Chapter 19. Putting It All Together: Notes on the Structure of Lojban Texts

This chapter is incurably miscellaneous. It describes the cmavo that specify the structure of Lojban texts, from the largest scale (paragraphs) to the smallest (single words). There are fewer examples than are found in other chapters of this book, since the linguistic mechanisms described are generally made use of in conversation or else in long documents.

This chapter is also not very self-contained. It makes passing reference to a great many concepts which are explained in full only in other chapters. The alternative would be a chapter on text structure which was as complex as all the other chapters put together. Lojban is a unified language, and it is not possible to understand any part of it (in full) before understanding every part of it (to some degree).

The following cmavo is discussed in this section:

Since Lojban is audio-visually isomorphic, there needs to be a spoken and written way of signaling the end of a sentence and the start of the following one. In written English, a period serves this purpose; in spoken English, a tone contour (rising or falling) usually does the job, or sometimes a long pause. Lojban uses a single separator: the cmavo i (of selma'o I):

The word “separator” should be noted. i is not normally used after the last sentence nor before the first one, although both positions are technically grammatical. i signals a new sentence on the same topic, not necessarily by the same speaker. The relationship between the sentences is left vague, except in stories, where the relationship usually is temporal, and the following sentence states something that happened after the previous sentence.

Note that although the first letter of an English sentence is capitalized, the cmavo i is never capitalized. In writing, it is appropriate to place extra space before i to make it stand out better for the reader. In some styles of Lojban writing, every i is placed at the beginning of a line, possibly leaving space at the end of the previous line.

An i cmavo may or may not be used when the speaker of the following sentence is different from the speaker of the preceding sentence, depending on whether the sentences are felt to be connected or not.

An i cmavo can be compounded with a logical or non-logical connective (a jek or joik), a modal or tense connective, or both: these constructs are explained in Section 9.8, Section 10.16, and Section 14.4. In all cases, the i comes first in the compound. Attitudinals can also be attached to an i if they are meant to apply to the whole sentence: see Section 13.9.

There exist a pair of mechanisms for binding a sequence of sentences closely together. If the i (with or without connectives) is followed by bo (of selma'o BO), then the two sentences being separated are understood to be more closely grouped than sentences connected by i alone.

Similarly, a group of sentences can be preceded by tu'e (of selma'o TUhE) and followed by tu'u (of selma'o TUhU) to fuse them into a single unit. A common use of tu'e…tu'u is to group the sentences which compose a poem: the title sentence would precede the group, separated from it by i. Another use might be a set of directions, where each numbered direction might be surrounded by tu'e…tu'u and contain one or more sentences separated by i. Grouping with tu'e and tu'u is analogous to grouping with ke and ke'e to establish the scope of logical or non-logical connectives (see Section 14.8).

The following cmavo are discussed in this section:

The paragraph is a concept used in writing systems for two purposes: to indicate changes of topic, and to break up the hard-to-read appearance of large blocks of text on the page. The former function is represented in both spoken and written Lojban by the cmavo ni'o and no'i, both of selma'o NIhO. Of these two, ni'o is the more common. By convention, written Lojban is broken into paragraphs just before any ni'o or no'i, but a very long passage on a single topic might be paragraphed before an i. On the other hand, it is conventional in English to start a new paragraph in dialogue when a new speaker starts, but this convention is not commonly observed in Lojban dialogues. Of course, none of these conventions affect meaning in any way.

A ni'o can take the place of an i as a sentence separator, and in addition signals a new topic or paragraph. Grammatically, any number of ni'o cmavo can appear consecutively and are equivalent to a single one; semantically, a greater number of ni'o cmavo indicates a larger-scale change of topic. This feature allows complexly structured text, with topics, subtopics, and sub-subtopics, to be represented clearly and unambiguously in both spoken and written Lojban. However, some conventional differences do exist between ni'o in writing and in conversation.

In written text, a single ni'o is a mere discursive indicator of a new subject, whereas ni'oni'o marks a change in the context. In this situation, ni'oni'o implicitly cancels the definitions of all pro-sumti of selma'o KOhA as well as pro-bridi of selma'o GOhA. (Explicit cancelling is expressed by the cmavo da'o of selma'o DAhO, which has the free grammar of an indicator – it can appear almost anywhere.) The use of ni'oni'o does not affect indicators (of selma'o UI) or tense references, but ni'oni'oni'o, indicating a drastic change of topic, would serve to reset both indicators and tenses. (See Section 19.8 for a discussion of indicator scope.)

In spoken text, which is inherently less structured, these levels are reduced by one, with ni'o indicating a change in context sufficient to cancel pro-sumti and pro-bridi assignment. On the other hand, in a book, or in stories within stories such as “The Arabian Nights”, further levels may be expressed by extending the ni'o string as needed. Normally, a written text will begin with the number of ni'o cmavo needed to signal the largest scale division which the text contains. ni'o strings may be subscripted to label each context of discourse: see Section 19.6.

no'i is similar in effect to ni'o, but indicates the resumption of a previous topic. In speech, it is analogous to (but much shorter than) such English discursive phrases as “But getting back to the point ...”. By default, the topic resumed is that in effect before the last ni'o. When subtopics are nested within topics, then no'i would resume the previous subtopic and no'ino'i the previous topic. Note that no'i also resumes tense and pro-sumti assignments dropped at the previous ni'o.

If a ni'o is subscripted, then a no'i with the same subscript is assumed to be a continuation of it. A no'i may also have a negative subscript, which would specify counting backwards a number of paragraphs and resuming the topic found thereby.

The following cmavo is discussed in this section:

The normal Lojban sentence is just a bridi, parallel to the normal English sentence which has a subject and a predicate:

In Chinese, the normal sentence form is different: a topic is stated, and a comment about it is made. (Japanese also has the concept of a topic, but indicates it by attaching a suffix; other languages also distinguish topics in various ways.) The topic says what the sentence is about:

The colon in the first two versions of Example 19.3 separate the topic (“this news”) from the comment (“I know already”).

Lojban uses the cmavo zo'u (of selma'o ZOhU) to separate topic (a sumti) from comment (a bridi):

Example 19.4 is the literal Lojban translation of Example 19.3. Of course, the topic-comment structure can be changed to a straightforward bridi structure:

Example 19.5 means the same as Example 19.4, and it is simpler. However, often the position of the topic in the place structure of the selbri within the comment is vague:

Is the fish eating or being eaten? The sentence doesn't say. The Chinese equivalent of Example 19.6 is:

which is vague in exactly the same way.

Grammatically, it is possible to have more than one sumti before zo'u. This is not normally useful in topic-comment sentences, but is necessary in the other use of zo'u: to separate a quantifying section from a bridi containing quantified variables. This usage belongs to a discussion of quantifier logic in Lojban (see Section 16.2), but an example would be:

The string of sumti before zo'u (called the “prenex”: see Section 16.2) may contain both a topic and bound variables:

To specify a topic which affects more than one sentence, wrap the sentences in tu'e…tu'u brackets and place the topic and the zo'u directly in front. This is the exception to the rule that a topic attaches directly to a sentence:

Note: In Lojban, you do not “want money”; you “want to have money” or something of the sort, as the x2 place of djica demands an event. As a result, the straightforward rendering of Example 19.9 without a topic is not:

where ri means loi jdini and is interpreted as “the mere existence of money”, but rather:

namely, the possession of money. But topic-comment sentences like Example 19.10 are inherently vague, and this difference between ponse (which expects a physical object in x2) and djica is ignored. See Example 19.45 for another topic/comment sentence.

The subject of an English sentence is often the topic as well, but in Lojban the sumti in the x1 place is not necessarily the topic, especially if it is the normal (unconverted) x1 for the selbri. Thus Lojban sentences don't necessarily have a “subject” in the English sense.

The following cmavo are discussed in this section:

Lojban questions are not at all like English questions. There are two basic types: truth questions, of the form “Is it true that ...”, and fill-in-the-blank questions. Truth questions are marked by preceding the bridi, or following any part of it specifically questioned, with the cmavo xu (of selma'o UI):

(Since the Lojban is tenseless, either colloquial translation might be correct.) Truth questions are further discussed in Section 15.8.

Fill-in-the-blank questions have a cmavo representing some Lojban word or phrase which is not known to the questioner, and which the answerer is to supply. There are a variety of cmavo belonging to different selma'o which provide different kinds of blanks.

Where a sumti is not known, a question may be formed with ma (of selma'o KOhA), which is a kind of pro-sumti:

Of course, the ma need not be in the x1 place:

The answer is a simple sumti:

A sumti, then, is a legal utterance, although it does not by itself constitute a bridi – it does not claim anything, but merely completes the open-ended claim of the previous bridi.

There can be two ma cmavo in a single question:

and the answer would be two sumti, which are meant to fill in the two ma cmavo in order:

An even more complex example, depending on the non-logical connective fa'u (of selma'o JOI), which is like the English “and ... respectively”:

An answer might be

(Note: A mechanical substitution of Example 19.20 into Example 19.19 produces an ungrammatical result, because * ... le zarci fa'u le briju is ungrammatical Lojban: the first le zarci has to be closed with its proper terminator ku, for reasons explained in Section 14.14. This effect is not important: Lojban behaves as if all elided terminators have been supplied in both question and answer before inserting the latter into the former. The exchange is grammatical if question and answer are each separately grammatical.)

Questions to be answered with a selbri are expressed with mo of selma'o GOhA, which is a kind of pro-bridi:

Here the answerer is to supply some predicate which is true of Lojban. Such questions are extremely open-ended, due to the enormous range of possible predicate answers. The answer might be just a selbri, or might be a full bridi, in which case the sumti in the answer override those provided by the questioner. To limit the range of a mo question, make it part of a tanru.

Questions about numbers are expressed with xo of selma'o PA:

The answer would be a simple number, another kind of non-bridi utterance:

Fill-in-the-blank questions may also be asked about: logical connectives (using cmavo ji of A, ge'i of GA, gi'i of GIhA, gu'i of GUhA, or je'i of JA, and receiving an ek, gihek, ijek, or ijoik as an answer) – see Section 14.13; attitudes (using pei of UI, and receiving an attitudinal as an answer) – see Section 13.10; place structures (using fi'a of FA, and receiving a cmavo of FA as an answer) – see Section 9.3; tenses and modals (using cu'e of CUhE, and receiving any tense or BAI cmavo as an answer) – see Section 9.6 and Chapter 10.

Questions can be marked by placing pau (of selma'o UI) before the question bridi. See Section 13.13 for details.

The full list of non-bridi utterances suitable as answers to questions is:

At the beginning of a text, the following non-bridi are also permitted:

Where not needed for the expression of answers, most of these are made grammatical for pragmatic reasons: people will say them in conversation, and there is no reason to rule them out as ungrammatical merely because most of them are vague.

The following cmavo is discussed in this section:

The cmavo xi (of selma'o XI) indicates that a subscript (a number, a lerfu string, or a parenthesized mekso) follows. Subscripts can be attached to almost any construction and are placed following the construction (or its terminator word, which is generally required). They are useful either to extend the finite cmavo list to infinite length, or to make more refined distinctions than the standard cmavo list permits. The remainder of this section mentions some places where subscripts might naturally be used.

Lojban gismu have at most five places:

Consequently, selma'o SE (which operates on a selbri to change the order of its places) and selma'o FA (which provides place number tags for individual sumti) have only enough members to handle up to five places. Conversion of Example 19.24, using xe to swap the x1 and x5 places, would produce:

And reordering of the place structures might produce:

Example 19.24 to Example 19.26 all mean the same thing. But consider the lujvo nunkla, formed by applying the abstraction operator nu to klama:

Example 19.27 shows that nunkla has six places: the five places of klama plus a new one (placed first) for the event itself. Performing transformations similar to that of Example 19.25 requires an additional conversion cmavo that exchanges the x1 and x6 places. The solution is to use any cmavo of SE with a subscript "6" (Section 19.6):

Likewise, a sixth place tag can be created by using any cmavo of FA with a subscript:

Example 19.27 to Example 19.29 also all mean the same thing, and each is derived straightforwardly from any of the others, despite the tortured nature of the English glosses. In addition, any other member of SE or FA could be substituted into sexixa and faxixa without change of meaning: vexixa means the same thing as sexixa.

Lojban provides two groups of pro-sumti, both belonging to selma'o KOhA. The ko'a-series cmavo are used to refer to explicitly specified sumti to which they have been bound using goi. The da-series, on the other hand, are existentially or universally quantified variables. (These concepts are explained more fully in Chapter 16.) There are ten ko'a-series cmavo and 3 da-series cmavo available.

If more are required, any cmavo of the ko'a-series or the da-series can be subscripted:

is the 4th bound variable of the 1st sequence of the da-series, and

is the 18th free variable of the 3rd sequence of the ko'a-series. This convention allows 10 sequences of ko'a-type pro-sumti and 3 sequences of da-type pro-sumti, each with as many members as needed. Note that daxivo and dexivo are considered to be distinct pro-sumti, unlike the situation with sexixa and vexixa above. Exactly similar treatment can be given to the bu'a-series of selma'o GOhA and to the gismu pro-bridi broda, brode, brodi, brodo, and brodu.

Subscripts on lerfu words are used in the standard mathematical way to extend the number of variables:

and can be used to extend the number of pro-sumti as well, since lerfu strings outside mathematical contexts are grammatically and semantically equivalent to pro-sumti of the ko'a-series. (In Example 19.32, note the required terminator boi after each xy. cmavo; this terminator allows the subscript to be attached without ambiguity.)

Names, which are similar to pro-sumti, can also be subscripted to distinguish two individuals with the same name:

Subscripts on tenses allow talking about more than one time or place that is described by the same general cmavo. For example, puxipa could refer to one point in the past, and puxire a second point (earlier or later).

You can place a subscript on the word ja'a, the bridi affirmative of selma'o NA, to express so-called fuzzy truths. The usual machinery for fuzzy logic (statements whose truth value is not merely “true” or “false”, but is expressed by a number in the range 0 to 1) in Lojban is the abstractor jei:

However, by convention we can attach a subscript to ja'a to indicate fuzzy truth (or to na if we change the amount):

Finally, as mentioned in Section 19.2, ni'o and no'i cmavo with matching subscripts mark the start and the continuation of a given topic respectively. Different topics can be assigned to different subscripts.

Other uses of subscripts will doubtless be devised in future.

The following cmavo are discussed in this section:

Numerical free modifiers, corresponding to English “firstly”, “secondly”, and so on, can be created by suffixing mai or mo'o of selma'o MAI to a number or a lerfu string. Here are some examples:

This does not imply that I go to the store before I go to the house: that meaning requires a tense. The sumti are simply numbered for convenience of reference. Like other free modifiers, the utterance ordinals can be inserted almost anywhere in a sentence without affecting its grammar or its meaning.

Any of the Lojban numbers can be used with MAI: romai, for example, means “all-thly” or “lastly”. Likewise, if you are enumerating a long list and have forgotten which number is wanted next, you can say ny.mai, or “Nthly”.

The difference between mai and mo'o is that mo'o enumerates larger subdivisions of a text; mai was designed for lists of numbered items, whereas mo'o was intended to subdivide structured works. If this chapter were translated into Lojban, it might number each section with mo'o: this section would then be introduced with zemo'o, or “Section 7.”

The following cmavo are discussed in this section:

Lojban has a complex system of “attitudinals”, words which indicate the speaker's attitude to what is being said. The attitudinals include indicators of emotion, intensity markers, discursives (which show the structure of discourse), and evidentials (which indicate “how the speaker knows”). Most of these words belong to selma'o UI; the intensity markers belong to selma'o CAI for historical reasons, but the two selma'o are grammatically identical. The individual cmavo of UI and CAI are discussed in Chapter 13; only the rules for applying them in discourse are presented here.

Normally, an attitudinal applies to the preceding word only. However, if the preceding word is a structural cmavo which begins or ends a whole construction, then that whole construction is affected by the attitudinal:

An attitudinal meant to cover a whole sentence can be attached to the preceding i, expressed or understood:

or to an explicit vau placed at the end of a bridi.

Likewise, an attitudinal meant to cover a whole paragraph can be attached to ni'o or no'i. An attitudinal at the beginning of a text applies to the whole text.

However, sometimes it is necessary to be more specific about the range of one or more attitudinals, particularly if the range crosses the boundaries of standard Lojban syntactic constructions. The cmavo fu'e (of selma'o FUhE) and fu'o (of selma'o FUhO) provide explicit scope markers. Placing fu'e in front of an attitudinal disconnects it from what precedes it, and instead says that it applies to all following words until further notice. The notice is given by fu'o, which can appear anywhere and cancels all in-force attitudinals. For example:

Here, only the blanu zdani portion of the three-part tanru blanu zdani ponse is marked as a belief of the speaker. Naturally, the attitudinal scope markers do not affect the rules for interpreting multi-part tanru: blanu zdani groups first because tanru group from left to right unless overridden with ke or bo.

Other attitudinals of more local scope can appear after attitudinals marked by FUhE; these attitudinals are added to the globally active attitudinals rather than superseding them.

The following cmavo are discussed in this section:

Grammatically, quotations are very simple in Lojban: all of them are sumti, and they all mean something like “the piece of text here quoted”:

But in fact there are four different flavors of quotation in the language, involving six cmavo of six different selma'o. This being the case, quotation deserves some elaboration.

The simplest kind of quotation, exhibited in Example 19.43, uses the cmavo lu (of selma'o LU) as the opening quotation mark, and the cmavo li'u (of selma'o LIhU) as the closing quotation mark. The text between lu and li'u must be a valid, parseable Lojban text. If the quotation is ungrammatical, so is the surrounding expression. The cmavo li'u is technically an elidable terminator, but it's almost never possible to elide it except at the end of text.

The cmavo lo'u (of selma'o LOhU) and le'u (of selma'o LEhU) are used to surround a quotation that is not necessarily grammatical Lojban. However, the text must consist of morphologically correct Lojban words (as defined in Chapter 4), so that the le'u can be picked out reliably. The words need not be meaningful, but they must be recognizable as cmavo, brivla, or cmene. Quotation with lo'u is essential to quoting ungrammatical Lojban for teaching in the language, the equivalent of the * that is used in English to mark such errors:

Example 19.44 is grammatical even though the embedded quotation is not. Similarly, lo'u quotation can quote fragments of a text which themselves do not constitute grammatical utterances:

Note the topic-comment formulation (Section 19.4) and the indicator applying to the selbri only (Section 19.8). Neither viska le nor viska lo is a valid Lojban utterance, and both require lo'u quotation.

Additionally, pro-sumti or pro-bridi in the quoting sentence can refer to words appearing in the quoted sentence when lu…li'u is used, but not when lo'u ... le'u is used:

In Example 19.46, ri is a pro-sumti which refers to the most recent previous sumti, namely le ninmu. Compare:

In Example 19.47, ri cannot refer to the referent of the alleged sumti le ninmu, because le ninmu cu morsi is a mere uninterpreted sequence of Lojban words. Instead, ri ends up referring to the referent of the sumti la tcarlis., and so it is Charlie who is alive.

The metalinguistic erasers si, sa, and su, discussed in Section 19.13, do not operate in text between lo'u and le'u. Since the first le'u terminates a lo'u quotation, it is not directly possible to have a lo'u quotation within another lo'u quotation. However, it is possible for a le'u to occur within a lo'u ... le'u quotation by preceding it with the cmavo zo, discussed in Section 19.10. Note that le'u is not an elidable terminator; it is required.

The following cmavo are discussed in this section:

The cmavo zo (of selma'o ZO) is a strong quotation mark for the single following word, which can be any Lojban word whatsoever. Among other uses, zo allows a metalinguistic word to be referenced without having it act on the surrounding text. The word must be a morphologically legal (but not necessarily meaningful) single Lojban word; compound cmavo are not permitted. For example:

Since zo acts on a single word only, there is no corresponding terminator. Brevity, then, is a great advantage of zo, since the terminators for other kinds of quotation are rarely or never elidable.

The cmavo zoi (of selma'o ZOI) is a quotation mark for quoting non-Lojban text. Its syntax is zoi X. text .X, where X is a Lojban word (called the delimiting word) which is separated from the quoted text by pauses, and which is not found in the written text or spoken phoneme stream. It is common, but not required, to use the lerfu word (of selma'o BY) which corresponds to the Lojban name of the language being quoted:

where gy stands for glico. Other popular choices of delimiting words are .kuot., a Lojban name which sounds like the English word “quote”, and the word zoi itself. Another possibility is a Lojban word suggesting the topic of the quotation.

Within written text, the Lojban written word used as a delimiting word may not appear, whereas within spoken text, the sound of the delimiting word may not be uttered. This leads to occasional breakdowns of audio-visual isomorphism: Example 19.50 is fine in speech but ungrammatical as written, whereas Example 19.51 is correct when written but ungrammatical in speech.

The text gy appears in the written word “gyrations”, whereas the sound represented in Lojban by jai appears in the spoken word “gyrations”. Such borderline cases should be avoided as a matter of good style.

It should be noted particularly that zoi quotation is the only way to quote rafsi, specifically CCV rafsi, because they are not Lojban words, and zoi quotation is the only way to quote things which are not Lojban words. (CVC and CVV rafsi look like names and cmavo respectively, and so can be quoted using other methods.) For example:

(A minor note on interaction between lo'u ... le'u and zoi: The text between lo'u and le'u should consist of Lojban words only. In fact, non-Lojban material in the form of a zoi quotation may also appear. However, if the word le'u is used either as the delimiting word for the zoi quotation, or within the quotation itself, the outer lo'u quotation will be prematurely terminated. Therefore, le'u should be avoided as the delimiting word in any zoi quotation.)

Lojban strictly avoids any confusion between things and the names of things:

In Example 19.53, zo .bab. is the word, whereas la bab. is the thing named by the word. The cmavo la'e and lu'e (of selma'o LAhE) convert back and forth between references and their referents:

Example 19.53 through Example 19.55 all mean approximately the same thing, except for differences in emphasis. Example 19.56 is different:

and says that Bob is both the name and the thing named, an unlikely situation. People are not names.

(In Example 19.53 through Example 19.54, the name bab. was separated from a preceding zo by a pause, thus: zo .bab.. The reason for this extra pause is that all Lojban names must be separated by pause from any preceding word other than la, lai, la'i (all of selma'o LA) and doi (of selma'o DOI). There are numerous other cmavo that may precede a name: of these, zo is one of the most common.)

The cmavo la'o also belongs to selma'o ZOI, and is mentioned here for completeness, although it does not signal the beginning of a quotation. Instead, la'o serves to mark non-Lojban names, especially the Linnaean binomial names (such as “Homo sapiens”) which are the internationally standardized names for species of animals and plants. Internationally known names which can more easily be recognized by spelling rather than pronunciation, such as “Goethe”, can also appear in Lojban text with la'o:

Using la'o for all names rather than Lojbanizing, however, makes for very cumbersome text. A rough equivalent of la'o might be la me zoi.

The following cmavo are discussed in this section:

English often uses strong stress on a word to single it out for contrastive emphasis, thus

is quite different from

The heavy stress on “George” (represented in writing by italics) indicates that I saw George rather than someone else. Lojban does not use stress in this way: stress is used only to help separate words (because every brivla is stressed on the penultimate syllable) and in names to match other languages' stress patterns. Note that many other languages do not use stress in this way either; typically word order is rearranged, producing something like

In Lojban, the cmavo ba'e (of selma'o BAhE) precedes a single word which is to be emphasized:

Note the pause before the name djordj., which serves to separate it unambiguously from the ba'e. Alternatively, the ba'e can be moved to a position before the la, which in effect emphasizes the whole construct la djordj.:

Marking a word with a cmavo of BAhE does not change the word's grammar in any way. Any word in a bridi can receive contrastive emphasis marking:

Emphasis on one of the structural components of a Lojban bridi can also be achieved by rearranging it into an order that is not the speaker's or writer's usual order. Any sumti moved out of place, or the selbri when moved out of place, is emphatic to some degree.

For completeness, the cmavo za'e should be mentioned, also of selma'o BAhE. It marks a word as possibly irregular, non-standard, or nonce (created for the occasion):

marks a Lojbanization of an English name, where a more appropriate standard form might be something like la ckiipyris., reflecting the country's name in Albanian.

Before a lujvo or fu'ivla, za'e indicates that the word has been made up on the spot and may be used in a sense that is not found in the unabridged dictionary (when we have an unabridged dictionary!).

The following cmavo are discussed in this section:

The cmavo to and toi are discursive (non-mathematical) parentheses, for inserting parenthetical remarks. Any text whatsoever can go within the parentheses, and it is completely invisible to its context. It can, however, refer to the context by the use of pro-sumti and pro-bridi: any that have been assigned in the context are still assigned in the parenthetical remarks, but the reverse is not true.

Example 19.66 implicitly redefines do within the parentheses: the listener is changed by doi frank. When the context sentence resumes, however, the old listener, Lisa, is automatically restored.

There is another cmavo of selma'o TO: to'i. The difference between to and to'i is the difference between parentheses and square brackets in English prose. Remarks within to ... toi cmavo are implicitly by the same speaker, whereas remarks within to'i ... toi are implicitly by someone else, perhaps an editor:

The sa'a suffix is a discursive cmavo (of selma'o UI) meaning “editorial insertion”, and indicating that the marked word or construct (in this case, the entire bracketed remark) is not part of the quotation. It is required whenever the to'i ... toi remark is physically within quotation marks, at least when speaking to literal-minded listeners; the convention may be relaxed if no actual confusion results.

Note: The parser believes that parentheses are attached to the previous word or construct, because it treats them as syntactic equivalents of subscripts and other such so-called “free modifiers”. Semantically, however, parenthetical remarks are not necessarily attached either to what precedes them or what follows them.

The cmavo sei (of selma'o SEI) begins an embedded discursive bridi. Comments added with sei are called “metalinguistic”, because they are comments about the discourse itself rather than about the subject matter of the discourse. This sense of the term “metalinguistic” is used throughout this chapter, and is not to be confused with the sense “language for expressing other languages”.

When marked with sei, a metalinguistic utterance can be embedded in another utterance as a discursive. In this way, discursives which do not have cmavo assigned in selma'o UI can be expressed:

Using the happiness attitudinal, .ui, would imply that the speaker was happy. Instead, the speaker attributes happiness to Frank. It would probably be safe to elide the one who is happy, and say:

The grammar of the bridi following sei has an unusual limitation: the sumti must either precede the selbri, or must be glued into the selbri with be and bei:

This restriction allows the terminator cmavo se'u to almost always be elided.

Since a discursive utterance is working at a “higher” level of abstraction than a non-discursive utterance, a non-discursive utterance cannot refer to a discursive utterance. Specifically, the various back-counting, reciprocal, and reflexive constructs in selma'o KOhA ignore the utterances at “higher” metalinguistic levels in determining their referent. It is possible, and sometimes necessary, to refer to lower metalinguistic levels. For example, the English “he said” in a conversation is metalinguistic. For this purpose, quotations are considered to be at a lower metalinguistic level than the surrounding context (a quoted text cannot refer to the statements of the one who quotes it), whereas parenthetical remarks are considered to be at a higher level than the context.

Lojban works differently from English in that the “he said” can be marked instead of the quotation. In Lojban, you can say:

which literally claims that John uttered the quoted text. If the central claim is that John made the utterance, as is likely in conversation, this style is the most sensible. However, in written text which quotes a conversation, you don't want the “he said” or “she said” to be considered part of the conversation. If unmarked, it could mess up the anaphora counting. Instead, you can use:

And of course other orders are possible:

Note the sa'a following each sei, marking the sei and its attached bridi as an editorial insert, not part of the quotation. In a more relaxed style, these sa'a cmavo would probably be dropped.

The elidable terminator for sei is se'u (of selma'o SEhU); it is rarely needed, except to separate a selbri within the sei comment from an immediately following selbri (or component) outside the comment.

The following cmavo are discussed in this section:

The cmavo si (of selma'o SI) is a metalinguistic operator that erases the preceding word, as if it had never been spoken:

means the same thing as ti mlatu. Multiple si cmavo in succession erase the appropriate number of words:

In order to erase the word zo, it is necessary to use three si cmavo in a row:

The first use of si does not erase anything, but completes the zo quotation. Two more si cmavo are then necessary to erase the first si and the zo.

Incorrect names can likewise cause trouble with si:

The Lojbanized spelling .esperanto breaks up, as a consequence of the Lojban morphology rules (see Chapter 4) into two Lojban words, the cmavo e and the undefined lujvo speranto. Therefore, two si cmavo are needed to erase them. Of course, .e speranto is not grammatical after la, but recognition of si is done before grammatical analysis.

Even more messy is the result of an incorrect zoi:

In Example 19.79, the first fy is taken to be the delimiting word. The next word must be different from the delimiting word, and gy., the Lojban name for the letter g, was chosen arbitrarily. Then the delimiting word must be repeated. For purposes of si erasure, the entire quoted text is taken to be a word, so four words have been uttered, and four more si cmavo are needed to erase them altogether. Similarly, a stray lo'u quotation mark must be erased with fy. le'u si si si, by completing the quotation and then erasing it all with three si cmavo.

What if less than the entire zo or zoi construct is erased? The result is something which has a loose zo or zoi in it, without its expected sequels, and which is incurably ungrammatical. Thus, to erase just the word quoted by zo, it turns out to be necessary to erase the zo as well:

The parser will reject zo .djan. si .djordj., because in that context djordj. is a name (of selma'o CMENE) rather than a quoted word.

Note: The current machine parser does not implement si erasure.

As the above examples plainly show, precise erasures with si can be extremely hard to get right. Therefore, the cmavo sa (of selma'o SA) is provided for erasing more than one word. The cmavo following sa should be the starting marker of some grammatical construct. The effect of the sa is to erase back to and including the last starting marker of the same kind. For example:

Since the word following sa is i, the sentence separator, its effect is to erase the preceding sentence. So Example 19.81 is equivalent to:

Another example, erasing a partial description rather than a partial sentence:

In Example 19.83, le blanu .zdan. is ungrammatical, but clearly reflects the speaker's original intention to say le blanu zdani. However, the zdani was cut off before the end and changed into a name. The entire ungrammatical le construct is erased and replaced by le xekri zdani.

Note: The current machine parser does not implement sa erasure. Getting sa right is even more difficult (for a computer) than getting si right, as the behavior of si is defined in terms of words rather than in terms of grammatical constructs (possibly incorrect ones) and words are conceptually simpler entities. On the other hand, sa is generally easier for human beings, because the rules for using it correctly are less finicky.

The cmavo su (of selma'o SU) is yet another metalinguistic operator that erases the entire text. However, if the text involves multiple speakers, then su will only erase the remarks made by the one who said it, unless that speaker has said nothing. Therefore susu is needed to eradicate a whole discussion in conversation.

Note: The current machine parser does not implement either su or susu erasure.

The following cmavo is discussed in this section:

Speakers often need to hesitate to think of what to say next or for some extra-linguistic reason. There are two ways to hesitate in Lojban: to pause between words (that is, to say nothing) or to use the cmavo .y. (of selma'o Y). This resembles in sound the English hesitation noise written “uh” (or “er”), but differs from it in the requirement for pauses before and after. Unlike a long pause, it cannot be mistaken for having nothing more to say: it holds the floor for the speaker. Since vowel length is not significant in Lojban, the y sound can be dragged out for as long as necessary. Furthermore, the sound can be repeated, provided the required pauses are respected.

Since the hesitation sound in English is outside the formal language, English-speakers may question the need for a formal cmavo. Speakers of other languages, however, often hesitate by saying (or, if necessary, repeating) a word (“este” in some dialects of Spanish, roughly meaning “that is”), and Lojban's audio-visual isomorphism requires a written representation of all meaningful spoken behavior. Of course, .y. has no grammatical significance: it can appear anywhere at all in a Lojban sentence except in the middle of a word.

The following cmavo is discussed in this section:

The cmavo fa'o (of selma'o FAhO) is the usually omitted marker for the end of a text; it can be used in computer interaction to indicate the end of input or output, or for explicitly giving up the floor during a discussion. It is outside the regular grammar, and the machine parser takes it as an unconditional signal to stop parsing unless it is quoted with zo or with lo'u ... le'u. In particular, it is not used at the end of subordinate texts quoted with lu…li'u or parenthesized with to ... toi.

The following list gives the cmavo and selma'o that are recognized by the earliest stages of the parser, and specifies exactly which of them interact with which others. All of the cmavo are at least mentioned in this chapter. The cmavo are written in lower case, and the selma'o in UPPER CASE.

The following list shows all the elidable terminators of Lojban. The first column is the terminator, the second column is the selma'o that starts the corresponding construction, and the third column states what kinds of grammatical constructs are terminated. Each terminator is the only cmavo of its selma'o, which naturally has the same name as the cmavo.