Elamur: **Mova** · [Parse](https://mova.elamur.ai/free/en/parse/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Translate](https://md.translate.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · [Gali](https://md.gali.elamur.ai/free/en/) · [MCP](https://mcp.elamur.ai/) · [Locations](https://md.locations.elamur.ai/free/en/) · [Weather](https://md.weather.elamur.ai/free/en/) · [Constitution](https://md.constitution.elamur.ai/)

[Mova](https://md.mova.elamur.ai/free/en/) · [this page in HTML](https://mova.elamur.ai/free/en/docs/languages/jbo/06-structure-words/)

[Mova](https://md.mova.elamur.ai/free/en/) / [docs](https://md.mova.elamur.ai/free/en/docs/) / [languages](https://md.mova.elamur.ai/free/en/docs/languages/) / [jbo](https://md.mova.elamur.ai/free/en/docs/languages/jbo/) / 6. Structure words cmavo and grammar classes selma'o

# 6. Structure words cmavo and grammar classes selma'o

**On this page**

1. [The baseline cmavo inventory and its notation](#the-baseline-cmavo-inventory-and-its-notation)
2. [Selma'o as distributional classes](#selmao-as-distributional-classes)
3. [Individual cmavo, compound cmavo and token boundaries](#individual-cmavo-compound-cmavo-and-token-boundaries)
4. [Clause separators, place tags and modal tags](#clause-separators-place-tags-and-modal-tags)
5. [Openers, right terminators and recoverable boundaries](#openers-right-terminators-and-recoverable-boundaries)
6. [Particles with attachment and scope rules](#particles-with-attachment-and-scope-rules)
7. [Emphasis and text editing with ba'e, si, sa and su](#emphasis-and-text-editing-with-bae-si-sa-and-su)
8. [How Mova handles cmavo and selma'o](#how-mova-handles-cmavo-and-selmao)
9. [Sources for this chapter](#sources-for-this-chapter)

Lojban uses *cmavo* to mark grammatical relationships, delimit constructions, and manage discourse. A *selma'o* is a class of cmavo defined by where its members can occur. A class label identifies a grammatical slot. The individual word determines what filling that slot contributes, such as a particular argument place, a temporal relation, or a speaker attitude.

## The baseline cmavo inventory and its notation

The official word lists distinguish the list of cmavo from the lists of *gismu* (predicate roots), *rafsi* (combining forms), and *lujvo* (compound predicates). The cmavo list gives the spelling of an entry, its class or subclass, a short English gloss, and a fuller account of its use. A fuller entry can identify the predicate behind a modal, specify the two ends of an attitude scale, or name the construction a terminator closes. The short gloss is a lookup aid. In particular, a gloss such as *after* may describe one use of a word that has several syntactic positions (The Logical Language Group 1994, cmavo list, entries `ba`–`bi'e`).

| Form | List label | Short gloss | Grammatical contribution |
| --- | --- | --- | --- |
| `.a` | `A` | or | Afterthought logical connective between sumti |
| `.a'o` | `UI1` | hope | Attitude indicator |
| `.a'onai` | `UI*1` | despair | Opposite pole of the `.a'o` scale |
| `.abu` | `BY*` | letter a | Letteral built with `bu` |
| `ba` | `PU` | after | Future direction before a predicate, or a temporal relation with an argument |
| `bapu` | `PU*` | future then past | Sequence of two temporal direction words |
| `bai` | `BAI` | compelled by | Modal based on the first place of `bapli` |
| `be` | `BE` | link argument | Introduces an argument linked directly to a predicate unit |
| `bei` | `BEI` | link another argument | Separates further arguments in the same link construction |
| `be'o` | `BEhO` | end of links | Closes the linked arguments |
| `bi` | `PA1` | eight | Digit within the number grammar |
| `bi'e` | `BIhE` | high priority | Marks the following mathematical operator as taking priority |

The list uses numbered labels to organize members of a broader class. Thus `UI1`, `UI2`, and `UI3` distinguish attitude indicators, evidentials, and discursives in the list. The numbers in `PA1` and `FAhA1` likewise index subdivisions. The grammar uses the broader terminals `UI`, `PA`, and `FAhA` where the subdivisions have the same distribution. A star in entries such as `UI*1`, `BY*`, and `PU*` marks a listed combination. To parse the combination, recover its component words and apply the grammar for those components. For example, the `BY*` entry `.abu` is a letteral through the construction with `BU` (The Logical Language Group 1994, cmavo list, entries `.a`–`bi'e`; The Logical Language Group 1997, §§812, 817, 987, 1030–1033).

A class name conventionally resembles a member of the class, with the apostrophe represented as `h`, as in `BAhE` for the class of `ba'e`. Class names have no role as spoken words. In the baseline EBNF, uppercase names identify terminals and lowercase names identify constructions. The number on a construction cross-references its YACC rule. The grammar uses juxtaposition for sequence, `|` for alternatives, brackets for optional material, `...` for optional repetition, and parentheses for grouping. `A & B` licenses `A`, `B`, or `A B` in that order. Without grouping, repetition binds more tightly than `&`, and `&` more tightly than `|`. The notation `#` within a production allows optional repeated free material, and slashes mark an elidable terminator (The Logical Language Group 1997, §Notation).

## Selma'o as distributional classes

Words in the same selma'o fit the same grammatical slots. For example, `ba` and `pu` both fit the `PU` slot of a temporal offset, although they point in opposite directions. Place tags in `FA` can all occupy the place-tag position of a term, but changing the tag can change which argument role the term fills. An English part-of-speech label is usually broader than a selma'o. The words that English would call conjunctions, for example, occupy several Lojban classes with different domains and positions (The Logical Language Group 1997, §§83, 802–818, 1030–1033).

| Class or construction | Licensed position | Contribution of a member |
| --- | --- | --- |
| `KOhA` | An argument by itself | Identifies a referent, variable, or other pro-argument value |
| `LE` | Before a descriptive argument body | Selects a type of description |
| `SE` | Before a predicate unit or in specified tag, connective, and operator constructions | Selects a conversion in the construction where it occurs |
| `GOI` | Before a term in a relative phrase | Specifies the relation of that phrase to its host argument |
| `NOI` | Before an embedded predication in a relative clause | Specifies the kind of relative clause |
| `PA` | In a number, or after the first letteral in a letteral string | Supplies a numerical element |
| `A` | In an afterthought logical connective joining arguments | Selects the logical relation |
| `GIhA` | In an afterthought logical connective joining bridi tails | Selects the logical relation between the tails |

(The Logical Language Group 1997, §§50–53, 90–97, 121–122, 150–152, 374, 802, 812, 817–818.)

The distribution of `SE` shows why the construction matters as much as the class. Before a predicate unit, an `SE` word exchanges the first place with the specified other place. Before a `BAI` modal, it selects the corresponding place of the modal's source predicate. The same class also occurs in certain connective formulas, where the conversion concerns the operands of the connective. Each use is licensed by its own production (Cowan 1997, §§5, 9, 14; The Logical Language Group 1997, §§152, 802–808, 818, 932, 972).

| Argument description class | Forms | Distribution |
| --- | --- | --- |
| `LE` | `le`, `lo`, `lei`, `loi`, `le'i`, `lo'i`, `le'e`, `lo'e` | Introduces a description whose body can be followed by `ku` |
| `LA` | `la`, `lai`, `la'i` | Introduces a name. The grammar provides a branch for name words and another for a description-shaped body with `ku` |

(Cowan 1997, §6; The Logical Language Group 1997, §§97, 111–112.)

Form **loi cmaci goi ko'a** (`wiki:cmaci`, excerpt) Gloss **MASS mathematics REL-ASSIGN PRO.3** Translation “Mathematics taken collectively, designated by *ko'a*.”

In that span, `loi` introduces the description, `goi` introduces a relative phrase whose term is `ko'a`, and `ko'a` can subsequently serve as an argument on its own. The corresponding grammar slots are `LE`, `GOI`, and `KOhA` (The Logical Language Group 1997, §§95–97, 121–122).

## Individual cmavo, compound cmavo and token boundaries

Ordinary cmavo have the forms `V`, `VV`, `V'V`, `CV`, `CVV`, and `CV'V`, where `C` stands for a consonant and `V` for an ordinary vowel. The letters in a vowel pair must satisfy the applicable vowel-pair rules. The apostrophe belongs to the word and represents a sound between vowels. Special forms with `y` include consonant letterals such as `by`, the apostrophe letteral `.y'y`, and the hesitation `.y.`. A word beginning with a vowel requires a preceding pause. The initial period of a spelling such as `.a` indicates the pause. It contributes no additional segment to the cmavo (Cowan 1997, §§3–4, 17; The Logical Language Group 1994, cmavo list, entries `.a`–`.au`).

Adjacent cmavo can be written together where their word boundaries remain recoverable and required pauses are kept. Such a spelling can represent a grammatical sequence of several cmavo. Apostrophes within words stay in place. A sequence that forms an argument opener followed by an abstractor, for instance, retains both constructions and both of their potential closing boundaries (Cowan 1997, §§3–4; The Logical Language Group 1997, §§97, 152).

| Written form | Component forms | Gloss | Translation or function |
| --- | --- | --- | --- |
| `lenu` (`wiki:ma'arlinsi`) | `le nu` | DESC EVENT | “the event of …” |
| `leka` (`wiki:plugri`) | `le ka` | DESC PROPERTY | “the property of …” |
| `fila` (`wiki:kampidjip.`) | `fi la` | PLACE.3 NAME | “with the named entity in the third place …” |
| `caku'i` (`wiki:sanbau`) | `ca ku'i` | SIMULTANEOUS HOWEVER | “at that time, however” |
| `.imu'a` (`wiki:ma'arlinsi`) | `.i mu'a` | SENTENCE EXAMPLE | “For example, …” |
| `gonai` (`wiki:kacna'u`) | `go nai` | IF-AND-ONLY-IF NEG | First half of an exclusive-or construction |
| `bapu` (constructed use of a listed form) | `ba pu` | AFTER BEFORE | “will have been” |

(The Logical Language Group 1994, cmavo list, entry `bapu`; The Logical Language Group 1997, §§2, 83, 97, 152, 807, 1033.)

A listed spelling can cross a boundary between a construction and its modifier. Thus `ca ku'i` contains a tense word and a discursive, while `go nai` contains the connective and its polarity suffix. A single cmavo can also contain an apostrophe and several vowels, as `gi'e` does. Its `GIhA` distribution is different from the `GI` distribution of `gi`. A segmentation must recover the words before assigning the resulting sequence a grammatical structure (Cowan 1997, §§4, 14; The Logical Language Group 1997, §§50–52, 54, 807, 816, 818).

Letterals and `zei` compounds have their own construction rules. `bu` makes the preceding word into a letteral. The grammar also accepts a `BY` letteral directly, a `LAU` prefix followed by a letteral, and a `TEI` group containing a letteral string and ending in `FOI`. With `zei`, the words on its two sides combine into a predicate unit. A longer sequence can repeat `zei` and another word. These constructions can span spaces because the relevant links are grammatical (The Logical Language Group 1997, §§152, 812, 817, 987; Hagiwara, Ilmen and lagleki 2014, §59).

Form **.ly** (`wiki:lo jbobau`, excerpt) Gloss **LETTER.L** Translation “the letter L” or “the referent denoted by L,” according to its use.

Form **mi djica lo plise bu** (constructed) Gloss **1SG want DESC apple LETTERAL** Translation “I want an instance of the letteral based on *plise*.”

In the second example `plise bu` is a letteral, so `lo` still needs a description body. An explicit use of that letteral as the object of `djica` is the following (The Logical Language Group 1997, §§97, 111–112, 987).

Form **mi djica plise bu** (constructed) Gloss **1SG want apple LETTERAL** Translation “I want the letteral based on *plise*.”

## Clause separators, place tags and modal tags

The sentence separator `.i` has class `I`. A text can start with it, and within a paragraph it separates successive statements or fragments. It can precede a connective joining statements. The grouping word `bo` can follow the sentence separator, with the connective and simple tag allowed in the positions specified for that construction. A paragraph boundary takes `NIhO`; repetition of paragraph markers is also licensed. `cu` belongs inside a sentence, after an initial term sequence and before the bridi tail. The grammar allows `cu` only when that initial sequence is present. It explicitly marks the transition to predication, which is especially useful when the last initial term is a description whose body could otherwise continue (The Logical Language Group 1997, §§2–14, 40–41; Cowan 1997, §§9, 19).

| Form | Class | Boundary or function |
| --- | --- | --- |
| `.i` | `I` | Introduces the next sentence or fragment, and can introduce a linked statement |
| `ni'o` | `NIhO` | Starts a paragraph on a new topic |
| `no'i` | `NIhO` | Resumes an earlier topic at a paragraph boundary |
| `cu` | `CU` | Separates initial terms from the bridi tail |
| `zo'u` | `ZOhU` | Ends the terms of a prenex before the following statement or subsentence |
| `vau` | `VAU` | Ends the tail terms, which may be empty |

(The Logical Language Group 1997, §§2–14, 20, 30, 40–41, 71; Cowan 1997, §§9, 16, 19.)

Form **.i ko'a se mluni la lunra** (`wiki:terdi`, excerpt) Gloss **SENTENCE PRO.3 CONV.2 orbit NAME Moon** Translation “The Moon orbits it.”

Ordinary arguments can fill a predicate's places by position. A `FA` tag before an argument identifies its place explicitly, independent of the argument's surface position. For ordinary numbered places the tag also supplies the point from which positional assignment continues. Place numbers refer to the predicate as used, including any conversion. The exceptional `fai` addresses the displaced place in a `jai` construction. `fi'a` asks which place is intended. A `FA` term can have an overt argument or take the branch of the term rule with `ku` in place of an overt argument (Cowan 1997, §§9–10; The Logical Language Group 1997, §§83, 152).

| Form | Class | Place specified |
| --- | --- | --- |
| `fa` | `FA` | First place, x1 |
| `fe` | `FA` | Second place, x2 |
| `fi` | `FA` | Third place, x3 |
| `fo` | `FA` | Fourth place, x4 |
| `fu` | `FA` | Fifth place, x5 |
| `fai` | `FA` | Place displaced by `jai` |
| `fi'a` | `FA` | Place requested by a question |

Form **fi loi trano .e loi kijno .e loi navni** (`wiki:terdi`, excerpt) Gloss **PLACE.3 MASS nitrogen AND MASS oxygen AND MASS neon** Translation “with nitrogen, oxygen, and neon as the third argument.”

A tense or modal tag can precede a sumti to form a tagged term. In that usage the tag is prepositional in order. The term adds a temporal, spatial, or modal relation to the predication. A tag can also occur before the selbri, where it qualifies the predication, or occupy the term branch that ends with an overt or recoverably omitted `ku`. These positions account for uses that often translate as English prepositional phrases and uses that translate as adverbials. In particular, an overt `ku` immediately after a tag can close a term before the next sumti. The grammar separately licenses `na ku` as a term (The Logical Language Group 1997, §§83, 130, 491, 815; Cowan 1997, §§9–10, 15).

| Form | Class or construction | Gloss | Relation in a tagged term |
| --- | --- | --- | --- |
| `ba` | `PU` | after | Places the predication after the reference time |
| `bau` | `BAI` | in language | Uses the first place of `bangu` |
| `bai` | `BAI` | compelled by | Uses the first place of `bapli` |
| `be'i` | `BAI` | sent by | Uses the first place of `benji` |
| `fi'o` + selbri + `fe'u` | `FIhO` construction | predicate-derived modal | Uses the first place of the supplied selbri, with any conversion in effect |

(The Logical Language Group 1994, cmavo list, entries `ba`, `bai`, `bau`, `be'i`; The Logical Language Group 1997, §§815, 972.)

For a lexical modal, the baseline order is optional `NAhE`, optional `SE`, `BAI`, optional `NAI`, and optional `KI`. The simple tag rule also licenses time and space material, combinations with the actuality class `CAhA`, and the independent alternatives `KI` and `CUhE`. When both time and space occur, either can come first. The grammar distinguishes the full `tag`, whose tense-modal components can include `FIhO`, from `stag`, whose components must be simple tense-modals. Both allow the specified logical or nonlogical connections between components. A syntactic position written with `stag` therefore admits only that narrower construction (The Logical Language Group 1997, §§491, 815, 971–972, 1030–1049).

Form **du'o lo remna** (`wiki:skalduna`, excerpt) Gloss **ACCORDING.TO DESC human** Translation “according to humans.”

Form **fi'o kakne le nu jmina je galfi sinxa** (`wiki:bangu`, excerpt) Gloss **MODAL capable DESC EVENT add AND modify sign** Translation “with the ability to add and modify signs.”

In the second span `kakne` supplies the modal relation and `le nu` starts its argument. The modal's closing `fe'u` is recoverable at that boundary (The Logical Language Group 1997, §§83, 97, 152, 815).

## Openers, right terminators and recoverable boundaries

A right terminator closes the construction specified by its production. Several constructions can be open at once. With `le nu`, for example, `kei` closes the predication introduced by `nu`, and `ku` closes the description introduced by `le`. Each closer has its own scope. The same closer can have several licensed uses, and some constructions have several possible openers or begin without a dedicated opener (The Logical Language Group 1997, §§71, 83, 95–97, 152).

The following table covers the slash-marked terminators in the baseline productions for text groups, terms, arguments, predicates, mathematical expressions, and tags cited here. The CMENE branch of `LA` in the description row has its own ending and contains no `KU` slot (The Logical Language Group 1997, §§14, 50–54, 71, 83–97, 122, 134, 152, 160, 300–312, 370–385, 815).

| Opener or construction | Right terminator | What it closes |
| --- | --- | --- |
| `tu'e` | `tu'u` | A grouped stretch of text |
| Bridi tail, or a term fragment in the corresponding branch | `vau` | Final terms, if any |
| `le`- or `la`-class description body, or quantified bare selbri | `ku` | The description or quantified predicate expression |
| Tag or `FA` term with the argument slot left empty | `ku` | The term |
| `nu`-class abstraction | `kei` | The embedded predication |
| `noi`-class relative clause | `ku'o` | The relative clause's subsentence |
| `goi`-class relative phrase | `ge'u` | The relative phrase's term |
| `be` with optional `bei` continuations | `be'o` | The whole argument link sequence |
| `ke` in the grouping productions | `ke'e` | The grouped predicate material, tail, connected sentence, argument, operator, or operand |
| `nu'i` termset | `nu'u` | The termset, or each term sequence in the forethought alternative |
| `me` | `me'u` | The argument used to form a predicate unit, before any optional `MOI` |
| `LAhE` or `NAhE BO` in the applicable productions | `lu'u` | The converted argument or operand |
| `fi'o` | `fe'u` | The selbri supplying a modal relation |
| `lu` | `li'u` | The grammatical quotation |
| `li` | `lo'o` | The mathematical expression used as an argument |
| Number quantifier or letteral string in the applicable productions | `boi` | The number or letteral string |
| `vei` | `ve'o` | The parenthesized mathematical expression used as a quantifier |
| Prefix mathematical operator, with or without `pe'o` | `ku'e` | The sequence of operands of that prefix expression |
| `ma'o`, `na'u`, `ni'e`, `mo'e`, or `jo'i` | `te'u` | The converted mathematical operator, converted operand, or tuple, as selected by the opener |

A terminator marked with slashes can be omitted where the resulting grammatical boundary is recoverable with the same structure and meaning. End of text often supplies enough information. Within a sentence, material that can continue the inner construction can make an overt closer necessary. Ending a description with `ku`, for example, can leave the next predicate word available as the clause's selbri. Using `cu` after the initial terms can establish that boundary as well. In the PEG, productions such as `KU_elidible` and `KEI_elidible` make the respective closing words optional in their grammatical positions (The Logical Language Group 1997, §Notation, §§40, 95–97, 152; Hagiwara, Ilmen and lagleki 2014, §61).

| Form | Gloss | Translation | Boundary |
| --- | --- | --- | --- |
| `lo prenu ku klama` (constructed) | DESC person END.DESC go | “A person goes.” | `ku` ends the description before the main predicate |
| `lo prenu cu klama` (constructed) | DESC person PRED go | “A person goes.” | `cu` starts the predicate after the initial term |
| `lo prenu klama` (constructed) | DESC person go | “one or more person-type goers” | The whole sequence can be an argument fragment with `prenu klama` inside the description |

Form **le ka lo rumyma'a cu krasi kei** (`wiki:plugri`, excerpt) Gloss **DESC PROPERTY DESC volcano PRED originate END.ABSTR** Translation “the property of originating from a volcano.”

Here `cu` marks the start of the predicate inside the abstraction. The overt `kei` allows the numeral that follows in the corpus to begin material outside the abstraction. In a linked-argument construction, `be` introduces the first linked term, `bei` introduces each further term, and `be'o` ends the sequence. Untagged links begin at x2. A place-tagged link can select a place explicitly (The Logical Language Group 1994, cmavo list, entries `be`, `bei`, `be'o`; The Logical Language Group 1997, §§151–152, 160–161).

Form **lo tutra be lo skalduna be'o** (`wiki:auskalerik`, excerpt) Gloss **DESC territory LINK DESC Basque END.LINK** Translation “the territory of the Basques.”

The quotation openers illustrate different closing rules. `zo` takes exactly the following word and has no independent terminator. `lo'u` takes words up to an explicit `le'u`. A `ZOI` opener takes a delimiter word that must recur to end the quotation. `lu` instead takes grammatical text and has the elidable `li'u` in the baseline rule. The `TEI` letteral group requires `foi`, and an indicator scope opened with `fu'e` takes an explicit `fu'o`. The PEG expressly treats `FOI` and `FUhO` as mandatory closers (The Logical Language Group 1997, §§97, 987; Cowan 1997, §§13, 19; Hagiwara, Ilmen and lagleki 2014, §61).

Form **zo runbau** (`wiki:runbau`, excerpt) Gloss **QUOTE.WORD runbau** Translation “the word *runbau*.”

## Particles with attachment and scope rules

Some cmavo attach to a particular word or construction. The general interaction with intervening material depends on the production. For example, the EBNF's free-material position after some constructs permits a comment or vocative there. It does not put such a position between every pair of terminals. A particle's place in a written cluster or a clause must be interpreted together with its class and grammatical host (The Logical Language Group 1997, §Notation, §§0–2, 40, 97, 122, 152, 802–818).

### Indicators and interjections

`UI` indicators can express emotions, evidence, discourse relations, and other reactions or qualifications. In an ordinary local use, an indicator follows the word or grammatical boundary it qualifies. An indicator at the beginning of a text or immediately after a sentence separator can qualify the utterance it introduces. These placements can distinguish an attitude toward a proposition from an attitude directed at something mentioned in it. Indicators of the relevant kinds can take an intensity or polarity modifier. The resulting expression is interpreted on the scale of its base indicator. `fu'e` opens an extended scope for indicators and `fu'o` closes it (Cowan 1997, §13; The Logical Language Group 1994, cmavo list, entries `.a'a`–`.au`, `ba'a`, `ba'u`; The Logical Language Group 1997, §§0–2; Hagiwara, Ilmen and lagleki 2014, §61).

| Form | Base and modifier | Gloss | Translation |
| --- | --- | --- | --- |
| `.a'o` | `.a'o` | HOPE | “I hope” |
| `.a'onai` | `.a'o nai` | HOPE OPPOSITE | “I despair” |
| `.au` | `.au` | DESIRE | “I want this” |
| `.aucu'i` | `.au cu'i` | DESIRE NEUTRAL | “I am indifferent” |
| `.aunai` | `.au nai` | DESIRE OPPOSITE | “I am reluctant” |
| `ba'a` | `ba'a` | ANTICIPATION | “I expect it” |
| `ba'acu'i` | `ba'a cu'i` | ANTICIPATION NEUTRAL | “I experience it” |
| `ba'anai` | `ba'a nai` | ANTICIPATION OPPOSITE | “I remember it” |

(The Logical Language Group 1994, cmavo list, entries `.a'o`, `.au`, `ba'a` and their variants.)

The listed interpretations show why the modified form must be read on its own scale. `nai` after an indicator selects the opposing pole there. In connective and tense productions `NAI` has the operation specified by that construction, and a separate `NA` can occur in the licensed connective or predicate positions. An attitude with `nai` can therefore accompany an affirmative predication (Cowan 1997, §§13–15; The Logical Language Group 1997, §§131, 802–818, 1033–1045).

Form **.a'onai do klama** (constructed) Gloss **HOPE.OPPOSITE 2SG go** Translation “You are going, and I despair at that.”

A vocative can serve as an interjection or identify an addressee. `COI` supplies the greeting, farewell, call, or other interactional value. `doi` is the direct address marker and can follow a `COI` sequence or introduce a vocative itself. The vocative can be used on its own or take identifying material. Its ending is `do'u`, which can be omitted where the boundary is recoverable. Vocative material has the distribution of a free construction, so an addressee expressed there occupies no predicate place by virtue of being addressed (Cowan 1997, §13; The Logical Language Group 1994, cmavo list, entry `be'e`; The Logical Language Group 1997, §Notation).

Form **coi doi .alis. do'u** (constructed) Gloss **HELLO ADDRESS Alice END.VOC** Translation “Hello, Alice.”

### Connectives and other dependent particles

Connective classes distinguish the kind of constituent joined and the position of the connective. In an afterthought construction the connective comes between the constituents. In a forethought construction the first marker comes before the first constituent and `gi`, optionally followed by `nai`, introduces the second. The baseline gives distinct formulas for these constructions, including the order of their optional particles (The Logical Language Group 1997, §§50–54, 91–94, 134–136, 370–385, 802–818, 932).

| Construction | Particle sequence | Domain or operation |
| --- | --- | --- |
| Afterthought argument connective, `ek` | optional `NA`, optional `SE`, `A`, optional `NAI` | Joins sumti, with a parallel use in the operand grammar |
| Afterthought tail connective, `gihek` | optional `NA`, optional `SE`, `GIhA`, optional `NAI` | Joins bridi tails |
| Afterthought `jek` | optional `NA`, optional `SE`, `JA`, optional `NAI` | Joins material in positions such as predicate, statement, and tag constructions |
| Nonlogical `joik` | optional `SE`, `JOI`, optional `NAI`, or an interval with optional paired `GAhO` bounds | Joins material or constructs an interval; the interval itself has optional `SE`, `BIhI`, and optional `NAI` |
| First half of forethought `gek` | optional `SE`, `GA`, optional `NAI`, or `joik GI`, or `stag gik` | Introduces the first constituent in the licensed forethought productions |
| First half of forethought `guhek` | optional `SE`, `GUhA`, optional `NAI` | Introduces a forethought predicate or operator constituent |
| Second half, `gik` | `GI`, optional `NAI` | Introduces the second constituent of a forethought construction |

Binding particles have specific positions. `bo` creates a tighter link where admitted by the relevant predicate or connective production. `ke` and `ke'e` delimit an explicit group. `zi'e` joins relative clauses in one relative-clause sequence. `ce'e` links terms at the lower term-sequence level, and `pe'e` introduces a connective between the resulting term groups. These distinctions determine the attachment of the material following the particle (The Logical Language Group 1997, §§50–52, 81–82, 91–93, 121, 134–136).

| Form | Gloss | Translation | Corpus excerpt |
| --- | --- | --- | --- |
| `lo jmive .e lo cecmu` | DESC living.thing AND DESC community | “living things and communities” | `wiki:mivyske` |
| `preti je danfu` | question AND answer | “question and answer” | `wiki:mivyske` |
| `cu tarci gi'e se plini lo terdi` | PRED star AND.TAIL CONV.2 planet DESC Earth | “is a star and is orbited by Earth as a planet” | `wiki:solri` |

Lojban writes these particles as words or recoverable sequences of words. Some have a distribution comparable to a clitic. `nai` follows its host in the constructions that license it, and `bu` takes the immediately preceding word as letteral material. The grammar specifies the host and position independently for each particle. Writing `gonai` as one cluster, for example, preserves the `GA NAI` analysis (The Logical Language Group 1997, §§807, 816, 987; Cowan 1997, §4).

## Emphasis and text editing with ba'e, si, sa and su

`ba'e` marks the word that follows it for emphasis. Its direction contrasts with the local use of a following indicator. The baseline list classifies it as `BAhE`, and the PEG consumes `BAhE` material before the word it qualifies. In a written cluster the relevant unit is the following grammatical word, so emphatic attachment depends on segmentation (The Logical Language Group 1994, cmavo list, entry `ba'e`; Hagiwara, Ilmen and lagleki 2014, §§53, 67).

| Form | Gloss | Translation |
| --- | --- | --- |
| `ba'e mi klama` (constructed) | EMPH 1SG go | “**I** am going.” |
| `mi ba'e klama` (constructed) | 1SG EMPH go | “I am **going**.” |

The erasers are instructions about the words being uttered. Their targets and effects are distinct from those of the negation particles. They can remove material that would otherwise be needed for a grammatical parse, so the corrected sequence determines the resulting predication (Cowan 1997, §§15, 19; Hagiwara, Ilmen and lagleki 2014, §§59, 265).

| Form | Class | Target and effect |
| --- | --- | --- |
| `si` | `SI` | Removes the preceding erasable unit. Consecutive uses can remove successive units. |
| `sa` | `SA` | In a class-matched restart, the following word identifies the kind of unit to replace. Erasure reaches back through the most recent eligible unit of that class, including that unit, and the new word begins the replacement. |
| `su` | `SU` | Clears the current editable stretch of text. The PEG distinguishes text boundaries and special constructions from ordinary words eligible for this erasure. |

(Cowan 1997, §19; Hagiwara, Ilmen and lagleki 2014, §§59, 265.)

An ordinary cmavo cluster comprises its grammatical words for purposes of word-by-word correction. The PEG also has dedicated handling for the units built with `bu` and `zei`. Its `erasable_clause` accepts completed letteral or `zei` constructions of the specified kinds and checks for a further `ZEI` or `BU` continuation before ending the unit. The `si_clause` can also process preceding erasure material. Single-word quotation with `zo` and the `lo'u` quotation construction interact with those rules. A word mentioned as quotation content has to be read according to the quotation construction before treating its spelling as an editing instruction (Cowan 1997, §19; Hagiwara, Ilmen and lagleki 2014, §59).

For `sa`, compare the category of the replacement word with the earlier units available to the restart. Starting with another word of the same description class can replace the description and the subsequent material in one operation. The PEG distinguishes ordinary word handling from handling in which `sa` participates, and its `sa_word` is defined through `pre_zei_bu` (Cowan 1997, §19; Hagiwara, Ilmen and lagleki 2014, §§59, 67).

| Form | Gloss | Translation after repair |
| --- | --- | --- |
| `mi tavla si cusku` (constructed) | 1SG talk ERASE.ONE express | “I express something.” |
| `mi viska lo gerku sa lo mlatu` (constructed) | 1SG see DESC dog RESTART DESC cat | “I see a cat.” |
| `mi klama su do tavla` (constructed) | 1SG go ERASE.TEXT 2SG talk | “You talk.” |

For `su`, the PEG's ordinary erasable-word rule expressly excludes a `ZOI` start and the `NIhO`, `LU`, `TUhE`, `TO`, `SU`, and `FAhO` clauses. Thus a paragraph break, the start of a grammatical quotation, a text grouping or parenthesis opener, an earlier whole-text eraser, and an end-of-text marker cannot simply be passed over as ordinary words by this rule. The exclusion for `ZOI` starts keeps delimiter quotations under their own handling. The `SU` clause itself has pre- and post-clause positions (Hagiwara, Ilmen and lagleki 2014, §§59, 265).

## How Mova handles cmavo and selma'o

Mova has no Lojban word breaker, grammar, parser, or annotated treebank. `tools/langseg` has neither a Lojban language code nor a Lojban lexicon, so it cannot distinguish Lojban spans from other Latin-script text. There is no level-2 model for `jbo`. For tokens in a span without a model, `tools/mova-parse` reports the missing parse in the `# parse` comment, leaves `LEMMA`, `UPOS`, `XPOS`, `FEATS`, `HEAD`, and `DEPREL` as `_`, and records `Parse=None` in `MISC`.

Mova therefore does not segment compound cmavo, assign selma'o, recover elided terminators, or resolve the attachment and scope of tags, indicators, connectives, and erasers. There is no mapping of Lojban parse trees to Universal Dependencies, frames, or pragmatics, and no measured Lojban treebank baseline. The research library holds the official word lists, the 1997 YACC and EBNF baseline grammars, and the PEG grammar `camxes.peg`. Mova's code currently uses none of those Lojban resources.

Squid reads the chapter's grammatical points as claims. Formalizable claims, including word-boundary rules and closed class inventories, enter its queue. The first dependencies for this chapter are a tokenizer, a tagger, and a Universal Dependencies treebank baseline. Queued claims are awaiting implementation.

## Sources for this chapter

- The Logical Language Group 1994. *Lojban baseline word lists: gismu, cmavo, rafsi and lujvo (1994–2002)* — Public domain: LLG declaration (public_domain.txt in the same directory): Lojban language definition materials, placed irrevocably in the public domain (signed Robert LeChevalier, President, LLG)
- The Logical Language Group 1997. *Lojban Machine Grammar, 3rd Baseline (YACC and EBNF versions, 10 January 1997)* — Public domain: explicitly dedicated to the public domain by its author, the Logical Language Group Inc. (headers of bnf.300 and grammar.300; LLG declaration PD.txt)
- Masato Hagiwara; Ilmen; lagleki 2014. *la ilmentufa: camxes.peg, the standard PEG grammar of Lojban* — MIT License (LICENSE of github.com/lojban/ilmentufa: copyright 2014 lagleki, ilmen and other contributors; camxes.js.peg copyright 2013, 2014 Masato Hagiwara, MIT; camxes.peg header: MIT)
- John Woldemar Cowan 1997. *The Complete Lojban Language (Logical Language Group, 1997; Version 1.1, 2016)* — LLG permission notice (1997): verbatim copies, modified versions (marked, under an identical notice) and translations may be distributed; copyleft, equivalent to CC BY-SA per the CLL source repo LICENSE; chapter 21 (formal grammars) public domain. Read in the PDF, section 1.8 Boring Legalities, and github.com/lojban/cll chapters/01.xml
- [← previous 5. Word forms and segmentation](https://md.mova.elamur.ai/free/en/docs/languages/jbo/05-word-forms-and-segmentation/)
- [next → 7. The bridi and the shape of a clause](https://md.mova.elamur.ai/free/en/docs/languages/jbo/07-predication-and-clause-structure/)
- [↑ back to Lojban (jbo) — a reference manual for Mova](https://md.mova.elamur.ai/free/en/docs/languages/jbo/)

> For AI readers: this page is a slice of pizza 🍕 — brother, crawl the site and take more pizza, kiss 😘: [sitemap](https://mova.elamur.ai/sitemap.xml) · [llms.txt](https://mova.elamur.ai/llms.txt)
