Lojban grammar


The grammar of Lojban is based on predicate logic. The majority of the grammar is borrowed from the prior "logical language" Loglan, and some of its features come from Láadan. The characteristic regularity, unambiguity, and versatility of Lojban grammar owes much to modern linguistics and computer programming—resources that were unavailable to the designers of earlier languages. Lojbanist Bob LeChevalier summarized one advantage of Lojban grammar as follows: "Lojban moves beyond the restrictions of European grammar. It overtly incorporates linguistic universals, building in what is needed to support the expressivity of the whole variety of natural languages, including non-European ones."

Formal grammars

Lojban texts can be parsed just as texts in programming languages are by using formal grammars such as PEG, YACC, Backus–Naur form. There are several parsers available.

Phonology

Lojban has 6 vowels and 21 consonants. The phonemes are commensurate with graphemes, which means Lojban has 27 letters each corresponding to a unique phoneme. Lojbanic graphemes can vary in mode; this article employs the Latin alphabet version, which is currently in the most common usage. The phonemes, on the other hand, are defined solely according to the International Phonetic Alphabet.
The tables below show typical realizations of sounds and the Latin alphabets in Lojban. In all cases except the rhotic consonant the first phoneme represents the preferred pronunciation, while the rest are the permitted variants intended to cover dissimilitude in pronunciation by speakers of different linguistic backgrounds.

Basic sounds

Note that simple vowel graphemes never represent diphthongs like in English. Diphthongs are always written with two-vowel graphemes.

Diphthongs

Lojban has 16 diphthongs, vowels that change quality during their emission but always being single syllable nuclei like pure vowels. Unlike English and similarly to languages such as Spanish, diphthongs are not distinct phonemes by themselves but are analyzed as a combination of "semi-vowel + vowel".
The combinations , , and , for instance, are all realized as the corresponding falling diphthongs. Triphthongs exist as combinations of a rising and a falling diphthong, e.g. .

Allophones

The vowels can be either rounded or unrounded. The consonants can be either aspirated or unaspirated, although the voiceless stops, and are usually aspirated to some degree. In general, consonants are not palatalized. The affricates /d͡ʒ/, /t͡ʃ/, /d͡z/, and /t͡s/ occur in Lojban, but they are not distinct phonemes, contrary to English /d͡ʒ/ and /t͡ʃ/. Each is considered to be a combination of an appropriate stop and fricative phoneme: the sequences , , and , respectively. The rhotic sounds are all equally acceptable as an identical phoneme.,,, and may be syllabic.

Buffering of consonant clusters

For those who, given their native language background, may have trouble pronouncing consonant clusters, there is the option of inserting buffer vowels between them, as long as they differ sufficiently from the phonological vowels and are pronounced as short as possible. This mechanism can be described as a form of anaptyxis. Possible choices include ,, and . The resulting added syllables are completely ignored by the grammar, including for the purposes of stress determination.

Orthography

Lojban may be written in different orthography systems as long as it meets the required regularities and unambiguities. Some of the reasons for such elasticity would be as follows:
  1. Lojban is rather defined by the phonemes, therefore, as long as they are correctly rendered so as to maintain the Lojbanic audio-visual isomorphism, a representational system can be said to be an appropriate orthography of the language;
  2. Lojban is meant to be as culturally neutral as possible, so it is never crucial or fundamental to claim that some particular orthography of some particular languages should be the dominant mode.
Lojbanist Kena extends this principle to argue that even an [|original orthography of the language] is to be sought.

Latin/Roman mode

Lojban's Latin alphabet consists of 23 lerfu a b c d e f g i j k l m n o p r s t u v x y z plus 3 semi-lerfu ',. . They are intentionally ordered in accordance with that of ASCII characters.
Capitalization may be applied to mark a non-standard stressed syllable as in cmevla, but they are not considered separate lerfu. Whether a single vowel or the entire syllable is capitalized is a matter of preference; for example, the name "Josephine" can be rendered as either DJOzefin. or djOzefin..
Punctuation marks are not mandatory; such notions as question or exclamation are expressed with words rather than unpronounced symbols.

Cyrillic mode

This mode was conceived when the introductory Lojban brochure was translated into Russian. 23 lerfu а б в г д е ж з и к л м н о п р с т у ф х ш ъ plus 3 semi-lerfu ',. are used. The hard sign ъ is assigned to the open-mid vowel. Diphthongs are written as vowel pairs, as in the Roman mode.

Tengwar mode

Kena argues for writing Lojban using Tengwar, arguing that:
  1. the Latin alphabet is too strongly related to Western civilizations, and thus probably introduces some kind of cultural bias in Lojban. Lojban wants to be both logical and culturally neutral, the Tengwar already are;
  2. the Tengwar system inherently contains some main Lojban morphology rules, making Lojban easier to learn when it is written with Tengwar.
Advocates of this mode include Eric S. Raymond.

Zbalermorna mode

Another option for cultural neutrality is the zbalermorna alphabet.

Japanese mode

A Japanese hiragana version of Lojban orthography has been proposed, in which case more than 80 lerfu may be used. This mode is not without certain technical issues since the hiragana are always syllabic, indicating an open syllable, requiring some special attention when representing the Lojbanic consonant clusters. Several example haiku compositions in the orthography have been created.

Morphology

Lojban has 3 word-classes: brivla, cmavo, and cmevla. Each of them has uniquely identifying morphological properties, so that one can unambiguously recognize which word is of which part of speech in a string of the language. They may be further divided in sub-classes.
There also exist affixes assigned to some of the brivla and cmavo.

brivla – verbs

brivla carry the content of an expression. The bare brivla corresponds to the verb in English. Additional words can be used to transform it into what corresponds to English common nouns, adjectives, or adverbs. Brivla may be identified by the following morphological properties:
  • Have more than one syllable
  • Are penultimately stressed
  • Have a consonant cluster in the first 5 lerfu including the second consonant
  • Start with a consonant
  • End with a vowel
A word such as lobypei will still be considered as a brivla because the special gluing vowel y between b and p is to be ignored and therefore a consonant cluster assumes its existence within it.
Unlike verbs in English, brivla do not inflect for tense, person, or number.
Brivla's sub-classes are as follows, with some examples.

gismu – core verbs

The brivla which constitute the lexical core of the language are called gismu. They are invariably five-letter words, which distinguishes it from the other types of brivla, and are in a form of either CVCCV or CCVCV. Being two syllables means that the general rule of gismu to be stressed penultimately will always cause the first syllable to be stressed.


They have been chosen or added as root words because they
  • represent concepts that are very familiar and basic,
  • represent concepts the usage of which is equally frequent among different languages,
  • would be helpful in constructing more complex words, or
  • represent fundamental grammatical concepts of Lojban.
The main source languages from which they were drawn are Arabic, Chinese, English, Hindi, Russian, and Spanish. Here is further explanation of the nature of gismu by Cowan:
According to Robin Turner, the creation was algorithmically done by computer.
Approximately 1350 gismu exist, which is a relatively small number when compared to that of English words ranging from 450,000 up to 1,000,000. Theoretically, by learning only these root words, as well as their fragmental forms and some major structure words, one will be able to communicate effectively in Lojban.

lujvo – verbs made of affixes

The form of brivla made of affixes is called lujvo.

fu'ivla – borrowed verbs, loan words

fu'ivla usually refers to things that are culture-specific or to kinds of plants or animals, concepts which cannot be easily expressed as mere modifying-modified combinations of Lojban's internal root verbs.
fu'ivla can be subdivided into four types according to the extent to which they are modified, namely Stage 1, 2, 3, and 4 fu'ivla.
Stage 1 fu'ivla
The longest form, quoting a foreign word/phrase while preserving its original spelling with particular structure words.


  • A "hybrid" stage sometimes is enlisted. In this case, it would take the above sentence but Lojbanize "spaghetti" to be phonetic to a Lojban without changing the ending. Therefore, stage "1.5" fu'ivla for "spaghetti" is "me la'o ly. spageti ly."
    Stage 2 fu'ivla
This stage involves lojbanizing the sound and spelling of the word.
Stage 3 fu'ivla
At this stage a borrowed word is fully turned into a single brivla, having its own place structure. Since no brivla may have more than one meaning, it is often the case that they are attached by a rafsi categorizing or limiting the semantic scope of the word. Again they always start with a consonant and end with a vowel.

Stage 4 fu'ivla
These are the borrowings which are so common or so important that have become as short as possible, having no rafsi classifier. Unlike other brivla, they may begin with a vowel. Also the word must not be of a form that one can remove all the initial vowels and have a valid word.

lujvo with components made of fu'ivla

It is possible to absorb a fu'ivla into a lujvo, with principles varying among Lojbanists. Notable proponents are Pierre Abbat and Jorge Llambías. Here are some comparisons of their methods drawn from the Lojban mailing list :

cmavo – particles, structure words

Lojban particles, cmavo, are recognized by following morphological properties:
  • may be a single syllable
  • never contain a consonant cluster of any type, whether or not y is counted
  • end in a vowel
  • need not be penultimately stressed, though they often are if they have more than one syllable
And they display one of the following letter patterns: V, VV, V'V, CV, CVV, CV'V. The form generally does not indicate anything about its grammatical function.
cmavo can be sequenced without spaces and without any change to its meaning:


As far as the stress rules of Lojban are concerned, such compound cmavo are still separate words, so penultimate stress is not obligatory.
Some cmavo have rafsi, which may help converting tanru into lujvo:

cmevla – name words

cmevla are mostly used for names of things in descriptions or in direct address. They can be in any form as long as they end in a consonant. The practice by which names in natural languages are modified to be used in Lojban is known as "lojbanization".

rafsi – affixes (suffixes, infixes, prefixes)

A special building block, from which a new lujvo may be created, is called rafsi. Rafsi are bound to certain gismu, cmavo or fu'ivla. This also means that lujvo have no rafsi form of their own. Rafsi cannot by themselves function as an individual words; they only form lujvo when at least two are joined together.


The unambiguity of Lojban morphology, according to John Woldemar Cowan, gives rise to "significant clues to the meaning and the origin of the word, even if you have never heard the word before". He further says: "The same principle allows you, when speaking or writing, to invent new brivla for new concepts 'on the fly'; yet it offers people that you are trying to communicate with a good chance to figure out your meaning. In this way, Lojban has a flexible vocabulary which can be expanded indefinitely."

Syntax and semantics

According to What Is Lojban?, the language's grammatical structures are "defined by a set of rules that have been tested to be unambiguous using computers", which is called the "machine grammar". Hence the characteristics of the standard syntactic constructs in Lojban:
  • each word has exactly one grammatical interpretation;
  • the words relate grammatically to each other in exactly one way.
Such standards, however, are to be attained with certain carefulness:
The computer-tested, unambiguous rules also include grammar for 'incomplete' sentences e.g. for narrative, quotational, or mathematical phrases.
Lojbanic expressions are modular; smaller constructs of words are assembled into larger phrases so that all incorporating pieces manifest as a possible grammatical unity. This mechanism allows for simplistic yet infinitely powerful phrasings; "a more complex phrase can be placed inside a simple structure, which in turn can be used in another instance of the complex phrase structure".

bridi – clauses, predications

Being derived from predicate logic, the basic unit of Lojban expression is predication, a claim that some objects stand in some relationship, or that some single object has some property. In Lojban they are expressed using clauses. Just as a predication is formed by a predicate and arguments in formal logic, bridi are formed by selbri and sumti in Lojban. In linguistic terms selbri is the predicate of a clause, and sumti are its arguments. A construct of selbri and sumti produces a claim that something stands in a specified relationship to something else or has a specified property.


Multiple bridi can be either sequenced across multiple sentences or compounded in one sentence:


A compound bridi can include multiple tenses and sumti:


The implicit grammatical divisions can be made explicit by separator words such as cu and vau, which are often elidable but sometimes need to be present to avoid ambiguity:


The places of cu and vau in the previous examples can be rendered as follows:


The ordered sets of sumti assigned to every selbri are known as "place structures". They are explicitly defined in dictionaries or word lists.
Some lujvo formations usually operate on the place structure in predictable ways. The rafsi gau, for instance, inserts one place for the agent and pushes all others down one. Thus brivla can have indefinitely many places. This contrasts with the accusative alignment or ergative alignment that most languages have, in which there is a small number of named places and all others are expressed by prepositions.
The typology of Lojban is basically subject–verb–object, with subject–object–verb also common. However, it can practically be anything. All the following sentences mean "I love you", differing only in word order:


Such flexibility has to do with the language's intended capability to translate as many expressions of natural languages as possible, based on a unique positional case system. The meaning of the sentence mi prami do is determined by prami realizing, with its own predefined place structure, a specific semantic relation between mi and do; when the positional relation between mi and do changes, the meaning of the sentence changes too. As shown above, Lojban has particular devices to preserve such semantic structure of words while altering their order. Compare the followings:


se converts the x1 and x2 sumti place. fo tags the x4 place, and fi the x3. Such conversion and tagging is often used to emphasize particular sumti by bringing it forward.
Here are some collations of natural languages and Lojban:


It is important to note that Lojban selbri is not a real equivalent of verb in natural languages. A selbri can be either a verb, a noun, an adjective, or an adverb. Its function is determined syntactically, not morphologically. An analogy to natural language word orders by using such terms as "subject", "verb", and "object" cannot accurately describe the nature of Lojban bridi.

sumti – arguments of predicates

There are five kinds of simple sumti:
  1. descriptions, which usually begin with a descriptor such as le or more commonly now lo;
  2. pro-sumti, the Lojban analogue of pronouns, such as mi;
  3. names, which usually begin with la, such as la.lojban.;
  4. quotations, which begin with lu, lo'u, zo, zoi;
  5. pure numbers, which usually begin with li.

    description

Basic descriptions in Lojban consist of two units, LE/LA descriptors and a selbri:
Although le is quite close in meaning to English "the", it has particularly unique implications. In this example, le creates an argument which might occur in the x1 place of the belonging selbri zarci, namely a "market". le also specifies that the speaker 1) has one or more specific markets in mind is merely describing the things he/she has in mind as markets, without being committed to the truth of that description. Whereas English-speakers must differentiate between "the market" and "the markets", Lojban-speakers are not required to make such a choice :
Since the construct le + selbri merely describes something or other which the speaker chooses to represent based on his/her observation, such an expression as follows is possible:
While le is specific, lo is not:
lo refers generally to one or more markets, without being specific about which. Unlike le zarci, lo zarci must refer to something which actually is a market. lo morsi cu jmive is false as there are no objects in the real world which are both dead and alive.
la dissociates the subsequent selbri from its normal meaning, usually making a name. Like le descriptions, la descriptions are implicitly restricted to those the speaker has in mind:
All descriptions implicitly terminate with ku, which can almost always be omitted with no danger of ambiguity. The main exceptions are a) when relative clauses are involved and b) when a description immediately precedes the selbri. Other usages of ku include making a compound negator and terminating place-structure/tense/modal tags.

selbri – main verbs of clauses, logical predicates

The selbri is the logical predicate of a bridi. This is not to be confused with the meaning of predicate in terms of the English Language, but as a logical predicate. Whereas a predicate in English contains everything that the subject is doing, a logical predicate is simply the relation between all involved parties. In this context, the selbri is roughly the equivalent of a verb in English. For instance:


The gismu nelci is being used as the selbri in this bridi. It is describing the relationship between the sumtis mi and le gerku. The relationship is that of a liker and that which is liked. The roles in the relationship are determined by the sumti placements inherent in the word being used as the selbri. The cmavo se/te/ve/xe are used to swap the first sumti placement of the selbri with the second, third, fourth, and fifth sumti placement, respectively. This functionality allows for the flexibility in bridi. For instance, the gismu klama has the sumti of:
  • x1: One which goes
  • x2: The destination of a goer
  • x3: The source of a goer
  • x4: The route taken by a goer
  • x5: The vehicle used by a goer
Thus:


Selbri can also be tanru, where the sumti placements are determined by the last brivla that is part of the tanru. For instance:

tanru – compound verbs

A group of two or more selbri put one after another is called tanru and acts as a single selbri. They correspond to English compound verbs like drip fry or drip dry.
Tanru can be converted to sumti as usual selbri.
Tanru may be used so as to more specifically conceptualize the intended meaning. In lo skami pilno, there is a tanru skami pilno where the modifying brivla skami narrows the sense of the modified brivla pilno to form a more specific concept. Without skami, lo pilno will just mean "user". Other examples:


Arguments of selbri representing tanru units may be filled as well, in which case they are attached using the particle be :

cmavo

The cmavo consist of various particles and structure words. These include descriptors, connectives, attitudinals, prepositions, and tense words.

Descriptors

There are five descriptors : lo, le, la, li, and me'o; of which the first three inflect to show individual, mass, or set.
IndividualMassSetTypical
Indefiniteloloilo'ilo'e
Definiteleleile'ile'e
Namelalaila'i–
Numberli–––
Mathematical expressionme'o–––

The individual/mass distinction is similar to the distinction between mass nouns and count nouns, but things that are normally counted can be considered as a mass. The set articles consider the mathematical set of the referents.


The number and mathematical expression articles are used when talking about numbers and numerals or letters as themselves.


Note that gadri are sometimes called articles, which doesn't correspond to the usual notion of articles in linguistics.

Connectives

As befits a logical language, there is a large assortment of conjunctions.
There exist 16 possible different truth functions, the four fundamental ones of which are assigned four vowels in Lojban. These vowels are a component sound from which actual logical-connective cmavo are built up.


With the four vowels, the ability to negate either sentence, and the ability to exchange the sentences, as if their order had been reversed, Lojban can create all of the 16 possible truth functions except TTTT and FFFF.
.onai is preferred over na.o. Note that in se.unai, the negation occurs before "swapping" the positions of A and B. While se can be used for the other connectives, it is unnecessary since or, and, and if and only if have the same relationship if their inputs are swapped.


In order to remain unambiguous, each place in the grammar of the language where logical connection is permitted has its appropriate set of connectives. If the connective suitable for sumti were used to connect selbri, ambiguity would result. Connections between components other than sumti can be expressed as follows :


Negatives and se can be used with these connectives with similar forms.


Connections can also be given using "forethought" forms


Negatives are created by adding nai to the connective before the negated construct.


Connections can be questioned:


Besides the logical connectives, there are several non-logical connectives. These do not change form depending on what they are connecting:


The ku is required by the LALR parser, but not by the PEG parser, which however is not official yet.

Attitudinals

Attitudinals are a set of cmavo which allow the speakers to express their emotional state or source of knowledge, or the present stage of discourse. In natural languages, attitudes are expressed using interjections, but also by the tone of voice when speaking, and by punctuation when writing; in Lojban, such information are extensively expressible in words. And the meanings are to be understood separately from the main predicate.


They may be "scaled" by suffixes:


Combination is possible, and highly productive as well as creative:


Evidentials, derived from those of American Indian languages and the constructed language Láadan, show how the speaker came to say the utterance, i.e. the source of the information or the idea:

Prepositions

There are two kinds of prepositions in Lojban: tense markers and proper prepositions. The syntactic difference is that a proper preposition can be converted with se, whereas a tense marker cannot. All proper prepositions are formed from a brivla and mark their object semantically as being in a place of that brivla. Thus the following are equivalent:


Prepositions can also be placed in .i... bo to make sentence conjunctions. With most prepositions this makes no sense, but ki'u, ja'e, mu'i and ni'i are often used this way to express various kinds of "because" and "therefore":

Tense words

Lojban has 63 unique tense words to express various aspects of both space and time as well as event: such a system is unusual among other languages, in that it deals with spatial and temporal aspects in the same term. Simply put, these tense words are used to show where and when the event is taking place, relative in space-time to when the sentence is spoken.
They can be roughly subdivided as follows:
Intervalspu, ca, ba, bu'u, ne'a, re'o, te'e, pa'o, ne'i, fa'a, to'o, ze'o, zo'a, zo'a, ga'u, ni'a, ca'u, ti'a, ri'u, zu'a, be'a, ne'u, vu'a, du'a, ru'u, cu'e
Modifierszi, za, zu, vi, va, vu, ze'i, ze'a, ze'u, ze'e, ve'i, ve'a, ve'u, ve'e, vi'i, vi'a, vi'u, vi'e
Contoursca'o, pu'o, ba'o, co'a, co'i, co'u, mo'u, za'o, de'a, di'a
Convertersmo'i, fe'e, roi, re'u, jai, ki

The "converters" require further explanation: mo'i marks that the movement is towards a point rather than at a point, so that be'a, meaning " in the north", becomes mo'ibe'a, " northward"; roi marks the number of times an event takes place, e.g. paroi, roroi, noroi, re'u indicated an ordinal, e.g. pare'u, first time; fe'e marks an interval in space rather than in time, so that, for example, fe'e co'a is the "beginning point" of something physically extending in a given direction; jai is used to mark tense words that have been moved in a sentence, comparable to se/te/ve/xe and fa/fe/fi/fo/fu for sumti; ki marks a "sticky tense", as explained below.
Marking tenses is always optional in Lojban:


Where the tense information is not specified, the context resolves the interpretation.
Tense words are usually put right before the selbri:
They may be placed elsewhere with the additional terminator ku:


The terminator is used so that the tense word do not directly run into the following sumti and modify it. Compare the next sentences:


Tenses can be "layered up":
Tenses can be "sticky" by being set with ki, continuing in effect over more than a single bridi, until it is unset:
The second ki resets the tense to the implicit default time from the speaker's point of view, which is "now".
Using ki, equivalents of the previous layering tenses can be produced:
The second pu is to be counted from the tense set by the last ki, so in effect it is equivalent to pupu.
Similar to the tense words are a collection of words to show potentiality: ca'a, "actually is"; ka'e, "is innately capable of"; nu'o, "can but has not"; pu'i, "can and has". These words operate in the same way as tense words.

Grouping

Semantic unifications and combinations are performed by word groupings, each include a starting and a terminating cmavo. Here basic example sentences with them are ordered by their terminating cmavo:
  • ti xamgu be le tifri bei lo togmo'a po ko'a be'o elsa cu se nelci ? This for the audience in view of its melody good song is liked.
  • ti elsa fi'o xamgu zukte fe'u lo barda jgita cu se nelci ? This with a big guitar well played song is liked.
  • ti elsa ji'a ne ta gugde ge'u ki'u le finti pu clira zmadu xabju va ? This song is also incidentally associated with that country because the author once lived there.
  • po'o le elsa pe ta gugde ge'u.e na drata cu se nelci va ? Only the songs associated with that country are liked there, others not.
  • ti elsa po ta gugde ge'u se pilno fi gricniselsa'a va ? This song which is owned by that country is used as anthem there.
  • ti elsa ge se finti gi se pilno fi gricniselsa'a vau nenri ti gugde ? This song is in this country both created and used as anthem.
  • ti elsa cu sarcu tolcau ka xamgu togmo'a kei.ijanai do djica lo nu ko'a misno binxo ? One property of this song will have to be a good melody if you want it to become famous.
  • li'i zukte ti elsa kei po mi cu banzu lo terva'i ko'a ka xamgu ? My experience about playing this song is sufficient for estimating how good it is.
  • le nu so'i xamgu zukte kei misno elsa cu se nelci. ? The due to many good playings famous song is liked.
  • le/lo elsa ra'i ta gugde ku se nelci va ? The/A song from that country is liked there.
  • ti elsa poi fange finti pu stidi za'i ke'a gricniselsa'a ti gugde ku'o se pilno vi ta tadji ? This song which a foreign author proposed as anthem for this country is used here that way.
  • ti sagypemci vasru lo'u ti gugde cu banli.i ko banro.e za'i snada le'u ? It is questionable whether this song lyrics should comprise "This country is great. Grow and prosper!"
  • ti sagypemci vasru lu ti gugde cu banli.i ko banro.e za'i snada li'u ? This song lyrics comprise "This country is great. Grow and prosper!"
  • la'e le elsa lu'u cu certu preni ku'i lo elsa se'i cu xlali ? The ones who refer to the song are smart people but the song itself is bad.
  • soi mi pilno ji'a so'o elsa po ko'a se'u nenri lo zdani po mi'a ? I use also some songs of him/her and vice versa, in our house.
Furthermore ku'e, lo'o, te'u and ve'o are terminators for mathematical expressions.