generally at the beginning or end of some complete construct; or, as in Example 8.74 to Example 8.76, as sentences by themselves.

As can be seen, the form of vocative phrases is similar to that of sumti, and as you might expect, vocative phrases allow relative clauses in various places. In vocative phrases which are simple names (after the vocative words), any relative clauses must come just after the names:

The restrictive relative clause in Example 8.78 suggests that there is some other Frank (perhaps on the Green Team) from whom this Frank, the one the speaker is greeting, must be distinguished.

A vocative phrase containing a selbri can have relative clauses either before or after the selbri; both forms have the same meaning. Here are some examples:

Example 8.79 and Example 8.80 mean the same thing. In fact, relative clauses can appear in both places.

For the most part, these are straightforward and uncomplicated: a sumti that is part of a relative clause bridi may itself be modified by a relative clause:

However, an ambiguity can exist if ke'a is used in a relative clause within a relative clause: does it refer to the outermost sumti, or to the sumti within the outer relative clause to which the inner relative clause is attached? The latter. To refer to the former, use a subscript on ke'a:

Here, the meaning of “IT-sub-2” is that sumti attached to the second relative clause, counting from the innermost, is used. Therefore, ke'axipa (IT-sub-1) means the same as plain ke'a.

Alternatively, you can use a prenex (explained in full in Chapter 16), which is syntactically a series of sumti followed by the special cmavo zo'u, prefixed to the relative clause bridi:

Example 8.83 is more verbose than Example 8.82, but may be clearer, since it explicitly spells out the two ke'a cmavo, each on its own level, and assigns them to the assignable cmavo ko'a and ko'e (explained in Section 7.5).

Relative clause introducers (selma'o NOI):

Relative phrase introducers (selma'o GOI):

Relativizing pro-sumti (selma'o KOhA):

Relative clause joiner (selma'o ZIhE):

Relative clause associator (selma'o VUhO):

Elidable terminators (each its own selma'o):