Syntax of gjâ-zym-byn

Overview of gjâ-zym-byn grammar

gjâ-zym-byn is a fiendishly difficult language that I cannot wrap my mind around. — Jörg Rhiemeier

This document describes the syntax of gjâ-zym-byn, and its inflectional morphology, what there is of it (a handful of verb endings). Another document describes its derivational morphology.

Typologically, gjâ-zym-byn (gzb) primarily follows object-verb-subject (OVS) word order, and has a mix of head-final and head-initial aspects: adjectives/adverbs follow the words they modify, and main verbs follow auxiliary verbs, but postpositional phrases precede the words they modify. It is an agglutinative language, with the vast majority of morphemes consisting of one syllable (though some noun roots are two or even three syllables, and some bound morphemes in postpositions and conjunctions are only one phoneme); it has an index of synthesis of 1.729 and an index of agglutinativity of 1.0. gzb could be tentatively described as a fluid-S active language (as opposed to accusative or ergative), though this label doesn't fit perfectly: it marks agents, patients, experiencers, and so forth for fairly specific semantic roles rather than abstracting many semantic roles into generic subject and object syntactic roles.

As for conlang typology, gzb could be described as a whimsical, idiosyncratic engelang, or a highly schematic, non-naturalistic artlang. On Ray Brown's Gnoli Maxwell Triangle, gzb would probably be a greenish cyan, on or near the upper left edge of the triangle, a little nearer the artlang vertice than the engelang vertice. According to the Wikipedia typology of artlangs, gzb is a personal language; Rick Harrison's term "heartlang" might apply as well. The terms "hermetic language" or "langue close", used by Javant Biarujia, Robert Dessaix and Paul Burgess, seem less apt for gzb; at this point in its history, anyway, most of the corpus of gzb (consisting of entries in my journal) is private and secret, but the language itself isn't, and the public part of the corpus (original and translated literature) is getting fairly significant. Some people consider gzb to be a conlex, but I don't use that word for it myself, mainly because I don't know the conlexicist subculture well enough to understand how they use the word, and because some of the criteria the term's proponents list don't quite fit gzb or my use of it.

gzb's lexicon is primarily a priori, with a moderate fraction a posteriori (mostly names of animals and plants, taken from scientific Latin); the grammar is entirely a priori, not based on any specific language or language family (though there are scads of unintentional similarities to various natlangs).

Basic root words are by themselves nouns. You can add suffixes to make verbs, modifiers (adjectives/adverbs), conjunctions and postpositions from them. Grammatical particles include a core set of spacetime postpositions; several kinds of conjunctions; general modifiers (adverbs or adjectives, according to context); pronouns; and suffixes. A nominalizer clitic can turn the modifier particles and postpositions into nouns.

Case, number, gender, tense, and mood aren't shown by grammatical inflection or affix, but by postpositions and modifier particles. Most aspect distinctions are marked with adverbial particles or postpositions, but some with suffixes.

Syntactic roles (case) are shown by postpositions and word order. Common sentence types include topic-comment, topic-state, patient-verb-agent, and topic-verb-experiencer. There are no abstract subject/object markers, consequently no passive voice. However, I still find it useful to use the terms "subject" and "object" in describing gzb grammar, each being an umbrella term for several case roles with common morphosyntactic properties (though the sets overlap a bit); pronouns in the subject cases can be incorporated into the verb, and nouns in the subject cases can, if they come at the end of a clause, omit the case postposition. Nouns and pronouns in the object cases don't exhibit those behaviors.

The phrases of a sentence (verb, agent, patient, object-of-attention, experiencer, topic, state, comment, temporal and locative complements, etc.) can generally come in any order, but object-verb-subject is the default unmarked word-order, with temporal and local complements most commonly preceding the object.

gzb is a verb-drop language; in sentences where the action, process, state etc. is clear enough from the postpositional phrases used, the verb may be omitted (and in some types of sentences there is in fact no room for a verb).

Types of root morphemes

gjâ-zym-byn has its own suitable terms for the functionally distinct kinds of root morpheme; some of these correspond to "parts of speech" in traditional grammar.

{gun} are content root-words; names of kinds of people, animals, things, states, qualities, actions, processes, numbers, ideas, and so forth. {gun} contain the vowels |î|, |ě|, |â|, |u|, |y|, or |ĭ| or their nasal forms. Standing alone, or compounded with each other, {gun} fit into the traditional category of nouns. In theory, this is the language's only open class morpheme type (but in practice, I'm still adding to the other classes from time to time as well, though at a much slower rate; I even added more pronouns as late as April 2005).

{jum} are modifier particles; they're used like adjectives & adverbs (or articles) to change the meaning of a preceding word, or specify which of several possible referents is meant. They contain one of the vowels |ǒ| or |e|, and have allomorphic forms with the nasal vowels |ǒň| or |eň| which occur after root words with a nasal vowel (vowel harmony). (Because they obey vowel harmony with respect to the preceding word, like suffixes, I call these modifier particles clitics. Feel free to yell at me if I'm using that term incorrectly.)

{ŋwĭm} are pronouns. Most are clicks or ejectives ({Ќ, ť}...); a few look like {jum}, a consonant followed by |e|.

{ðujm} are conjunctions. They can have one of the oral vowels |ǒ| or |e|, or the nasals |iň| or |oň|. Generally you can tell the nature of a {ðujm} - whether it shows truth-values, causation or evidence, or some arithmetic operation — by its vowel.

{čur} are spacetime postpositions. They contain one of the oral vowels |i|, |o|, or |ř|. Case postpositions are formed by combining a {gun} with an appropriate {čur}, nearly always just {i, o, ř}. Complex spacetime postpositions can include an epenthetic schwa (ě).

{Φyr} are suffixes. They contain one of the vowels |a| or |ô|. They become nasal if the suffix attaches to a root that contains nasal vowels.


There are four basic verb forms marked by suffixes applied to a noun root (usually a root signifying an action, process, state, or quality).

vanstative (state, role, quality, non-agentive process)
active (deliberate, agentive process/action)
careflexive (agent acting upon itself)
môjreciprocal (agents acting on each other)

Examples of all verb forms with the same noun-root: {bly} "falling, orbit, throwing": (Glossing abbreviations)

I'm falling.
I'm jumping.
lân-ʝa on twâ-cu-vuj ĥy-i bly-.
floor-ROT sentence-system-concrete PAT-at throw-V.ACT
I throw the book at the wall.
bly-môj pe bly-θaj-môj bly-ķĭm-tla tu-i.
throw-V.RECP and throw-OPP1-V.RECP throw/jump-exercise-professional AGT-at
The acrobats throw and catch each other.

{Ќ} "I, me" is the default agent, experiencer or topic, so it isn't expressed explicitly in the first three examples above.

Time, aspect, mood, etc. are optionally shown with modifier particles following the verb, such as

mweoptative, imperative, hortative, jussive
źǒnegative imperative/optative
šemaybe [facts]
bemaybe [intentions]
de nowadays; lately; (with {mje}) in those days (habitual aspect, extended tense)

If a temporal complement specifies a particular time when the action of the sentence takes place, {mje} or {ler} is usually unnecessary.

-ŋla i sâŋ ĥy-i ķârm-zô.
three-ORD.D on blood few PAT-at cough-V.ACT.
On Tuesday I coughed up a little blood.

Or if the time of a narrative has been specified by a postpositional phrase like the above in one clause, one can generally infer that each subsequent clause's action takes place a little after that of the last, and again {mje} is not needed except to show relative time, that this clause's action takes place before the time of the surrounding narrative; in this use (its most common use in modern gzb) it's more like a perfect aspect marker than a past tense marker. (Early on, I used {mje} and {ler} a lot more often.)

A day of the week mentioned generally refers to the past instance of that day, unless the next such instance is specified by {ler}:

fy-ŋla ler i gâm-ʝĭl kwǒ -i ruŋ- -rĭm-zô.
seven-ORD.D FUT on picture-motion some ATT-at go-V.ACT attention-see-V.ACT
This coming Saturday I'm going to see a movie.

Verb argument structures and the case postpositions

Verb arguments can be marked in five ways:

Within the first two types, there are many specific postpositions and a handful of subordinating conjunctions which can mark verb arguments.

Case-like postpositions can be derived from almost any root word followed by one of the three basic spatial postpositions (most commonly {i}, "at, in"). These are some of the case-like postpositions used most frequently.

ĥy-ipatient (object affected by action)
kâ-iobject of attention
jâ-iin such a state
jâ-řceasing to be, changing from

In active sentences, {-zô} marks the verb and {tu-i} and {ĥy-i} typically mark the agent and patient. These are not the same as subject and object in English and other Indo-European languages; there is no passive voice for verbs. {tu-i} always denotes an animate being who is intentionally doing something. {ĥy-i} always denotes something that is affected by the action of the verb.

Some of the uses of the passive (e.g., saying that something happens without saying who does it) can be rendered by use of {mĭ-i} and {jâ-o}.

bĭm ĥy-i šyj-zô ƥ tu-i.
tub PAT-at clean-V.ACT 3 AGT-at
He cleans the tub.
bĭm mĭ-i šyj-bô jâ-o.
tub TOP-at clean-ADJ state-to
The tub becomes clean.

If the object of the verb is not really acted upon in some way by the agent, another role marker is used for it: for instance, {mĭ-i} or {kâ-i} for object of thought or attention:

ť kâ-i rĭm-van.
2 ATT-at see-V.STATE
I see you.
ljâw-gjâ mĭ-i zym-zô. TOP-at think-V.ACT
I think about linguistics.

If the subject is not actively, intentionally doing something, then it is typically marked with {ʝâr-i} "experiencer" (if animate) or {mĭ-i} "topic" (if inanimate, or if the semantics of the verb are not consistent with experiencer marking).

šî'fy mâ-dân kâ-i ku-van de kâ'θij-ram ʝâr-i.
spirit person-formerly ATT-at hear-V.STATE HAB Cathy-NAME EXP-at
Cathy hears ghosts.
ij'mâks-gam mĭ-i sjum-van terij-ram ʝâr-i.
Emacs-NAME.G TOP-at thankful-V.STATE Terry-NAME EXP-at
Terry is grateful for Emacs.

{ŋâw-o} is used for the object (addressee or listener) of a communication-verb:

dejv-ram ŋâw-o twâ-zô Φǒ hǒ}.
Dave-NAME call-to say-V.ACT QUOTE 2 VOC
I said "Hey!" to Dave.

If the object of the verb didn't already exist, but is created by the action, it's marked by {ķĭn-o} (being constructed, put together from physical materials), {krĭ-o} (being thought up, written, composed, etc.), or {bĭŋ-o} (coming into existence).

mě'hu ķĭn-o -faj-fwa-ƥ-zô.
stew construction-to digest-able-CAUS-3-V.ACT
He's cooking a stew.

gjâ o-m gun krĭ-o zym-zô krĭ-gjâ-pja tu-i.
language to-part.of root.word create-to think-V.ACT create-language-amateur AGT-at
The conlanger thinks up words for [into] the language.

gî'bu i θě'ku tu-i krĭ-zô fî'suň pe mu ble bĭŋ-o.
beginning at God AGT-at create-V.ACT Earth and universe rest.of existence-to.
In the beginning God created the heavens and the earth.

The topic of a topic-comment sentence and the "object" of some verbs of thinking and feeling are marked the same way, with the postposition {mĭ-i}, which was translated in some sentences above as "about".

râm mĭ-i pâ-źa-bô ŋĭn-i.
cat TOP-at restless-AUG-ADJ CMT-at
The cat is hyperactive.
râm mĭ-i zym-zô.
cat TOP-at think-V.ACT

I'm thinking about the cat.

râm tu-i zym-zô.
cat AGT-at think-V.ACT

The cat is thinking [about something unspecified].

râm kâ-i rĭm-van.
cat ATT-at see-V.STATE

I see the cat.

źum-la-zô râm ĥy-i.
touch-AFF-V.ACT cat PAT-at

I stroke the cat.

Some other theta-role postpositions:

ĉul-i transitory object of performance
rjâ-i object of quest or desire
čĭ-ř copied from, extracted from, quoted from
θĭ-o on behalf of, for the benefit of
θĭ-ř by/with the help of
ĥun-i with, together with, in company with
pě'ŝlĭ-i focused body part or mental faculty
jĭrn-i quantity, measure

{ĉul-i} is used for the direct objects of verbs like "sing", "read aloud", "perform", and "play": transitory processes that are over when the action of the verb is over, or which are the action of the verb re-conceptualized as nouns, or transitory process-instantiations of more permanent abstract informational entities.

kuln reŋ ĥun-i {oklahomě-wam-ŋa} ĉul-i hwâwm-ŝrun- ƥ tu-i.
friend many meeting-at Oklahoma-NAME.P-ATD.surprise perform-at acting-music-V.ACT 3 AGT-at
She performed/sang in "Oklahoma!" with several friends.
twâ-cu čĭ-ř du -pa ĉul-i lju-gju-zô.
sentence-system that copy-from section one-ORD perform-at read-speak-V.ACT
I read aloud the first part of that poem.
go-gam ĉul-i pĭw-môj Ќ ke ĭnza-ram.
Go-NAME.G perform-at play-V.RECP 1 and Ynza-NAME
Ynza and I played Go.

{rjâ-i}, object of quest, is used for the objects of verbs of searching, desiring, requesting, etc., that are not necessarily present or even existent, in contrast to {kâ-i}, object of attention, where the object is generally within range of the experiencer's senses.

{Codex Seraphinianus} rjâ-i sru-θô-van.
(title) quest-at desire-DIM-V.STATE
I would like a copy of the Codex Seraphinianus (but am not likely to get it).
sâln-grâm mew rjâ-i kâj-tla ŋâw-o -;
ticket-message also quest-at exchange-professional call-to request-V.ACT
mǒj te-ta-ƥ-van.
but 3.INAN-without-3-V.STATE
I asked the clerk for some postage stamps, too; but she had none.

{θĭ} "helping" derives two different benefactive postpositions:

-ma-ĵĭn θĭ-o θuň krĭ-o lju-θaj-zô.
sibling-meta-young help-to story create-to read-OPP1-V.ACT
I wrote a story for my little cousin.
ĭnza-ram θĭ-ř gjâ-krĭ kujm-o Φyr-tôn-cu krĭ-o.
Ynza-NAME help-from language-create that purpose-to suffix-GNR-system create-to
With Ynza's help I created an inflection paradigm for that conlang.

{pě'ŝlĭ} signifies the body part or mental faculty one's attention is currently focused upon, or which is clamoring for attention. The postposition {pě'ŝlĭ-i} is used in place of {kâ-i} or {mĭ-i} for body parts in some contexts.

tâlm pě'ŝlĭ-i jyn--van.
head focal.part-at pleasure-OPP2-V.STATE
My head hurts. / I have a headache. = lit., I suffer pain focused in my head.
pě'ŝlĭ-i žâw-van plyn-.
digestion focal.part-at feel-V.STATE full-ADJ
I feel full / I don't feel hungry. / lit., I feel fullness focused in the digestive system.

{jĭrn-i} marks a measured quantity that normally applies to the action of the main verb; for instance,

gě'dĭm i twâ-θy gâr-tĭm- jĭrn-i
day DEM1 at sentence-element ten-hundred-ADJ quantity-at
θuň krĭ-o lju-θaj-zô.
story DEM3 create-to read-OPP1-V.ACT
I wrote a thousand words of that story today.
mě'tyr-źa - jĭrn-i ƴâ-cjaj-.
meter-AUG three-ADJ quantity-at go-SPEC-V.ACT
I walked three kilometers.

Polysemy of {mĭ-i}

{mĭ-i} can mark different semantic roles depending on the particular verb or the other postpositions it's used with. This was not originally intended; it's one of the aspects of the language that developed from use rather than being deliberately designed. When I find that some part of the language is more complex or irregular than I intended, I don't necessarily change it; if I've already learned to use it fairly fluently, I leave it alone and just try to document the way I find that I actually use it.

The original design of gzb's case-postposition system was influenced indirectly by Japanese and similar topic-prominent languages — not directly by a serious study of the language, but by a few example sentences and a brief discussion in a linguistics textbook. It was also influenced by my then-misunderstanding of the linguistics term "topic", used in various senses by different schools of linguists and in different branches of linguistics. The result of these misunderstandings is that {mĭ-i}, though I've customarily glossed it as the "topic postposition", has four or five main senses, only one of which even roughly corresponds to the topic marker of a true topic-prominent language like Japanese.

Predication with quality/state term:

huw jâ-i byn-pja mĭ-i.
happiness state-at hack-amateur TOP-at
The hacker is happy.

Predication with quantity:

jĭrn-i θĭl mĭ-i.
eleven amount-at potato TOP-at
There are eleven potatoes (here).

Predication with entity term:

ʝâr-ku-zym ŋĭn-i dejvĭd-ram mĭ-i.
experiencer-hear-thought CMT-at David-NAME TOP-at
David is a passive telepath.

Predication of companionship:

râm ĥun-i suŋ-kě'ĝu-tla mĭ-i.
cat meet-at TOP-at
The wizard is accompanied by a cat.

With verb of thinking:

mluj krĭ-gjâ-za mĭ-i zym-gâw-.
small.conference create-language-ADJ2 TOP-at think-plan-V.ACT
I'm planning for the [next] Language Creation Conference.

With emotion verb:

blě'tâ mĭ-i ħum-van tîm'θĭ.
cockroach TOP-at fear-V.STATE Timothy
Timothy is afraid of cockroaches.

With communication verb:

takahiro-ram ŋâw-o flyr-ĵwa mĭ-i gju-ť- mwe.
Takahiro-NAME call-to flower-place TOP-at speak-2-V.ACT IMP
Talk to Takahiro about the garden.

As topical genitive:

plutarĥǒs-ram dâm-ř lě'kjân mĭ-i twâ-cu kâ-i lju-zô.
Plutarch-NAME authorship-from Alexander TOP-at sentence-system ATT-at read-V.ACT
I'm reading Plutarch's life of Alexander.
hĭnrij-ram ðy--pa mĭ-i gâm krĭ-o ķĭn- hanz-ram holbin-šam ĵĭn-sra- tu-i.
Henry-NAME five-three-ORD TOP-at picture create-to construct-V.ACT Hans-NAME Holbein-NAME.F young-CMP-ADJ AGT-at
Hans Holbein the Younger painted a portrait of Henry VIII.

Inanimate subject of stative verb:

grĭ--van flu-fwa-ĉa pě'lâ- mĭ-i.
function-OPP2-V.STATE flow-CAUS-tool old-ADJ TOP-at
The old pump is malfunctioning.

Animate subject of locative verb:

ħun-tôn-daj tyn-van lîn'kjĭ pe vîl'pun mĭ-i.
pine-GNR-mass DEM1 inside place-V.STATE bobcat and fox TOP-at
There are bobcats and foxes in this forest.

Argument structure classes

In the lexicon, many verb entries (though far from all, unfortunately) contain an explanation of the verb's argument structure, like this:

žy-zô to show, point out (person {tu-i} shows thing {kâ-i} to person {ĥy-i})
ĵulm-van to deserve, to merit; {mĭ-i} deserves {ðĭ-i}

In the table below, these verbs' structures are indicated as "kâ-i ~~~ ĥy-i ~~~ tu-i" or "ðĭ-i ~~~ mĭ-i". A few other notes: {o/ř} represents any motion postposition, and {i} represents any stationary locative postposition. The arguments are listed in the order they would typically appear in a sentence, with the subject at the end. Note that the argument structures of verbs aren't cast in stone; for instance, many argument structures with a {mĭ-i} place can substitute {ʝar-i} instead if the argument is animate, and often some or even all of the arguments can be omitted. Temporal and locative complements can usually be added to almost any verb; {i} or {o/ř} are listed as arguments below only if they often occur with a certain class of verbs.

Argument structure Typical verbs
mĭ-i bly-van, flâň-van
tu-i ķĭm-ca, pjylm-syl-môj
ʝâr-i huw-van, fĭm-van
ĥy-i ~~~ tu-i ĥâ-zô, vâ-oŋ-zô, šâ-zô, tru-zô
kâ-i ~~~ tu-i kâ-rĭm-zô, lju-zô, suŋ-hôw-ca
rjâ-i ~~~ tu-i rě'ju-zô
mĭ-i ~~~ tu-i zym-zô
mĭ-i ~~~ ʝâr-i sjum-van, blâl-van
kâ-i ~~~ ʝâr-i rĭm-van, ku-van, lym-van
rjâ-i ~~~ ʝâr-i sru-van
ðĭ-i ~~~ mĭ-i slân-van, wuŋ-van, ĵulm-van
krĭ-o ~~~ tu-i lju-θaj-zô, gâm-zô
ķĭn-o ~~~ tu-i vâ-oŋ-faj-fwa-zô
kâ-ř ~~~ tu-i zym-ʝǒ-zô
ĥun-i ~~~ mĭ-i gju-môj, ĝu-ðĭl-môj
pě'ŝlĭ-i ~~~ ʝâr-i jyn-cô-van, žâw-van
kâ-i ~~~ ĥy-i ~~~ tu-i žy-zô, ku-fwa-zô
(ŝâj-o) ~~~~ (ŝâj-ř) ~~~ ĥy-i ~~~ tu-i kâj-zô
ŝâj-o ~~~ ĥy-i ~~~ tu-i bwĭl-zô, ₣yjm-zô
ŝâj-ř ~~~ ĥy-i ~~~ tu-i tâň-zô
ŝâj-ř ~~~ ĥy-i ~~~ ʝâr-i bwĭl-θaj-van
o/ř ~~~ tu-i ruŋ-zô, ƴâ-zô, zyŋ-zô
o/ř ~~~ ĥy-i ~~~ tu-i fyn-zô, ₣yjm-zô, čuj-zô
o/ř ~~~ mĭ-i flu-van
i ~~~ mĭ-i tyn-van, ty-van, su-van
ŋâw-o ~~~ mĭ-i / hǒŋ / Φǒ / mĭ-šar ~~~ tu-i twâ-zô, gju-zô, grâm-zô
ŋâw-o ~~~ rjâ-i / hǒŋ ~~~ tu-i lâ-zô
ĉul-i ~~~ tu-i pĭw-zô, ŝrun-zô, lju-gju-zô
ŋâw-o ~~~ čĭ-ř ~~~ tu-i grâm-zô, ĵĭ-zô
ŋĭn-i ~~~ mĭ-i ~~~ tu-i kĭ-zô, zym-zô, twâ-zô
(jâ-o) ~~~ (jâ-ř) ~~~ ĥy-i ~~~ tu-i ₣âl-zô, ĉâm-zô
ðǒŋ mĭ-van, nî'sâ-van, le-tǒj-van
hǒŋ ~~~ tu-i zym-zô
hǒŋ ~~~ ʝâr-i hyw-van, gwě'vu-van

This list is still incomplete, and I'm not sure there's not a simpler way to analyze verbs' arguments structures or a more concise way to present them.


{-van} verbs are not necessarily intransitive, and {-zô} verbs are not necessarily transitive. The distinction is partly between nonvolitional and volitional, and partly between static and dynamic. {-zô} verbs always imply an animate agent. An agentive, dynamic process will always be denoted with a {zô}-verb, unless it's reflexive or reciprocal, and an agentless state will be denoted by a {van} verb (if by a verb at all; or possibly by a postpositional phrase or an adjective). Agentless processes (such as involuntary acts like breathing and seeing, or "acts" of inanimate objects like water flowing) are also denoted by {-van} verbs. I haven't yet worked out the detailed rules for handling agentive states (if it even makes sense to speak of such).

Changes in argument structure with opposite-suffixes

The opposite-suffixes {-θaj} and {-cô} have a more or less predictable effect on the argument structures of verbs. With prototypical ditransitive verbs such as {bwĭl-zô}, "to give", {-θaj} makes the source or recipient of the basic verb the experiencer of a derived verb, leaving the patient the same, and allowing the agent to be omitted, while {-cô} reverses the action of the verb and leaves its argument structure the same.

With verbs of perception, {-θaj} makes the object-of-attention of the stem verb into the topic of the derived verb, and allows the experiencer to be omitted.

lyn kâ-i rĭm-van krĭ-gâm-tla ʝâr-i.
moon ATT-at see-V.STATE create-picture-professional experiencer-at
The artist sees the moon.
rĭm-θaj-van lyn mĭ-i.
see-OPP1-V.STATE moon TOP-at
The moon is visible.

Changes in argument structure with causative suffixes

The causative suffixes {-fwa} and {-hôw} are discussed at greater length in the derivational morphology document. They can be used to derive either modifiers or verbs. Here, I'll briefly describe how the syntax of a verb changes when it becomes causative.

If a verb normally has an experiencer argument and an object of attention argument, making it causative makes the experiencer into a patient, and adds an agent argument who is the one that is causing the patient to undergo the action of the verb. The object of attention argument remains the same. E.g.:

nĭŋ kâ-i ku-van θě'mâ ʝâr-i.
ringing.sound ATT-at hear-V.STATE Thomas EXP-at
Thomas heard a ringing sound.
nĭŋ kâ-i θě'mâ ĥy-i ku-fwa- mî'rĭj tu-i.
ringing.sound ATT-at Thomas PAT-at hear-CAUS-V.ACT Mary AGT-at
Mary made Thomas heard a ringing sound.

A verb of thinking, feeling or knowing, which typically has a topic argument (with {mĭ-i}) or sometimes an object of attention argument, along with an experiencer or agent argument depending on the activeness of the verb, acts in a similar way. The topic or object of attention argument remains unchanged while the experiencer or agent becomes a patient, and a new agent argument appears for the causer.

râ-zla mĭ-i kun-van reŋ lě'kjân ʝâr-i.
event-COLL TOP-at know-V.STATE much Alexander EXP-at
Alexander knows a lot about history.
râ-zla mĭ-i lě'kjân ĥy-i kun-hôw-zô reŋ zě'ĥâr tu-i.
event-COLL TOP-at Alexander PAT-at know-CAUS2-V.ACT much Zachary AGT-at
Zachary taught Alexander a lot about history.

Sometimes the object of attention or topic in a causative of feeling is omitted, in which case we can pragmatically assume that the causer is trying to make the causee feel in such a way about the causer themselves. E.g.:

tîm'θĭ mĭ-i -van lî'zě'bĭ ʝâr-i.
Timothy TOP-at love-V.STATE Ellizabeth EXP-at
Elizabeth loves Timothy.
lî'zě'bĭ ĥy-i fâ-hôw-zô tîm'θĭ tu-i.
Ellizabeth PAT-at love-CAUS2-V.ACT Timothy AGT-at
Timothy is flirting with/courting Elizabeth.

However, if the verb normally takes as patient argument, the above process would result in two patient arguments, it being unclear (especially if both are of the same animacy) which is being caused to do something and which is having something done to them. Some languages allow this with their causatives, but in gzb, the original patient argument becomes a generic {ðĭ-i} argument, as follows:

gî'sĭr ĥy-i vâ-oŋ-zô hě'lî'sjy tu-i.
spaghetti PAT-at digestion-into-V.ACT Alice AGT-at
Alice ate spaghetti.
gî'sĭr ðĭ-i hě'lî'sjy ĥy-i --hôw- sě'râ tu-i.
spaghetti relationship-at Alice PAT-at digestion-into-CAUS2-V.ACT Sarah AGT-at
Sarah fed spaghetti to Alice.

Note that in all these causatives, as well as in lexical causatives like {žy-zô} "to show" (i.e. "cause to see"), the intermediate cause is always marked as the patient, and the causative verb always takes the {-zô} active verb ending.

Argument structures of verbless sentences

Not only topic-comment and experiencer-state structures allow verbless sentences; several of the specific object postpositions allow verbless agentive sentences, with the verb implicit in the semantics of the object and the postposition.

ŝrun-twâ ĉul-i.
music-saying performance-at
I sing a song.
ŝrun-twâ krĭ-o pî'tĭr.
music-saying create-to Peter
Peter composes a song.

{ķĭn-o} in particular is likely to occur in verbless clauses; in my electronic corpus, it occurs with a verb only four times, versus nine times in verbless clauses.

Ќ-ť θĭ-o rî'mâ-daj pe rî'mâ hum-cô-bô ķĭn-o ~~~
1-2 help-to house-place and house deep-OPP2-ADJ make-to ...
Let's build ourselves a city and a tall house...

Default arguments of verbs

As noted before, {Ќ} (I, me) is the default topic/experiencer/agent of a sentence. This applies at the beginning of a text or conversation, when there is no previous context.

I read, am reading.

I'm happily confused.

When there is previous context, however, the default topic, experiencer or agent of each sentence is the same entity as was last explicitly marked as topic, experiencer, or agent. (This also goes for subordinate clauses, whose subject defaults to be the same as in the main clause.)

kâj-kô o ruŋ-zô tam-ram tu-i.
exchange-place to go-V.ACT Tom-NAME AGT-at
re i gâ-rjâ ĥy-i tru-zô heŋ.
3.PLACE at thing-quest PAT-at find-V.ACT not

Tom went to the store. [He] didn't find what he was looking for there.

A vocative phrase sets the default subject, as well:

?najĝěl-ram , râm ðĭ-i ħě'nâw-van zǒn.
Nigel-NAME VOC cat relationship-at allergy-V.STATE Q.YN
Nigel, are you allergic to cats?
kĭn-ram hǒ, ty-o ruŋ- vjurm- mwe.
Ken-NAME VOC 1 home-to go-V.ACT visit-V.ACT IMP
Ken, come visit me at home.

This means that agentless processes and states have to be expressed otherwise than with subjectless verbs, as in Esperanto's "pluvas", "necesas ke...". If I translated those literally with just a bare verb, the result would mean "I rain, I am necessary that...", or depending on the context, it might attribute these actions to any random entity that was recently mentioned. Nor do I use a dummy subject as in English "It's raining".

bly-van pwĭm mĭ-i
fall-V.STATE water TOP-at

Water is falling = It's raining.

jâln-van purj mĭ-i.
hot-V.STATE environment TOP-at

It's hot.

If the subject of the sentence comes last (as it usually does) then the final postposition (agent, experiencer, or topic) may be omitted.

gju-zô kun-hôw-tla.
speak-V.ACT know-CAUS2-professional
The teacher speaks.
žuln-van byn-pja. hack-amateur

The hacker is pleased with his work.

This applies also if the subject of a clause is the last thing in that clause, even if it's not the end of the sentence; for instance,

twâ-zô dě'dâ-lô Φǒ {bly-van drě'mâj reŋ.}
say-V.ACT Dada-member QUOT fall-V.STATE emu many
The Dadaiste said "It's raining emus."

{dě'dâ-lô} immediately precedes the quote-conjunction {Φǒ}, so it needn't be explicitly marked as subject (in this case, agent) of {twâ-zô}. The same is true for subjects preceding other clause-scope conjunctions, such as {kiň} "and", {hǒŋ} "that", and {vĭj-šar} "then, next".

Ditransitive verbs

In Indo-European languages, the term "ditransitive" is used for verbs that are pretty much required to have both a direct and an indirect object. I use the same term for verbs in gzb that usually if not always occur with two different objects, though with gzb's theta-role marking system, it's not always easy to say which of the core arguments for a given gzb ditransitive would correspond to a direct vs. an indirect object in another language; and gzb, like English, is prone to dropping arguments that are obvious from context (e.g., "Give it here" where the theoretically required "to me" indirect object argument is omitted).

Probably the prototypical ditransitive verb in most languages is the equivalent of "give": gzb {bwĭl-zô},

vělĭriě-ram ŝâj-o wrym-θym mâŋ-paj ĥy-i bwĭl- ĝejsn-ram.
Valeria-NAME having-to ornament-torus hand-for PAT-at give-V.ACT Jason-NAME
Jason gave a bracelet to Valeria.

Here the gift is marked with {ĥy-i}, patient, and the recipient of the gift is marked with {ŝâj-o}, coming-into-the-possession-of. Straightforward enough. The verb {kâj-zô} "buy/sell/trade" has a similar argument structure, but can have additional arguments.

gâm prym-fwa -i Ќ ĥy-i žy- krĭ-gâm-tla pǒ.
picture aesthetic.appreciation-CAUS ATT-at 1 PAT-at show-V.ACT create-picture-professional DEM3
That artist showed me a beautiful picture.

Here, the person being shown something is the patient {ĥy-i}, and the thing being shown gets the attentive case postposition, {kâ-i}. The simpler English equivalent is ambiguous (in how one would analyze the cases of "me" and "a beautiful picture", I mean, not in what it means), but the more formal version would be "...showed a beautiful picture to me", i.e. the picture is the direct object; and the same would be true in the other Indo-European languages I'm familiar with.

The thinking verb {kĭ-zô} "to deem / consider X to have quality Y" adds an agent to what would otherwise be a simple topic-comment sentence. E.g.,

ħâl-fwa ŋĭn-i ₣â -i.
nervous.fear-CAUS CMT-at task DEM3 TOP-at
This task induces the jitters.
ħâl-fwa ŋĭn-i ₣â -i - pǒlin-ram tu-i.
nervous.fear-CAUS CMT-at task DEM3 TOP-at deem-V.ACT Pauline-NAME AGT-at
Pauline considers this task a fearful one.

In other words, the "direct" and "indirect" objects of {kĭ-zô} and similar verbs are marked as topic and comment. ({kĭ-zô} also has an alternate argument structure, where it takes an object subordinate clause marked with {hǒŋ}.)

The verb {ðĭl-zô}, "to type or transcribe", is potentially ditransitive in the second sense:

twâ-cu-hân lju-i pě'pâ-ga om ðĭl-Ќ-zô.
sentence-system-old DEM1 read-at page-METAPH into type-1-V.ACT
I'm transcribing this old book into an electronic document.

Here the objects are marked as {lju-i}, a kind of performative case more specific than {ĉul-i}, and {om}, becoming-part-of. However, I'm not sure {ðĭl-zô} actually qualifies as ditransitive since the {om} argument is optional.

The opposite-suffix {-θaj} as used with some stems that form ditransitive verbs makes the source/recipient of the basic verb the experiencer of a derived verb, leaving the patient the same, and allowing the agent to be omitted.

wrym-θym ĥy-i bwĭl-θaj-van vělĭriě-ram ʝâr-i.
ornament-torus PAT-at give-OPP1-V.STATE Valeria-NAME experiencer-at
Valeria received a bracelet as a gift.

Giving and receiving are one process; {bwĭl-zô} and {bwĭl-θaj-van} simply focus on different aspects of it from different participants' perspective. The other opposite-suffix {-cô} does not necessarily affect the theta-roles of the participants in the action of the verb, but it changes the meaning of the underlying action:

vělĭriě-ram ŝâj-ř wrym-θym ĥy-i bwĭl-- tesě-ram tu-i.
Valeria-NAME possession-from ornament-torus PAT-at give-OPP2-V.ACT Tessa-NAME AGT-at
Tessa stole a bracelet from Valeria.

Reflexive and reciprocal verbs

A reflexive verb can occur with an explicit patient, topic or attentive postpositional phrase; usually this signifies a body part or faculty of the agent or experiencer, e.g.:

!maŋ ĥy-i šyj-ca mwe ť.
hand PAT-at clean-V.REFL IMP 2

Wash your hands.

Ќ im tâlm vin kâ-i rĭm-ca-ĉa syj-i rĭm-ca.
1 part.of head front-surface ATT-at see-V.REFL-tool use-at see-V.REFL

I see my face in [using] the mirror.

gě'dĭm pen šin žâj-ŋĭw kâ-i byn-ca mwe Ł.
sleep.wake.cycle every end-of ATT-at poke.around-V.REFL IMP 3.GEN

One should examine one's conscience every night.

Because gjâ-zym-byn does not have a sharp distinction between direct objects and oblique objects, the reflexive and reciprocal verbs formed with {-ca} and {-môj} sometimes have as their reflexive objects things which would be expressed with oblique objects or complements in other languages. A few verbs tend to almost always take {-môj} when the subject is plural.

hyr srǒ il gju-môj tam-ram pe ser'ě-ram.
hour several during speak-V.REFL Tom-NAME and Sarah-NAME

Tom and Sarah talked [with each other] for hours.

Logically, perhaps, {gju-môj} ought to mean "to talk about each other". But one of my design principles for gjâ-zym-byn is not to change something if I've already learned to use it fluently, just because I later decide it's not perfectly logical. I did not design any irregularity into the language deliberately, but since my goal is to learn to use it fluently myself, and not to devise a language that's easy for people in general to learn, I'm perfectly happy with keeping any irregularity that creeps into the language through my occasional carelessness, if I don't notice it's irregular until I've already learned it.

Sequential verbs

Where the typical Indo-European language would use an auxiliary verb followed by a particple or infinitive, gjâ-zym-byn just uses two verbs in sequence. The first verb roughly corresponds to an auxiliary in English or French and the second verb roughly corresponds to an infinitive, though it gets no special marking. Either verb can be marked with {-van} or {-zô} according to its meaning.

âθ'ĭnz-wam o sru-van ruŋ-zô.
Athens-NAME.P to want-V.STATE go-V.ACT

I want to go to Athens. [= Athens, Georgia; the Greek one is {a'θen'aj'ǒs-wam}, the one in Kentucky is {ej'θĭnz-wam}.]

dlu-van heŋ huw-Ł-van.
right-V.STATE NEG happy-3.GEN-V.STATE
One doesn't have a right to be happy.
mǒj dlu-van vǒm rjâ-zô huw-van.
but right-V.STATE yes seek-V.ACT happy-V.STATE

But one does have a right to seek to be happy.

Note that modifiers to the main verb can come between it and the second verb, as with {heŋ} above.

Sometimes the first of a sequence of verbs is not a typical auxiliary verb.

gâm-ʝĭl -i ruŋ- rĭm-van ler. that ATT-at go-V.ACT see-V.STATE FUT
I'm going to see that movie.
mî'ħâ-van krĭ-šĭm- byn-pja.
obsession-V.STATE create-algorithm-V.ACT hack-amateur
The hacker is obsessively coding.

Sometimes arguments of the second verb in a sequence come between the two verbs:

gâw-zô gjâ-krĭ bâm- krĭ-o byn-zô.
plan-V.ACT language-create new-ADJ create-at hack-V.ACT
I plan to tinker with my new conlang.
suŋ-ƥ-van fîn'cĭ ĥun-i gju-môj. penguin meeting-at talk-V.RECP
She knows how to talk with penguins.

But if there are multiple arguments to the verb, or one argument with a lot of words, it's probably better style/less confusing to put the second verb and its arguments in a subordinate clause (with the appropriate conjunction for the main verb's meaning) or relative clause, e.g.,

gâw-zô hǒŋ fy-ŋla il gjâ-krĭ bâm- krĭ-o byn-zô.
plan-V.ACT that seven-ORD.D through language-create new-ADJ create-at hack-V.ACT
I plan to tinker with my new conlang all day Saturday.
suŋ-ƥ-van ru i fîn'cĭ-ĵĭn sru-źě'fy- ĥun-i gju-môj. manner REL at penguin-young desire-success-ADJ meeting-at talk-V.RECP
She knows how to talk with ambitious young penguins.

Subject Pronoun Incorporation

gjâ-zym-byn can optionally incorporate a subject pronoun into the verb; it affixes between the verb stem and the verb suffix. For serial verbs, the pronoun will generally only be incorporated into the first of the series.

You're beautiful.
twâ-cu -i vy-ƥ-van lju-zô.
sentence-system DEM3 ATT-at intend-3-V.STATE read-V.ACT
She intends to read that book.

Such pronoun incorporation is usually done only when

kelij-ram ĥun-i gju-môj. Ќ ŋâw-o twâ-ƥ-zô,
Kelly-NAME meet-at talk-V.RECP 1 call-to say-3-V.ACT
hǒŋ fĭm--ť-van.
that healthy-OPP2-2-V.STATE
I talked with Kelly. She told me you were sick.

First-person plural subject attraction

In gzb, sometimes a comitative phrase expressed with {ĥun-i}, "with", will influence the verb form and/or the subject, making the verb reciprocal or the subject plural, thus:

vâl-ram ĥun-i ŝě'ĥâ-môj.
Val-NAME meeting-at chess-V.RECP
lit., We played chess with Val.
tam-ram ĥun-i re o ruŋ- Ќ-ƥ.
Tom-NAME meeting-at 3.PLACE to go-V.ACT 1-3
lit., We went there with Tom.

English would express these as "I played chess with Val" or "I went there with Tom."


gjâ-zym-byn does not have grammatical category of aspect as such, but several aspectual distinctions are commonly marked by affixes or by root words compounded into verbs.


ť dâm-ř grâm kâ-i lju-sun-zô mje θǒ.
2 authorship-from message ATT-at read-finish-V.ACT past immediate
I've just finished reading your letter.


vâ-oŋ-vĭj-zô ƥ.
digestion-into-time.period-V.ACT 3
He goes on eating.


pĭw-gĭn-zô θǒ Ќ-ƥ, nu-šar vě'ty-θaj ĥy-i trâw-zô kwǒ.
play-begin-V.ACT immediate 1-3 moment-CONJ doorway-OPP PAT-at strike-V.ACT person some
We had just started playing when someone knocked at the door.


kyl-pwĭm-daj rol-lol čâ-ra-zô lu'ĭs-ram.
box-water-mass across.through-hither.through swim-repeat-V.ACT Louis-NAME
Louis swam [laps] across the pool several times.


ķarm-nu-zô ku-faj-źa râm.
cough-moment-V.ACT hear-able-AUG cat
The cat coughed once loudly.

The adverb {de} already mentioned marks a habitual aspect.

tâŋ i, nĭvĭn-šam dâm-ř θuň reŋ kâ-i lju-zô de.
life.period DEM3 at Niven-NAME.F authorship-from story many ATT-at read-V.ACT HAB
I was reading a lot of stories by Niven in those days.

{de} can also mark e.g. the day of the week when something is regularly done or regularly happens.

kru-ŋla de i ĥrî'cu-ķam-vuj -i tî'šâ-zô.
cross-ORD.D HAB at Christ-NAME.T-physical ATT-at worship-V.ACT
On Fridays, I go to Eucharistic adoration.

See also the section on the qualifier {jǒm} "most of, mostly" for examples of its aspectual use with verbs.

Gerunds and Participles

There is no need for special morphology to mark gerunds, since the root words for actions, events and processes are already nominal.

ty ruŋ š-i-j, vâ-oŋ-zô.
home into going after-at-near digestion-into-V.ACT.
Soon after coming home, I ate.

hwâwm mĭ-i suŋ-hôw-zô rěbekě-ram tu-i.
acting TOP-at Rebecca-NAME AGT-at
Rebecca teaches acting.

When a verb is derived indirectly (e.g. from a postpositional phrase), there is no root noun that means the same thing as the verb, so one can use the nominalizer clitic {tǒj} to obtain such a gerund:

ĥun-pĭw ðij vâ-oŋ-tǒj mĭ-i gâw-zô, mǒj ce heŋ.
meeting-play before digestion-into-NMZ TOP-at consider-V.ACT but this not.
I considered eating before the party, but decided not to.

Any direct object must immediately precede the gerund; the verb or comment on the gerund clause usually comes after the gerund.

₣ĭŋ kyl-plâŋ-za ĥy-i lĭn žu-bô mĭ-i hum-ga-van.
string box-foot-ADJ2 PAT-at linking careful-ADJ TOP-at deep-MET-V.STATE
It's important to tie [one's] shoestrings carefully.

Use {tu} "agent" and {ĥy} "patient" to form nominal participles.

ljuact or process of reading
lju-zôto read; I read, he reads, ...
tu-lju reader; person reading
ĥy-ljuthe thing read

Note that this use of {ĥy} is not entirely consistent with the way the verb {lju-zô} is used. Reading may affect the physical book {twâ-cu-vuj} (in terms of slight wear and tear) but it does not affect the text of the book {twâ-cu} (abstracted from its instantiation in particular printed copies). So normally one would use the attentive case postposition instead of the patient case:

twâ-cu ĵyn-fwa kâ-i lju-zô.
sentence-system interest-CAUS ATT-at read-V.ACT
I'm reading an interesting book.

twâ-cu-vuj hân-bô nâ-cô-bô ĥy-i lju-zô žu-bô mwe Ł.
sentence-system-physical old-ADJ common-OPP2-ADJ PAT-at read-V.ACT careful-ADJ IMP 3.GEN
One must read rare old books carefully.

The first form (with {kâ-i}) emphasizes the content of the book (and doesn't specify its format, whether it is printed, an etext, or even an audiobook). The second, with {ĥy-i}, emphasizes the physical act of handling the book, turning the pages. The first is by far the more common way of marking the "direct object" of {lju-zô}. So does {ĥy-lju} refer mainly to a physical book, magazine, etc.? Not necessarily. {kâ-lju} would mean something very different: "reading attention", or "act of attention characterized in some way by reading". So it could not refer to the "thing read" in the sense of the content of a book as distinct from its embodiment in a particular copy of a particular edition. Therefore {ĥy-lju} has to do double duty for both senses, and in short {ĥy} is not so specific when acting as a participle base as when it is acting as a postposition base. (If necessary, one can be more specific by referring to {ĥy-lju-vuj}, physical thing read, or {ĥy-lju-vuj-cô}, abstract thing read.)

One can add {-bô} to these nominal participles to form modifer participles:

ƴâw-bâm tu-pĭw-bô kâ-i pym-van.
dog-new AGT-play-ADJ ATT-at amusement-V.STATE

I'm amused at the puppy playing.

rjuŋ kâ-i ħun-tôn-daj kiŋ tru-zô tu-pĭw-bô mâ-ĵĭn.
dragon ATT-at pine.tree-GNR-mass among find-V.ACT AGT-play-ADJ person-young

The children found a dragon [while] playing in the forest.

ĥy-tru v-ř ruŋ-zô ĵwy-bô.
PAT-find front-from go-V.ACT fast-ADJ

They ran away from what they found.

Experiencer participles

The root word {ʝâr}, "experiencer", also forms a kind of participles.

ʝâr-pym one who is experiencing amusement
ʝâr-fĭm-cô one who is experiencing sickness; a sick person or animal
ʝâr-bly one who is falling
ʝâr-ħĭn one who is experiencing restrictions; a prisoner
ĝyl-fyn -i ʝâr-rĭm - ĥy-i tru- Ќ-ɱ gwe.
interruption-drive TOP-at experiencer-see three-ADJ PAT-at find-V.ACT 1-3 already
We've found three witnesses to the accident so far.
Ќ o ruŋ- mwe ʝâr-ĵyj-fja pen pe tu-šâ-wâj pen,
1 to come-V.ACT IMP experiencer-vigor-minimum all and AGT-carry-heavy all
kiň ť ĥy-i ĵyj-rjâ -o Ќ.
and 2 PAT-at vigor-seeking state-to 1
Come to me, all who are tired and carrying heavy things, and I will give you rest.


The core postpositions are:

iat, in, near, with; during
řfrom, out of; since
oto, toward; until

One can make them more specific with various other single-phoneme morphemes prefixed (for orientation) or suffixed (for proximity). These prefixes and suffixes occur only with these core spacetime postpositions, and never affix to any other morpheme.

Suffixes: being near, far or inside:

-min (part of)
in (contained by)
-ntouching the outside of
-rfar from
-lthrough, throughout, all through

So, for instance,

řŋout of
impart of
ojtoward but not (yet) at
irfar from
řlthrough (coming this direction, toward the speaker)

These morphemes show orientation about a center:


v-in front of
k-among, between
ĉ-all around, surrounding
r-at, to, from the other side of
l-at, to, from this side of
š-after, later part of (time)
ð-before, early part of (time)



Example spacetime postpositions:

sijabove (not touching)
sinon (touching the surface of)
siŋin the upper part of
(rî'mâ siŋ pě'pâ-daj, papers in the attic)
sim in the upper part of
(Ќ sim šĭm-ŋĭw, my brain)
sogoing above
sřnoff of
θijunder (not touching)
θinunder (touching)
θo going under
θř from under
θoŋ into the lower part of
ĵi on the right side of
ci on the left side of
vi in front of
hi behind
hiŋ in the back part of
from in front
vo to in front
kinbetween (touching the things it's between, e.g. a bookmark between pages)
kiŋthroughout (mushrooms scattered through a forest)
ilthrough (a road going through a forest)
olthrough (a man walks through a forest)
rirfar beyond
roncoming to touch the far side of
lion this side of
ĉi surrounding
ĉoŋinto from all sides
źi on the west side of
ħř from the south of
i(ŋ)during, while
ðo(n)until, up to
šř(n)since, from that time

Complex directions can be specified by using two of the prefixes and inserting an epenthetic schwa between them, thus:

běźir far to the northwest of
sěviŋ in the upper front part of

Uses of "before" and "after" postpositions {ði} and {ši}

These "before" and "after" postpositions (and their derivatives) are used not only with nouns and noun phrases denoting time periods, but with words for other things that are conceived of as having their extension primarily in time rather than space.

frâ š-i-m gjâ-θy {zǒn} rej {srem}
question after-at-part.of language-element "zǒn" or "srem"
tyn-van ʝel, pwiň frâ i-m bu-kyr š-i-m tyn-te-van.
place-V.STATE generally or question at-part.of phase-verb after-at-part.of place-3.INAN-V.STATE
The particle "zǒn" or "srem" is placed at the end of a yes/no question, or at the end of a verb phrase within the question.
θuň-ba ð-i-m tyn-van -fĭw ĵyn-fwa -i,
story-ATD3 that before-at-part.of place-V.STATE person-fictional interest-CAUS TOP-at
mǒj te ĝy-i-m tâň-van ƥ.
but 3.INAN middle-at-part.of removal-V.STATE 3
Several interesting characters appear in the early part of that story, but they disappear in the middle of it.

A few of the 357 postpositions one can form in this system don't make any sense. But most of them are potentially usable in some situation or other. For instance, {šom}, "becoming part of the ending of" could be used if one is talking about reforming a calendar system and reassigning some days from the beginning of one month to the end of the previous month, or, less farfetchedly,

mluj š-o-m ĥun-frâ ĥy-i tâň-θaj-Ł-zô.
convention end-to-part.of meeting-question PAT-at take-OPP1-3GEN-V.ACT
They added a question and answer session at the end of the convention.

Postpositions in {ĉ-}

Most of the spacetime postpositions are fairly straightforward, but those in {ĉ-} require some more explanation. "ĉi" refers to a position surrounding the object on all sides, "ĉo" to motion of something that begins to surround the object. Neither refers to going around something, circumnavigating it. The adverbs "ŝwe" (widdershins) and "ŝwe-θaj" (clockwise) can be used together with "ĉi" or "ĉo" to indicate such motion.

rîmâ-źa ĉ-i tyn-van pwĭm-daj.
house-AUG around-at place-V.STATE water-mass
There is a moat around the castle.
rîmâ-źa ĉ-o ruŋ-zô sî'ðyr-tla-cu.
house-AUG around-to go-V.ACT fight-professional-system
The army surrounded the castle.
rîmâ-ĵwa jeriĥo-wam ĉ-i ŝwe ƴâ-zô fy-bô jisrael-tam-cu.
house-place Jericho-NAME.P around-at widdershins walk-V.ACT seven-ADJ Israel-NAME.E-system
The Israelites marched around Jericho seven times.

Serial postpositions

Sometimes two postpositions in a row are used, the second postposition modifying the first.

swyŋ s-i-n ĥy-i ĉârn- šyj-zô.
table top-at-contact PAT-at abrade-V.ACT clean-V.ACT
I scrub the surface of the table.
mruň ħ-i-m Φâ -i kujm-o re o ruŋ-zô.
mountain south-at-part.of form ATT-at motive-to there to go.V.ACT
I went there to see the carvings in the south side of the mountain.
žĭr gân-ř luw-tâlm i-ŋ -i rĭm-van
humming cause-from bone-head at-inside ATT-at see-V.STATE
kiň te i-ŋ pî'dâ-daj -i tru-zô.
and 3.INAN at-inside bee-COLL ATT-at find-V.ACT
Because of the humming noise I looked inside the skull, and found a swarm of bees in it.

Sometimes a serial postposition can be analyzed as involving an omission of an obvious default noun between the first and second postposition; for instance, the common sequence {dâm-ř kâ-i} following an author's name:

taměs-ram pejn-šam dâm-ř -i lju- jǒj.
Thomas-NAME Payne-NAME.F authorship-from ATT-at read-V.ACT again
I'm reading Thomas Payne [= some salient book he wrote] again.

This could probably be interpreted as an abbreviated form of

taměs-ram pejn-šam dâm-ř {Describing Morphosyntax} -i lju- jǒj.
Thomas-NAME Payne-NAME.F authorship-from [book_title] ATT-at read-V.ACT again

or of

taměs-ram pejn-šam dâm-ř twâ-cu -i lju- jǒj.
Thomas-NAME Payne-NAME.F authorship-from sentence-system ATT-at read-V.ACT again
I'm reading a book by Thomas Payne again.

A time-postposition following another postposition is a common pattern, especially with the state-transition postpositions {jâ-ř} and {jâ-o}, but also with some others; e.g.,

mwĭl -ř šin šyj-ca θǒ.
sleep state-from immediately.after clean-V.REFL next
After waking up I immediately bathed/showered.
blâl hǒl -o ðij, tyn ř ruŋ- mwe.
frustration total state-to shortly.before place DEM1 from go-V.ACT IMP
I should get out of here before I get totally frustrated.

These before/after time postpositions can follow various object-case postpositions, in which case a default verb appropriate to that case is usually implied; for instance,

{Hamlet} -i ši, {Rosencrantz and Guildenstern are Dead} -i lju- de.
[title] ATT-at after [title] ATT-at read-V.ACT HAB
Typically after [I read] "Hamlet", I read "Rosencrantz and Guildenstern are Dead".

Purposive postpositions such as {rjâ-i} "in quest of" and {kujm-o} "for the purpose of" can also commonly follow other postpositions, including the state transition postpositions:

prym -o rjâ-i ŝrun -i -ku-zô. state-to quest-at music ATT-at attention-hear-V.ACT
I listen to music seeking to enter a state of appreciation-of-beauty.
gâm-ʝĭl -i kujm-o vě'ty-rĭm ĥy-i kě'ĝu-ť- mwe.
picture-motion ATT-at purpose-to door-seeing PAT-at hide-2-V.ACT IMP
Close the window-blinds so we can focus on the movie.

Indicating posture and position

A sequence of two spacetime postpositions — usually a part-whole postposition followed by a contact postposition, both built on {i} "at" — can be used to indicate posture and position. Usually a locative verb of some kind is used with this sequence. For instance:

Ќ him sin su-ʝa-van
1 back.part.of on standing-ROT-V.STATE
I'm lying on my back. / lying supine.
Ќ vim sin su-ʝa-van
1 front.part.of on standing-ROT-V.STATE
I'm lying on my front. / lying prone.

Or if the second postposition in the sequence is a motion postposition (built on {o} "to" or {ř} "from"), it can indicate getting into or out of a given position.

ɱ ĵim son mlĭr-ʝa-ca zě'ĥâr.
3 left.part.of onto rotate-ROT-V.REFL Zachary
Zachary rolled over onto his right side.
ɱ him sřn su jâ-o tîm'θĭ.
3 back.part.of off.of standing state-to Timothy
Timothy stands up from a supine position.

More precise paths of motion

Two sequential motion postpositions, one built on {ř} "from" followed by one built on {o} "to", can describe the path of motion relative to the head of the postpositional phrase more precisely than a single spacetime postposition can do. For instance,

rî'mâ θoŋ ruŋ-zô.
house bottom.into go-V.ACT
I go into the lower part of the house / into the basement.

The {θoŋ} alone doesn't specify whether I'm starting out upstairs, and going downstairs, or whether I'm starting out outdoors, and entering the house on the basement level. Sometimes context distinguishes, but one can also use two sequential postpositions:

rî'mâ sřŋ θoŋ ruŋ-zô.
house top.out.of bottom.into go-V.ACT
I go downstairs.
rî'mâ źřj θoŋ ruŋ-zô.
house west.from.near bottom.into go-V.ACT
I enter the basement from the west side of the house.

The {ź-} prefix isn't necessary here; replacing {źřj} with simple {řj} would yield a less precise but still accurate meaning.

This is also how gjâ-zym-byn describes directions of motion more generally, in situations where English, for instance, would use a -ward adverb such as "north(ward)", "forward", etc.

purj břŋ ħoŋ ruŋ-źa-zô num.
environment north.out.of south.into go-AUG-V.ACT wren
The wren voyages south.
ɱ ĵř co ƴâ-zô dě'pyw.
3 right.from walk-V.ACT crab
The crab walks to its left.

In the first example, the southward motion is described relative to the environment as a whole; in the second, it's relative to the crab's own body, with the cataphoric pronoun {ɱ} pointing forward to the crab, which hasn't been mentioned yet.

Similarly a combination of two before/after postpositions is used to indicate direction of time-travel, e.g.,

sî'ðyr-źa-cu ðoj šř ruŋ-ƴu-zô.
fight-AUG-system after.from go-time-V.ACT
I time-traveled back to shortly before the war.

If the speaker just used {ðoj} by itself, the sentence would be ambiguous as to whether they were going to that point in time from some point even earlier before the war, or from a future point.

Abstract postpositions

The abstract case markers all derive from a root word followed by a basic {i, o, ř} postposition.


The relationships shown by the English possessive or the Greek genitive are shown in various ways in gjâ-zym-byn:

ŝâjhaving stuff
ŝâj-iof (belonging to)
lĭwpersonal relationship
lĭw-iof (related to)
dâm-řof (by)

The partitive genitive would sometimes be translated with the suffix {-na} ("made of" the substance described by the root). A few other concepts denoted by prepositions in other languages are denoted by suffixes here as well: {-ta} "without", {-ja} "according to, fitting".

Some other useful non-spacetime postpositions:

muw-i subset of; one of; out of; among
syj use, utility
syj-i with, using
gân cause, reason
gân-ř because of, on account of
kujm motive, goal, purpose, reason
kujm-o in order to, for the purpose of
ðĭ relationship
ðĭ-i in some unspecified relationship with

{ðĭ-i} roughly corresponds to Esperanto's generic preposition "je". Its most common use is to mark the object of a stative verb when the subject must be marked with {mĭ-i} rather than {ʝâr-i}, and no other postposition seems more precisely fitting for the object. The possession and ownership verbs {ŝâj-van} and {wuŋ-van} are the most common such verbs.

hajnlajn-šam dâm-ř twâ-cu
Heinlein-NAME.F authorship-from sentence-system
ðy-ðy-lwa-bô ðĭ-i wuŋ-van.
five-five-approximate-ADJ relationship-at own-V.STATE

I own about twenty-five books by Heinlein.

{muw-i} is used to indicate that the entity or group of entities denoted by one noun phrase is a member or subset of another set. It translates among other things some uses of the English phrases "one of", "some of", "among" and "out of":

lju-θaj-tla muw-i tu-i twâ- Φǒ:
read-OPP1-professional subset-at person certain AGT-at say-V.ACT QUOTE
{hyw-hôw-tla hǒ, frâ-θaj- žuln-fwa ť tu-i.}
know-CAUS-professional VOC ask-OPP1-V.ACT satisfaction-CAUS 2 AGT-at
Then one of the scribes said, "Teacher, you have answered well."
-ŝy muw-i ť -i ₣urŋ- ŋĭn-i ~~~
person-female subset-at 2 TOP-at blessed-ADJ CMT-at ...
Blessed art thou among women...

When used to describe the proportion of a given set with a certain property or engaging in a certain action, the noun used before {muw-i} doesn't need to be repeated after it, just the numerator.

fĭm-hôw-tla ðy- muw-i ĉu-ĉu- ʝâr-i
health-CAUS-professional five-ADJ subset-at two-two-ADJ experiencer-at
blâl-van ŋî'bĭ fĭw- -i.
annoyance-V.STATE number fiction-ADJ TOP-at
Four out of five doctors get annoyed at made-up statistics.

There is an implied {fĭm-hôw-tla} between {muw-i} and {ĉu-cŭ-bô} here, as in English "Four [doctors] out of five doctors..."

Lexical dissimilation

When an abstract postposition would occur after the same noun that it's built upon, it is generally replaced with a plain spacetime postposition for the sake of euphony, or sometimes replaced with a less specific/accurate abstractpostposition. E.g., substituting a simplified postposition:

? i gju-ŋĭn-zô θě'mâ.
topic Q.WH at speak-explain-V.ACT Thomas
What topic did Thomas lecture about?

— as opposed to {?mĭ nǒ mĭ-i}.

Substituting a less specific postposition:

ŝĭw syj-i bĭm ĥy-i šyj-ť- mwe.
substance this use-at sink PAT-at clean-2-V.ACT IMP
Clean the sink using this substance.

— as opposed to {ŝĭw kǒ ŝĭw-i}. {ŝĭw-i} is normally used for the materials one uses up to do the action of the verb, while {syj-i} is used for tools that are relatively unaffected by a single use. But when the substanceused is so vaguely described as above, {syj-i} is used to avoid repeating {ŝĭw}so close together.

I call this phenomenon "lexical dissimilation." A similar kind of dissimilation occurs withthe suffix {-ma} where it replaces a repeated instance of the same morpheme in a derived word. Some of the other postpositions this occurs most commonly with include {kujm-o} "for the purpose of," and {gân-ř} "because of."

kujm o bâň-van heŋ ðǒŋ tě'bâ ĥy-i kyw--Ł-.
purpose this to allow-V.STATE not that tobacco PAT-at lung-into-3.GEN-V.ACT
That's why it is not allowed to smoke tobacco.
gân kwǒ ř grĭ-van heŋ flu-fwa-ĉa.
cause some from function-V.STATE not flow-CAUS-tool
For some reason the pump's not working.

Note on why I derive abstract postpositions from spacetime postpositions

A lot of natlangs and naturalistic conlangs cover a lot of their needs for abstract relationship adpositions by metaphorically stretching the basic meaning of a locational or temporal adposition. gzb does a lot less of that (I won't say none at all); instead of, for instance, using {si} to mean both "physically on top of" and "about the topic of", or "hi" to mean both "physically behind" and "being the cause of", it forms abstract relationship postpositions by compounding an abstract noun root with one of the basic spatial relationship postpositon. The spatio-temporal meaning of {i} remains intact when it stands alone or combines with spacetime prefixes and suffixes; it generally only gets stretched into abstract territory when it's combined with an abstract noun, and then the noun it's combined with generally indicates pretty clearly what that stretched meaning is.

Hypothetically, I could have simply extended the meanings of those abstract nouns into postpositions by zero-derivation, the way English does a lot of its derivation between the categories of noun, verb and adjective. But the way gzb syntax works, that would have made a lot of sentences hard to parse, I think. Or I could have lexicalized the most common thematic-relationship concepts simply as postpositions, and then used a nominalizer when I needed to use the corresponding abstract noun; but then I would have either needed a more complex morphophonology, to give room for more allowed wordforms in the postposition class, or I still would have needed some way to derive less-common thematic-relationship postpositions from abstract nouns.

Or I could have formed abstract postpositions from abstract nouns with one or more normal suffixes, the way I produce conjunctions from noun stems with {-šar}, adjectives with {-bô} and others, etc. But it seemed more intuitive to use the core spacetime postpositions, which were already there, instead of adding more morphemes and making words that were already longer than most natlang equivalents even longer (the shortest possible suffix of normal form is two phonemes, CV). Also, I would have needed three postposition-making suffixes to produce a system as flexible and expressive as the one gzb has now — notice the way the variation between the spacetime postpositions {i, o, ř} marks aspect in {jâ-i}, {jâ-o}, and {jâ-ř}; there are other triplets of related case postpositions as well.

I intended that the postposition system should have its inevitable complexity all on the surface, manifesting as a large number of postpositions with fairly specific meanings and uses, instead of it being hidden, as in most natlangs, which have a smaller number of adpositions and cases but each has a bewildering arrray of meanings and rules for when to use it. However, I'm not sure I succeeded; in the process of pulling most (not all) of the unavoidable complexity to the surface, I may have created some otherwise avoidable complexity.

Modifiers (adjectives/adverbs)

Modifiers are formed from root substantives by addition of appropriate suffixes. gjâ-zym-byn does not make a morphological distinction between adjectives and adverbs, though in practice some modifiers always modify verbs or modifiers. Modifiers, whether non-derived particles or derived words, always follow their heads, with a handful of exceptions: interjections like {hwǒ}, evidentiality adverbs derived with {-pôm}, and ordinal adverbs derived with {-saw}, which can optionally come at the beginning of a clause instead of after the main verb (or the most verb-like postpositional phrase, in a sentence with no verb).

If the root noun denotes a quality or state, use {-bô} to form the adjective meaning "having this quality, being in this state". Other suffixes can also be used with this kind of root.

bâm-zaof newness
bâm-tanlike new
bâm-côage, non-newness

If the root noun denotes a concrete entity or type of entity, {-bô} and {-cô} wouldn't be appropriate, but other modifier derivations are possible:

râm-zapertaining to cats
râm-tanresembling a cat
pwĭm-zaof water
pwĭm-tanlike water



{-rô} specifies an idiomatically selected quality of the root substantive. It's similar to "-um" in Esperanto — not all concrete roots have a defined {rô}-adjective.

If the root denotes an action or relation, certain other suffixes are appropriate.

lju-fwacausing to read
lju-fajreadable, legible
lju-gôworth reading

If the root denotes a mindstate, an adjective formed with {-fwa} describes the circumstances or qualities that conduce to it, and an adjective formed with {-bô} describes the person who experiences it. In gjâ-zym-byn most or all subjective qualities are named by a root mindstate-word plus {-fwa}.

prymappreciation of beauty
prym-bôin awe of something beautiful
prym-côdistaste for ugliness

For more details on derivation of modifiers, see the corresponding sections of the derivational morphology document.


Modifiers as attributes normally follow the nouns they modify.

gâm ny-bô small picture
gâm pân-kwa multicolored picture
gâm prym-fwa beautiful picture

However, with the conjunction {če}, a modifier can be bound to its head in reverse order:

ny-bô če gâm = gâm ny-bô

The same rule goes for modifiers of verbs:

ruŋ-zô ĵwy-bô to move quickly
ruŋ-zô žu-bô to move carefully

With the reversal conjunction,

ĵwy-bô če ruŋ-zô = ruŋ-zô ĵwy-bô

Usually {če} is used as an error correction mechanism.

Here is a sentence with a modified noun as patient and a modified verb:

gâm ny- ĥy-i tâň-zô ƴum-bô -ħa kwǒ.
picture small PAT-at take-V.ACT fraud-ADJ person-ATD4 some
Some scoundrel has taken the small picture in a fraudulent way.


Modifiers as predicates normally occur with the postpositions {ŋĭn-i} (comment) or {jâ-i} (state), with the noun phrase whose referent they're predicated of marked by {mĭ-i} or (more typically) left unmarked in final subject position.

gâm mĭ-i pân-kwa ŋĭn-i.
picture TOP-at all-color CMT-at
The picture is multicolored.
pân-kwa ŋĭn-i gâm.
all-color CMT-at picture
The picture is multicolored.

Note, however, that in predicate position a root or stem signifying a state or quality doesn't have to take the {-bô} suffix, as it does in attributive position.

ny ŋĭn-i gâm .
smallness CMT-at picture DEM3
That picture is small.

(There are archaic sentence in many texts, and in these grammar documents, where a {-bô} adjective is used in predicate position. Updating the grammar document and all its sample sentences to reflect the current state of the language is an ongoing task.)

prym-fwa ŋĭn-i gâm ny-.
aesthetic.pleasure-CAUS CMT-at picture small-ADJ
The small picture is beautiful.

The state transition postpositions {jâ-o} (becoming) and {jâ-ř} (ceasing to be) are used similarly.

hwǒ, pî'râ jâ-o gâm.
INTJ flame state-to picture
Oh no, the picture's caught on fire!
prym-fwa jâ-ř gâm ĥy-pî'râ-.
aesthetic.pleasure-CAUS state-from picture PAT-flame-ADJ
The burned picture ceases to be beautiful.

{ŋĭn-i} and {jâ-i} are interchangeable in many contexts; one rule of thumb is that {ŋĭn-i} is used with more subjective qualities, while {jâ-i} with more objective ones.

ĥĭn-fwa ŋĭn-i tě'bâ.
disgust-CAUS CMT-at tobacco
Tobacco is disgusting.
ĥě'fâŋ-fwa jâ-i tě'bâ.
cancer-CAUS state-at tobacco
Tobacco is carcinogenic.

{jâ-i} also tends to be used for more transitory qualities, and {ŋĭn-i} with more durable ones. Also, if the subject is an animate entity marked by {ʝâr-i} "experiencer", then the predicate modifier is marked by {jâ-i} or the state transition variants thereof, not {ŋĭn-i}.

mwĭl jâ-i gî'lu ʝâr-i.
sleep state-at wolverine experiencer-at
The wolverine is asleep.
mwĭl jâ-ř gî'lu ʝâr-i.
sleep state-from wolverine experiencer-at
The wolverine is waking up.
ĥul jâ-o gî'lu ʝâr-i.
anger state-to wolverine experiencer-at
The wolverine is getting angry.


Comparative forms of modifiers and (especially stative) verbs are formed with the suffixes {-sra} "more" and {-ĵar} "less". Applied to modifiers formed from quality-stems, they come before the {-bô} adjectivizing suffix:

hum depth
hum-bô deep
hum-sra quality of being deeper than something else
hum-sra-bô deeper
hum-ĵar-bô less deep
hum-cô height
hum-cô-bô high, tall
hum-cô-sra quality of being higher/taller than something else
hum-cô-sra-bô higher, taller
hum-cô-ĵar-bô less high, shorter

Applied to other modifiers, formed with suffixes other than {-bô}, the {-sra} and {-ĵar} normally come at the end of the word. For instance, with the causative suffix {-fwa}:

prym appreciation of beauty
prym-fwa beautiful
prym-fwa-sra more beautiful

but this is also possible, though rarer:

prym-sra more intensely/vividly appreciating something's beauty
prym-sra-fwa causing someone to more vividly appreciate something's beauty

With most other modifier suffixes this inversion wouldn't make sense, though:

râm-tan-sra more catlike
žâj-dô-ĵar less sinful
pwĭm-da-sra wetter
ðâ-ja-ĵar less logical
lju-gô-sra more worth reading

When turning one of these comparative modifiers into a stative verb, replace {-bô} with {-van} or add {-van} after the comparative suffix:

hum-sra-van to be deeper
prym-fwa-sra-van to be more beautiful

gjâ-zym-byn doesn't have a morphologically distinct comparative and superlative. If a comparative modifier or verb form occurs with an explicit standard of comparison, it would generally be translated into English as "more/less X" or "X-er"; if it occurs with no standard of comparison, the implicit comparison may be, depending on context, to the same entity in the past, or to some other recently mentioned entity, or to all other things of the kind, or all things of that kind that are in context at the moment. In the latter cases it would be translated as "most/least X" or "X-est". Inexplicit comparison can be in attribute form (a modifier applying to a head noun or verb within a single noun phrase or verb phrase) or predicate form (the modifier being in a separate comment or state postpositional phrase, applying to a head noun that's in a topic or experiencer postpositional phrase):

mwĭl-ŝra-van ĵĭn-sra ʝâr-i.
sleep-tending-V.STATE sibling young-COMP experiencer-at
The youngest of the siblings is sleepy.
sâr-ĵar ŋĭn-i pě'pâ-daj pǒ.
order-COMP.NEG CMT-at paper-mass that
That mass of papers is less orderly [than other such] / is the least orderly.
huw-sra -i -bâm kǒ.
happy-COMP state-at person-new this
This baby is happier [than other [recently mentioned?] babies] / is the happiest.

Or a comparative modifier can be an attribute of a verb, i.e. an adverb:

!ƴâ-ť- ĵwy-sra- mwe.
go.adjustable-2-V.ACT fast-COMP-ADJ IMP
Walk/run faster!

gzb has two ways of relating the head noun of the comparative modifier or the subject of the comparative verb to the standard it's being compared with. One is with the comparative conjunction {θe} (as, than); this is archaic, the comparative conjunction usually being used now only for equality-comparison. In this form the subject of the comparison is linked within its topic postpositional phrase with the standard of comparison:

*hwǒ, lâŋ-sra-van twâ-cu-θuň θe {gormenħast-wam} -i.
INT long-COMP-V.STATE sentence-system-story this than Gormenghast-NAME.P TOP-at
Whoa, this novel is longer than Gormenghast!

Nowadays (since 2008), the standard of comparison is put in a separate postpositional phrase marked with {dî'fu-i}, "compared with". ({dî'fu} derives from the name of the Unix command "diff".)

šun i mîr'su dî'fu-i -sra ŋĭn-i hěm'lu -i.
region this at mulberry compare-at common-COMP CMT-at magnolia TOP-at
Around here magnolias are more common than mulberry trees.

The comparative suffixes can also occur with such quantifier clitics as {reŋ} (many) and modifier clitics as {mje} (in the past):

ƥ dâm-ř twâ-cu reŋ-sra ðĭ-i wuŋ-Ќ-van.
3 authorship-from sentence-system many-COMP relation-at own-1-V.STATE
I own more books by her [than someone else does / than by some other author / than I used to].
ƥ ĥun-i gju-môj mje-ĵar.
3 meet-at speak-V.RECP past-COMP.NEG
I have spoken with him more recently [than someone else has / than with other people recently mentioned].

There are a couple of kinds of qualified comparatives; they occur only as predicates, not as attributes. Subset comparison corresponds to English "one of the most/least...", and uses the subset postposition {muw-i}:

rî'mâ hum--sra- muw-i rî'mâ-źa sĭrz-gam -i.
building deep-OPP2-COMP-ADJ subset-at building-AUG Sears-NAME.G TOP-at
The Sears Tower is among the tallest buildings.

Ranked comparison corresponds to English "the second-most X", "the third-least Y", etc. It is unique in not actually using the suffixes {-sra} and {-ĵar} at all, but an ordinal number in a comment phrase with the compared quality in a generic relational phrase:

Φu ðĭ-i ĉu-pa ŋĭn-i tam-ram -i.
mass relationship-at two-ORD CMT-at Tom-NAME TOP-at
Tom is the second most massive. (= With respect to mass, Tom is second.)
ŋy ðĭ-i se-ĉu-pa ŋĭn-i kentâwrus-wam -i.
distance relationship-at minus-two-ORD CMT-at Centaurus-NAME.P TOP-at
Centaurus is second least distant / second nearest. [In a discussion of stars; Sol of course is the nearest.]

Subset and ranked comparison can be combined to give an explicit standard of comparison, e.g.:

terě-ram lĭw-i fru muw-i ĵu ðĭ-i -pa ŋĭn-i lǒrě-ram -i.
Tara-NAME relation-at child subset-at maturity relation-at three-ORD CMT-at Laura-NAME TOP-at
Laura is the third-oldest of Tara's children.

Explicit comparisons can be made with {dî'fu-i} in relation to a quantified standard, e.g.:

jǒm dî'fu-i ĵlân-sra-van hektor-ram.
person most.of compare-at wise-COMP-V.STATE Hector-NAME
Hector is wiser than most people / than the majority of people.

Explicit superlatives generally use {dî'fu-i}, with a standard quantified with {ble}, "the rest of":

mruň ble dî'fu-i hum--sra-van evřest-wam.
mountain rest.of compare-at deep-OPP2-COMP-V.STATE Everest-NAME.P
Everest is taller than the rest of the mountains / is the tallest mountain.

And a locative or temporal phrase or {muw-i} subset phrase can give a more explicit context within which the superlative is valid:

mâw-źa-cu i mruň ble dî'fu-i hum--sra-van olîmpǒs-mons-wam.
ball-AUG-system DEM1 at mountain rest.of compare-at deep-OPP2-COMP-V.STATE Olympus.Mons-NAME.P
Olympus Mons is the tallest mountain in our solar system.
tâŋ i kĭlm ble dî'fu-i pwĭ-fwa-sra ŋĭn-i ĥun-pĭw kǒ.
personal.era DEM1 at party rest.of compare-at delight-CAUS-COMP CMT-at meet-game DEM1
This gaming party is the most enjoyable party [I've been to] this whole tâŋ.

The comparative suffixes generally don't apply to entity stems. However, they are occasionally used with number or unit-of-measure stems, meaning "more than" or "less than" the amount specified by the stem:

tĭm-sra a set with more than a hundred members
tĭm-sra-bô more than a hundred
gĭ-sra-gla i later than eleven o'clock
kî'grâ-ĵar less than a kilogram

With verbs (whether active or stative) derived from process stems, the comparative suffixes mean "to do said action/undergo said process more/less intensely":

grĭ-ĵar-van šĭm-ĉa kǒ.
function-COMP.NEG-V.STATE algorithm-tool this
This computer functions less well [than other computers / than it used to].
te ĥy-i -sra-ť- mwe šyj-fwa-zô.
Try harder to get it clean.

These comparative action/process verbs can have explicit standards of comparison with {dî'fu-i} or {muw-i} as well.

Some ambiguous comparisons can be disambiguated by using serial postpositions, putting a case postposition before {dî'fu-i} to indicate the role the standard of comparison plays in the sentence (besides being the standard of comparison):

sě'râ dî'fu-i mî'rĭj mĭ-i -sra-van jě'nu ʝâr-i.
Sarah compare-at Mary TOP-at love-CMP-V.STATE Joan EXP-at
Joan loves Mary more than Sarah.

Does she love Mary more than she loves Sarah, or does she love Mary more than Sarah does? Adding {ʝâr-i} or {mĭ-i} between {sě'râ} and {dî'fu-i} disambiguates whether we're comparing two experiencers or two topics:

sě'râ mĭ-i dî'fu-i mî'rĭj mĭ-i -sra-van jě'nu ʝâr-i.
Sarah TOP-at compare-at Mary TOP-at love-CMP-V.STATE Joan EXP-at
Joan loves Mary more than she loves Sarah.
sě'râ ʝâr-i dî'fu-i mî'rĭj mĭ-i -sra-van jě'nu ʝâr-i.
Sarah EXP-at compare-at Mary TOP-at love-CMP-V.STATE Joan EXP-at
Joan loves Mary more than Sarah does.

An example with an active verb:

gwî'jum tu-i dî'fu-i ƥ ŋâw-o gju-ŋy-zô reŋ-sra lě'kjân.
William AGT-at compare-at 3 call-to talk-distant-V.ACT much-CMP Alexander
Alexander calls her on the phone more often than William does.
gwî'jum ŋâw-o dî'fu-i ƥ ŋâw-o gju-ŋy-zô reŋ-sra lě'kjân.
William call-to compare-at 3 call-to talk-distant-V.ACT much-CMP Alexander
Alexander calls her on the phone more often than he calls William.

Equality comparison uses the comparative conjunction {θe} and some sameness or similarity adverb on the stative verb, adjective, or state-postposition:

twâ-cu θe twâ-cu -i lâŋ-van sâm-bô.
sentence-system DEM1 as sentence-system DEM2 TOP-at long-V.STATE same-ADJ
This book is just as long as that book.
blĭn -i ₣um- vlym-srĭw θe vlym-srĭw pǒ.
tightness state-at similar-ADJ clothes-leg DEM1 as clothes-leg DEM3
These pants are about as tight as those.

For more examples of equality comparison, see the section of the derivational morphology document on mathematical equation sentences.

One can also make an explicit comparison with an entity as it was in the past:

pym-fwa-van sâm- tam-ram θe ƥ mje.
amusement-CAUS-V.STATE same-ADJ Tom-NAME as 3 past
Tom is as funny as his past self / is as funny as ever.
ɱ mje dî'fu-i žu--sra -i vĭrě-ram.
3 past compare-at care-tending-COMP state-at Vera-NAME
Vera is more careful than her past self / than she used to be.

Or one could use {ler} "future" or a postpositional phrase specifying a particular time period to be more specific about what other point on an entity's world-line you're comparing its present self to.


The simple personal pronouns are:

ЌI, me
Ł"one", "they" (generic) "someone" (unspecified)
ƥ, ɱhe, she, they (refers to spirits, humans, animals)
te, ŋeit, they (plants, inanimate objects, abstractions)

{ƥ} points backward to a previously mentioned person or group, {ɱ} forward to somone(s) not already mentioned by name. Similarly {te} and {ŋe} point backward and forward to their referents. Technically, {ƥ} and {te} are anaphoric pronouns, while {ɱ} and {ŋe} are cataphoric pronouns.

Note there is no plural first person pronoun. "We" could be expressed by:

Ќ-ť inclusive we: I and you
Ќ-ƥ / Ќ-ɱ exclusive we: I and someone else, I and some others
Ќ-ť-ƥ / Ќ-ť-ɱ very inclusive we: I and you and other(s)

The other pronouns can be marked plural by appending a quantifier clitic, or a number adjective, but in some contexts this is not necessary; ť, ƥ, ɱ can refer to plural antecedents even without such explicit pluralization. Ќ would not be pluralized with a number or quantifier, only in one of the ways mentioned above (unless the speaker were a group mind or hive, perhaps). ("Royal We" would be translated by appending a respectful affix to the first-person pronoun.)

I have not been perfectly consistent about using {ƥ, ɱ} vs. {te, ŋe} for body parts. I am leaning toward consistently using the animate pronouns, but for now the animate and inanimate pronouns are interchangeable for reference to body parts.

{Ł} is glossed above as "one," but it has a broader meaning than "one" in formal English or "on" in French. It can refer to a generic, nonspecific person, in a statement that applies to people in general or a broad class of people, but it can also refer to a specific person whose identity is unknown or irrelevant to the current discussion, similar to the way the passive voice in many Indo-European languages deemphasizes or leaves out the agent. In this sense it could be glossed as "someone."

ƴâw ŝâj-o čĭm ĥy-i --fwa-Ł- źǒ.
dog posses-to chocolate PAT-at digestion-into-CAUS-3.GEN-V.ACT IMP.NEG
One mustn't give chocolate to dogs.
Ќ sim ʝym-daj ĥy-i ĥâ-Ł-.
1 top.part.of hair/leaf-mass PAT-at cut-3.GEN-V.ACT
I got my hair cut / Someone cut my hair.

gzb has no interrogative or relative pronouns per se, but uses the interrogative particle {nǒ}, or the relativizer {lǒ}, with various nouns or noun phrases to serve those purposes.

mâ lǒwho, whom
mâ nǒwho? whom?
gâ lǒwhich
gâ nǒwhat? which?

Almost any noun phrase can be interrogated or relativized, but in practice a few generic nouns tend to get used frequently with these particles. See under Information questionse and Relative clauses for details.

The non-personal pronouns are:

cethis, that; stands for a whole situation described previously
žethis, that; stands for a fact or situation about to be described
rethere, thither, thence; stands for a recently mentioned place

{že} typically points forward to the content of a {hǒŋ} subordinate clause.

že mĭ-i gju-zô ƥ tu-i, hǒŋ gjâ mĭ-i
this TOP-at speak-V.ACT 3 AGT-at that language DEM1 TOP-at
syj-faj heŋ źe ŋĭn-i.
use-able not very CMT-at.

She talked about how useless my language was.

ce mĭ-i sjum-van, wǒj Ќ ĥy-i hyw-fwa-zô ce gân-ř
this TOP-at thankful-V.STATE because 1 PAT-at know-CAUS-V.ACT this cause-from
luŋ mĭ-i.
detachment TOP-at

I was thankful for that, because it taught me something about detachment.

{re} is a place pronoun, similar to "y" in French. It is normally followed by a spacetime postposition to express its role in the sentence, just like a noun phrase referring to a place.

âthĭnz-wam o ruŋ-źa- pî'tĭr pe zě'ĥâr.
Athens-NAME.P to go-AUG-V.ACT Peter and Zachary
re i tyn-van bĭr-ĵwa reŋ.
there at place-V.STATE beer-place many
Peter and Zachary traveled to Athens, Georgia. There are a lot of bars there.

However, {re} can be the object of an abstract case postposition as well.

tâŋ reŋ ði sânfrânsĭsko-wam-la o ruŋ-źa-.
personal.era many ago San.Francisco-NAME.P-ATD1 to go-AUG-V.ACT
re kâ-i prym-źa-van, mǒn pwĭm-by-da ŋĭn-i re.
there ATT-at aesthetic.appreciation-AUG-V.STATE although water-air-full CMT-at there
Many years ago I traveled to San Francisco. I found it very beautiful, although it was foggy.

In the second clause, {re} is the object of attention, and in the third, {re} is the topic (the topic postposition being left out in sentence-final position).

One can add modifiers to pronouns, indicating number, gender, age, etc., if necessary to clarify which of several previously-mentioned entities is intended.

pen tu-i ruŋ-zô mwe mruň on.
you all AGT-at go-V.ACT IMP mountain to

Y'all go to the mountain.

vlym bâm-bô ĥy-i kâj-zô ƥ-ŝy srǒ tu-o ƥ-mym ŝâj-o.
clothing new-ADJ PAT-at exchange-V.ACT 3-female several AGT-to 3-self possession-to.

They buy new clothes.


{gjâ-zym-byn} has several kinds of conjunctions.

One shows the relative truth or falsity of two independent clauses. Words of this type are derived by compounding phonemes from a truth table:

first clausesecond clauselogic function
T T ŝ = TT, k = TF, p = FT, f = FF
F T w = T, (null) = F
F F oň = T, iň = F

So for instance:

kiňand - TFFF
ŝwiňor (inclusive) - TTTF
pwiňor (exclusive), unless - FTTF
foňneither/nor - FFFT
koňequivalence; if and only if — both are true or both false. TFFT

(In practice, these are rarely used, except for {kiň}, and I don't think I've ever used any of the 11 other conjunctions one could theoretically form from this table.)

Another kind links two clauses and shows their causal relation (or surprising lack thereof); like "because, therefore, however, but" in English. They're also formed with a matrix.

Prefix elements:
ŝ-logical cause(therefore, because)
w-effective cause(therefore, because)
ʝ-evidence, inference(therefore, because)
m-not hindered(however, even though, in spite of, but)

Suffix elements:
-ǒn1st clause, therefore (however) 2nd clause
-ǒj1st clause, because (even though) 2nd clause

ĉu pe ðy θe fy mĭ-i sâm-van, ŝǒn
2 plus 5 = 7 TOP-at same-V.STATE therefore.logically
fy se ðy θe ĉu mĭ-i sâm-van.
7 minus 5 = 2 TOP-at same-V.STATE.

2 + 5 = 7, therefore 7 minus 5 = 2.

lju-sô ŋĭn-i, wǒn kâj-zô twâ-cu-vuj reŋ ĥy-i. CMT-at, transact-V.ACT sentence-system-concrete many PAT-at

I am readful, so I buy many books.

{ʝǒj} and {ʝǒn} show a relationship between two clauses where the truth of one clause is inferred or deduced from the more obviously evident truth of the other.

tyn ř ruŋ-ƥ- gwe, ʝǒj
place this from go-3-V.ACT already because.inference
ƥ wuŋ-i ƴâ-ĉa -i rĭm-Ќ-van heŋ.
3 owning-at go-tool ATT-at see-1-V.STATE not
He must have already gone, because I don't see his car.


zym-zô, ʝǒn bĭŋ-van.
think-V.ACT therefore.inference exist-V.STATE

I think, therefore (I deduce that) I am.

The {-ǒj} forms reverse the causal order:

râm mĭ-i pwĭ-cô-van, wǒj ƥ mĭ-i pwĭm-da ŋĭn-i.
cat TOP-at delight-OPP2-V.STATE, because.fact 3 TOP-at water-full CMT-at

The cat is miserable because it's wet.

{mǒn, mǒj} correspond to "although" and "but"; there is a mirative element in one of the two clauses joined by these, the truth of one clause being unexpected or surprising in light of the truth of the other clause:

helenike-lam mĭ-i suŋ-van heŋ, mǒn kun-hôw-ca mje te kâ-i.
Greek-NAME.L TOP-at not, although know-CAUS2-V.REFL PAST 3 ATT-at

I'm not fluent in Greek though I studied it awhile ago.


vlym-srĭw ĥy-i šyj- žu-bô, mǒj te im ver ħĭwm gǒ.
clothing-legs PAT-at clean-V.ACT care-ADJ but 3.INAN part.of still stain behold
I washed the pants carefully, but look, the stain is still there.
The conjunctions {mǒn} and {mǒj} are sometimes used in combination:
mǒn sru-van ť ty-o ruŋ-zô,
although desire-V.STATE 2 home-to go-V.ACT

mǒj Ќ ŝâj-i ƴâ-ĉa -i grĭ--van.
but 1 possession-at move-tool TOP-at function-OPP2-V.STATE
Although I would like to come to your house, my car is broken.

A third type of conjunction primarily works with numbers to show arithmetic operations. Some of them are used by analogy with other words and phrases.

peplus; additive 'and'
kemultiplied by; synergetic 'and'
seminus; 'except'
ðedivided by, per; contrasted with
meraised to the power of
zej range operator: "X zej Y", the set of numbers from X to Y inclusive

Examples of the math/phrasal conjunctions:

ɱ-ŝy pe Ќ tu-i vâ-oŋ-zô.
3-female and 1 AGT-at digestion-into-V.ACT
She and I ate together.

{pe} can be used to link verbs within a single clause which have the same subject and object.

mě'hu kwǒ ĥy-i jâln-fwa- pe --zô.
stew some PAT-at heat-CAUS-V.ACT and digestion-into-V.ACT
I heated up and ate some stew.
Ќ lĭw-i -ma-bâm -i -van pe pym-van.
1 relationship-at sibling-meta-new TOP-at love-V.STATE and amusement-V.STATE
I love my baby cousin and find [her] amusing.

In these cases the two verbs not only have the same subject but have the same kind of relationship to their objects: patient in the first case, focus or topic in the second. But between verbs with different subjects, or with the same subject but different relationships to their object, the clausal conjunction {kiň} must be used to translate "and".

θuň-bâm krĭ-o lju-θaj- kiň te ĥy-i byn-zô.
story-new create-at read-OPP2-V.ACT and 3.INAN ATT-at tinker-V.ACT
I am writing a new story and revising it.

In English one might say "I am writing and revising a new story", "a new story" being the direct object of both verbs; but gzb has no "direct objects" as such, just patients, objects-of-result, objects-of-attention, and so forth. Here each verb has a different relationship to its object and requires a different postposition to mark it, so the pronoun {te} is used resumptively and {ĥy-i} shows that {byn-zô} is modifying a now existing story, not creating a new one like {lju-θaj-zô}.

Besides its core meaning of "multiplied by", {ke} is used between non-mathematical nouns to mean "and", but implying a closer, synergistic connection between the nouns or noun phrases linked by it than {pe}.

ɱ-ĉu mĭ-i rě'ĵy ke rě'ĵy-θaj jâ-i.
3-two TOP-at wife and wife-OPP1 state-at

They are wife and husband.

{ke} can also be used to link proper names of husband and wife, co-authors, or collaborators; in this case the names are compounded into one word with -ke- as a kind of hyphen, and the name suffix {-ram} or {-šam} is usually only used once, not after each name.

Φě'ĥu-cu -i krĭ-gâm- tam-ke-ser'ě-ram.
elephant-system TOP-at create-picture-V.ACT Tom-and-Sarah-NAME
Tom and Sarah painted a picture of a herd of elephants.

{se} "minus" is incorporated into number-words to form negative numbers, and used phrasally to express subtraction. With non-numbers, it means "except", "with the exception of":

dejvĭd-ram mĭĉěl-šam dâm-ř θuň-lâŋ pen se
David-NAME Mitchell-NAME.F author-from story-long all except
bâm-sra kâ-i lju-zô gwe.
new-COMP ATT-at read-V.ACT already
I've read all of David Mitchell's novels except his newest.

{ðe} "divided by" is used with number-words to derive words/phrases for fractions. With other kinds of words, it means "contrasted with":

bě'lâm ðe mwĭň mĭ-i dlâw-Ł-van dî'fu-zô.
embarrassment.privacy.violation CONTRAST embarrassment.subject TOP-at duty-3.GEN-V.STATE distinguish-V.ACT
One must distinguish embarrassment at violated privacy from embarrassment at talking about emotionally sensitive subjects.

{me} "raised to the power of" is incorporated into kinship terms to indicate more distant relationships. It's also used with units of measure and dimensional terms to indicate higher dimensions, e.g. to derive volume from a basic word for distance.

kyn parent
kyn-ma grandparent
kyn-me-dâ great-grandparent
ŋy distance
ŋy-ma area
ŋy-me-dâ volume
mě'tyr meter
mě'tyr-ma square meter
mě'tyr-me-dâ cubic meter

{se} and {ðe} are most commonly used in the corpus with numbers ({se} is most commonly used in the phrase {gě'dĭm se-cĭ-pa}, day minus-one'th = yesterday), while {pe} and {ke} are far more commonly used with ordinary stems. This is probably because simply concatenating numbers in compounds multiplies or adds them, according to their increasing or decreasing magnitude.

The range conjunction {zej} has primarily been used with number words, signifying a range of numbers from one to another inclusive:

ðy zej fy five to seven, i.e. the integers 5, 6 and 7 or the real numbers from 5.0 to 7.0 depending on context
ðy-pa zej fy-pa the fifth through seventh items in a sequence
ðy-gla zej fy-gla from five a.m. to seven a.m. in the morning


tĭm-ĉu-fy zej tĭm-ĉu-ĝu im
hundred-two-seven RANGE hundred-two-thirteen part.of
tyn-van heŋ ŋî'bĭ gĭ-rô.
be.located-V.STATE not number eleven-QUAL
There are no prime numbers from 114 to 126.
θuň-lâŋ-kwĭ im ĉu-pa zej ðy-pa kâ-i lju-zô.
story-long-series DEM3 part.of two-ORD RANGE five-ORD ATT-at read-V.ACT
I've read the second through fifth novels in that series.
-gla zej ħy-gla i gjâ ĥy-i byn-zô.
seventeen-ORD.T RANGE nineteen-ORD.T at language DEM1 PAT-at tinker-V.ACT
I tinkered with this language from 5pm to 7pm.

It can also signify an approximate, uncertain number that's somewhere in a certain range, depending on context.

twâ-cu gâr-tĭm- zej ĉu-gâr-tĭm- ðĭ-i wuŋ-van.
sentence-system ten-hundred-ADJ RANGE two-ten-hundred-ADJ relationship-at own-V.STATE
I own around a thousand to two thousand books.

I've recently (May 2010) discovered that {zej} can be used with any pair of modifiers in the same semantic domain, not just with numbers. It can express a transition from one state or quality to another within a common continuum, or that various members of the set denoted by the head noun have values for the meta-quality in a given range; for instance:

gě'dĭm-zla il kâj-ha-fja zej kâj-ha-da jâ-i lî'zě'bĭ.
day-whole.set throughout exchange-stuff-MIN RANGE exchange-stuff-full STATE-at Elizabeth
During her life Elizabeth ranged from poor to rich.
pjylm-pwĭm-daj i vlym-fja zej vlym-ta jâ-i .
border-water-mass DEM3 at clothing-MIN RANGE clothing-without STATE-at person
On that beach, the people range from scantily dressed to naked.


A fourth kind of conjunction are individual words to fill miscellaneous needs.

hǒŋ introduces object subordinate clauses: 'I think that...', 'He asked whether...'
ðǒŋ introduces subject subordinate clauses: 'It's obvious that...', 'It's uncertain whether...'
Φǒintroduces quotations: 'He said "..."'
θecomparative (as, than)
šej or; also known as; that is to say
rej exclusive or (phrasal, contrast with clausal {pwiň})
hej inclusive or (phrasal, contrast with clausal {ŝwiň})
ĉǒ if (conditions beyond speaker's control)
if (conditions within speaker's control)
dlen else, otherwise
ĉe concatenation conjunction

Examples of some of the subordinate clause conjunctions:

twâ-zô hǒŋ ruŋ-zô ƥ.
say-V.ACT that come-V.ACT 3

I said that he's coming.

twâ-zô Φǒ {ruŋ-zô ƥ.}
say-V.ACT QUOTE come-V.ACT 3

I said "He's coming."

See also the main section on subordinate clauses.

The conjunction {šej} can sometimes be glossed as "also known as"; it links two different names or descriptions of the same entity, or a main title and a subtitle.

?lju-ť- gwe zǒn arue-šam šej volteŕ-ram
read-2-V.ACT already Q.YN Arouet-NAME.F AKA Voltaire-NAME
dâm-ř θuň -i.
authorship-from story ATT-at
Have you read any stories by Arouet, also known as Voltaire?
ɱ lĭw-i lĭm šuŋ--sra šej sij'sĭl-ram ty-o
3 relation-at friend new-OPP2-COMP AKA Cecil-NAME home-to
ruŋ- vjurm- sĭr'ĭl-ram.
go-V.ACT visit-V.ACT Cyril-NAME
Cyril went to visit his oldest friend Cecil.

It can also link two clauses that describe the same event or situation in different ways.

vlym tyn-ca šej Ќ ĉon vlym ĥy-i tyn-zô.
clothing into place-V.REFL AKA 1 around clothing PAT-at place-V.ACT
I put myself into clothes, or surround myself with clothes. [= I get dressed.]

{ĉǒ}, {bǒ} and {dlen} are discussed under Conditional clauses.

{ĉe} is the concatenation conjunction, used between number words to build up identifier strings (e.g., phone numbers or other arbitrary identifiers with little or no intended mathematical signififance).

ðy-dâ ĉe fy ĉe ĉe ĉu-ĉu ŋĭn-i mĭlm-kě'ĝu.
five-three CAT seven CAT three CAT two-two CMT-at identifier-hidden
The PIN number is 8734.

Various additional conjunctions can be derived from various stems with the suffix {-šar}:

huw-fwa-ƥ-ca ru-šar ŝrun -i ku-van.
happy-CAUS-3-V.REFL way-CONJ music ATT-at hear-V.STATE
She cheers herself up by listening to music.
suŋ-van pǒl-ram ru-šar pjân- prym-fwa. Paul-NAME way-CONJ piano-V.ACT aesthetic.appreciation-CAUS
Paul is skilled at playing the piano beautifully.
gju-ƥ- -šar mâl--van ķe ħĭ -i.
talk-3-V.ACT topic-CONJ precise-OPP2-V.STATE too.much experiment TOP-at
He talked about how the experiment was too imprecise.
gâm-ĵwa o ruŋ- kujm-šar aňsor-šam dâm-ř
picture-place to go-V.ACT purpose-CONJ Ensor-NAME.F authorship-from
gâm-daj -i rĭm-van.
picture-collection ATT-at see-V.STATE
I went to the art museum to see the Ensor exhibit.

Note that when a {kujm-šar} clause is negated, it is with the {źǒ} imperative negative (see under "Epistemic particles"):

mâ-bâm ĉin vlym-θĭm vâlm-cô-bô ĥy-i
person-new around clothing-thigh permanent-OPP2-ADJ PAT-at
ŋwĭm-ť- mwe kujm-šar ħâň-ƥ- ver źǒ.
replace-2-V.ACT IMP purpose-CONJ yell-3-V.ACT still IMP.NEG
Change the baby's diaper so they will stop yelling.

Note the distinction in syntax between postpositions and conjunctions formed from the same root:

mluŋ ŋwĭm-i rî'zĭ ĥy-i --zô.
bread substitute-at rice PAT-at digestion-into-V.ACT
I eat rice instead of bread.
ty tyn-vĭj-van kiň lju-zô, ŋwĭm-šar
home inside place-while-V.STATE and read-V.ACT substitute-CONJ
ĥun-pĭw o ruŋ-zô.
meeting-game to go-V.ACT
I stayed home and read instead of going to the gaming party.

See also the discussion of {-šar} in the derivational morphology document.

Correlatives, demonstratives, interrogative and relative pronouns

Question-words and relative pronouns are formed by attaching a clitic to any root word. For instance,

what?, which?
mâ-nǒwho? [which person?]
nu-nǒwhen? [at which moment?]
vĭj-nǒwhen? [during what period?]
ru-nǒhow? [in what manner?]

{nǒ} can be embedded into a verb.


What am I missing?

?ryň--zôť tu-i.
action-Q.WH-V.ACT 2 AGT-at

What are you doing?

Relative pronouns are formed in a similar way with the clitic {lǒ}. (Note that gjâ-zym-byn uses distinct relative and interrogative pronouns. English, French, and Esperanto all make do with a single series of pronouns for both relatives and interrogatives in wh-, qu-, and ki- respectively.)

which, that
mâ-lǒwho, that

The demonstrative clitics (this, that...) are similar in form.

this near me
that near you
that far from us

Terms like "here", "there", "now", "then", and so forth are formed by applying these clitics to various root words; usually a postposition is required as well.

tyn kǒ ihere (at this place)
šun pǒ ithere (in that region)
tyn tǒ othither (toward you)

The distinction between {kǒ} and {tǒ} can be interesting with respect to time:

nu kǒ inow (at this moment when I am speaking/writing this)
vĭj tǒ inow (during the period when you are hearing/reading this)
vĭj i grâm ĥy-i pî'râ-zô mwe.
time DEM2 during message DEM1 PAT-at fire-V.ACT IMP

Burn this message now [when you read it].

Terms like "anyone, everyone, no one", etc., are formed by following ordinary root words with various quantifier particles or compounding with number root words.

mâ peneveryone
mâ-bâno one
mâ kwǒsomebody, anyone
tyn peneverywhere


gzb does not mark definiteness pervasively with articles or inflections as do English, French, Basque, and some other languages. Nouns are by default unmarked for definiteness, though context will usually clarify whether any instance of a class or a particular member is meant. When context is insufficient, gzb has several particles that can be used to mark definiteness or indefiniteness, including the demonstrative particles and some of the quantifiers.

rî'mâ rjâ-i.
house quest-at
I'm looking for a/some/the house. [ambiguous in itself, context may clarify]
rî'mâ kwǒ rjâ-i.
house some quest-at
I'm looking for a house.

Any house will do, perhaps; or the speaker is vague about what house or kind of house he's looking for.

rî'mâ rjâ-i.
house certain quest-at
I'm looking for a certain house.

Here the speaker knows the specific house he's looking for but isn't sure if the listener does.

rî'mâ rjâ-i.
house that.3 quest-at
I'm looking for the/that house.

Here the speaker and listener have already agreed on what house they're talking about.

rî'mâ ʝel -i zym-zô.
house in.general TOP-at think-V.ACT
I'm thinking about houses in general. [about the category of houses]


Yes/no questions

Yes/no factual questions are formed by following the main verb with {zǒn} (roughly equivalent to "ĉu" in Esperanto, "-li" in Volapük, "-kah" in Malaysian, etc.). The questioned verb is often though not always fronted.

?ruŋ-zô zǒn ť tu-i pjylm-pwĭm-daj o.
go-V.ACT Q.YN 2 AGT-at border-water-mass to

Are you going to the beach?

Ordinarily the locative complement would come first, but questioning the verb overrides this and shunts it to another position.

Questions with pairs of alternatives are formed using one of the "or" conjunctions (ŝwiň, pwiň, hej, rej) and placing the question particle {zǒn} just after the last of the questioned alternatives.

?mjyl rej čĭm zǒn kâ-i jyn-sra-van ť ʝâr-i.
honey or chocolate Q.YN ATT-at pleasure-COMP-V.STATE 2 EXP-at

Do you get more pleasure from honey or chocolate?

Questions expecting a "yes" answer may insert {vǒm} (yes, indeed, certainly) between the main verb and {zǒn}; similarly with questions expecting "no" and the negative particle {heŋ}:

?mjyl kâ-i jyn-van vǒm zǒn ť ʝâr-i.
honey ATT-at pleasure-V.STATE yes Q.YN 2 EXP-at

You like honey, don't you?

?fĭm-cô-van heŋ zǒn ƥ ʝâr-i.
health-OPP2-V.STATE not Q.YN 3 EXP-at

She isn't sick, is she?

All of the aforementioned kinds of questions are usually answered with {vǒm} (yes) or {heŋ} (no). However, {mwe} (necessary, imperative) and {źǒ} (negative imperative) could be used as emphatic forms in response to factual questions.

Questions asking for permission or advice either place the question particle {zǒn} after an auxiliary verb,

?dlu-van zǒn ruŋ-zô tyn ř.
right-V.STATE Q.YN go-V.ACT place DEM1 from
May I be excused?

...or use {mwe zǒn} or {źǒ zǒn} after the main verb,

?vlym čâ-ja ĥy-i šâ-zô mwe zǒn.
clothing swim-suitable.for PAT-at carry-V.ACT IMP Q.YN
Should I bring a swimsuit?
?Φě'ĥu ĥy-i čĭn-Ł- źǒ zǒn.
elephant PAT-at poke-3.GEN-V.ACT IMP.NEG Q.YN
One shouldn't poke the elephant, should one?


Questions asking someone for agreement with a proposed plan use the question particle {srem}.

lĭw-o rě'ĵy jâ-o srem ť.
1 relation-to wife state-to Q.YN.PLAN 2
Will you marry me?

(Note that, as there is no verb in the above sentence, the question particle follows the chief postpositional phrase.)

[Translation of xkcd #386

?tĭw-mwĭl son srem ť-la.
chair-sleep onto Q.YN.PLAN 2-ATD1
Are you coming to bed?
ðu-Ќ-van heŋ. *hum-ga-van ce.
able-1-V.STATE not deep-METAPH-V.STATE behold this
I can't. This is important.
*sĭŋ-flu-kô i twâ-zô ĵlân- .
information-flow-place at say-V.ACT wisdom-violation person certain
Someone is wrong on the Internet.

Information questions

Information questions or question-word questions (who, what, where, etc.) are formed with the particle {nǒ} following a general word for the kind of entity whose nature is questioned, and followed by a postposition for the role said unknown entity plays in the sentence; the questionized postpositional phrase almost always comes first, even if it would normally follow other consitutents in an indicative sentence.

? tu-i kĭlm o ruŋ-zô.
person Q.WH AGT-at party to go-V.ACT
Who all is coming to the party?

(Normal order of an indicative sentence would be locative, verb, agent.)

? mĭ-i -ť-van.
person Q.WH TOP-at love-2-V.STATE
Whom do you love?
? ʝâr-i ĝyl-fyn kâ-i rĭm-van.
person Q.WH EXP-at interruption-driving ATT-at see-V.STATE
Who saw the accident?
?tyn ř ruŋ-zô -ŋa .
place Q.WH from come-V.ACT thing-ATD5 this
Where did this thing come from?
?ru i dĭn-ť- .
manner Q.WH at knot-2-V.ACT that.2
How do you/did you tie that knot?
?gân ř tě'θru-van kwě'kyr.
cause Q.WH from collapse-V.STATE oak
Why (from what cause) did the oak fall?
?kujm o twâ-cu lju-θaj-ť-.
goal Q.WH to sentence-system that.2 write-OPP1-2-V.ACT
Why (for what purpose) are you writing that work?

Note the way the cause and purpose question-phrases relate to the corresponding postpositions {gân-ř} "because" and {kujm-o} "for the purpose of", derived from the same roots.

? syj-i te ĥy-i ķĭn-ƥ-.
thing Q.WH use-at 3.INAN PAT-at build-3-V.ACT
What did he use to build it?
?ŋĭw pě'ŝlĭ-i jyn--ť-van.
body.part Q.WH focal.part-at pleasure-OPP2-2-V.STATE
What body part hurts?
?twâ-cu ĉul-i hwâwm-ƥ-zô.
sentence-system Q.WH perform-at roleplaying-3-V.ACT
What play are they performing?

Some words take on a broader sense when used with the question-clitic {nǒ} than they have in other contexts. For instance, {mâ} refers to human persons, but {mâ nǒ} can question the identity of any animate agent more generally; the speaker in the first three questions above isn't necessarily assuming that the answer will refer to a human, though it probably will. The most generic WH-question phrase is {?gâ nǒ}, "thing which?"; but in most contexts a more specific word would be selected; e.g.,

? -i lju-ť-zô.
thing Q.WH ATT-at read-2-V.ACT
What thing are you reading?

is less likely than

?twâ-cu -i lju-ť-zô.
sentence-system Q.WH ATT-at read-2-V.ACT
What written work are you reading?

{gâ nǒ} is even a hypernym of {mâ nǒ}, in contexts where the questioner is unsure whether the entity he's asking about is even animate much less human.

The indefinite particle {kwǒ} is used along with {nǒ} to indicate bewilderment or consternation:

?ru kwǒ i -van ce.
manner Q.WH some/any at event-V.STATE this
How the heck did this happen?
?kujm kwǒ o žâ-ť-van.
purpose Q.WH some/any to wait-2-V.STATE
Why are you waiting? What are you waiting for?


{zǒn} and more rarely {srem} can question other elements of a sentence besides the verb:

?mě'zâ ĥy-i zǒn vâ-oŋ-zô ť.
corn PAT-at Q.YN digestion-into-V.ACT 2

Is that corn you're eating?

Note the difference from {nǒ}, and the different placement of these particles relative to the postposition:

?mě'zâ ĥy-i vâ-oŋ-zô ť.
corn Q.WH PAT-at digestion-into-V.ACT 2

Which corn [or, what kind of corn] are you eating?

Subordinate clauses

Subordinate clauses are introduced with {ðǒŋ} or {hǒŋ} (like "that" or "whether" in English), or one of the causal (because/therefore) conjunctions ({wǒn}, {wǒj}, etc.).

{hǒŋ} is used to introduce subordinate clauses which are the object of the main clause.

?kun-van zǒn ť tu-i, hǒŋ tyn o ruŋ-zô ƥ tu-i.
know-V.STATE Q.YN 2 AGT-at that place DEM1 to go-V.ACT 3 AGT-at

Did you know that he's coming?

Here, the subordinate clause introduced by {hǒŋ} is the object of the main verb {kun-van} "know".

kun-van heŋ, hǒŋ tyn o ruŋ-zô zǒn ƥ tu-i.
know-V.STATE not, that place DEM1 to go-V.ACT Q.YN 3 AGT-at

I don't know whether she's coming.

Note the use of the question particle {zǒn} within the subordinate clause; this makes {hǒŋ} mean "whether" instead of "that". It works much the same with {srem} as well:

ƥ-ŝy ŋâw-o frâ- ƥ- tu-i, hǒŋ
3-female call-to ask-V.ACT 3-male AGT-at that
ƥ- lĭw-o rě'ĵy -o srem ƥ.
3-male relation-to wife state-to Q.YN.PLAN 2

He asked her whether she would marry him.

The default subject in a {hǒŋ}-subordinate clause is the subject of the main clause:

twâ-zô tam-ram tu-i, hǒŋ mwĭl-ŝra-van, wǒn ty o ruŋ-zô.
say-V.ACT Tom-NAME AGT-at that therefore home to go-V.ACT

Tom said he was sleepy, so he was going home.

{ðǒŋ} is used to introduce subordinate clauses which are the subject (typically the topic) of the main clause. A {ðǒŋ} construction is equivalent to a similar construction with {hǒŋ} having the forward-reference pronoun "že" in the main clause as a dummy subject (like "it" in similar senteces in English).

že -i huw-fwa ŋĭn-i, hǒŋ Ќ ty-o ruŋ- ler ť tu-i.
this TOP-at happy-CAUS CMT-at that 1 home-to go-V.ACT FUT 2 AGT-at

is equivalent to:

huw-fwa ŋĭn-i, ðǒŋ Ќ ty-o ruŋ- ler ť tu-i.
happy-CAUS CMT-at that 1 home-to go-V.ACT FUT 2 AGT-at
It makes [me] happy that you are coming to my home.

Relative clauses

In a relative clause, the relativizer {lǒ} is placed after a noun phrase referring to the relativized element in the main clause, and is followed by a spacetime or case postposition as appropriate for its role in the relative clause. The relativized element comes first in the relative clause, whatever its role. Any element in the main clause can be relativized, as well as non-overt elements (e.g., time or place where a main clause doesn't have an overtly expressed temporal or locative complement). And the relativized element can have any role in the relative clause.

Relative clauses come after the main clause; they do not embed in it like in English.

The noun phrase that is relativized doesn't have to be fully repeated in the relative clause; the noun phrase preceding {lǒ} just has recognizably refer to a unique element of the main clause. (E.g., usually {mâ} is used when the relativized element refers to the only person mentioned in the matrix clause, but if more than one person was mentioned, some other noun (probably a {mâ-} based compound) is used instead.)

Relativized element is object of main clause and subject of the relative clause:

lî'zě'bĭ ĥun-i ŝě'ĥâ-môj, tu-i
Elizabeth meeting-at chess-V.RECP person REL AGT-at
gě'dĭm-zla hǒl θje il pĭw- de.
sleep.wake.cycle-COLL whole almost through play-V.ACT DEM1 HAB
I played chess with Elizabeth, who had been playing almost her whole life.

Relativized element is locative complement of main clause and object of relative clause:

mwĭl-kô-ĵwa hân- i drulm i tyn-nî'šĭm-van,
sleep-place-place old-ADJ DEM3 at short.period certain at
tyn ĥy-i ħulŋ-źa-Ł- θǒ.
place REL PAT-at damage-AUG-3.GEN-V.ACT just
I once stayed in that old hotel they just demolished.

Relativized element is not overtly present in main clause:

kun-van heŋ, tyn i ty-ƥ-van.
know-V.STATE not place REL at home-3-V.STATE
I don't know where he lives.
Note that the equivalents of some of these relative clauses would not be considered relative clauses in English, but maybe complement clauses introduced by "when", "while", "where" etc.
sě'râ kâ-i ĥun- vĭj i žĭlm rjâ-i kâj-kô o ruŋ-zô.
Sarah-NAME ATT-at meet-V.ACT time REL at butter quest-at exchange-place to go-V.ACT
I met Sarah when I went to the store for butter.

If you have multiple relative clauses in a sentence, each comes at the end of the clause whose element it relativizes.

ĉě'θâ i mâ-vĭ ĥy-i zuň--fwa- ĝâj-ja
day DEM1 at person-male PAT-at life-OPP2-CAUS-V.ACT
tyrn tu-i, tu-i ƥ lĭw-i rě'ĵy ĥy-i
government AGT-at person REL AGT-at 3 relationship-at wife PAT-at
zuň--fwa-, mâ-ŝy tu-i hĭj-rě'ĵy--.
life-OPP2-CAUS-V.ACT person-female REL AGT-at sacrament-wife-violation-V.ACT
The man who murdered his wife who cheated on him was executed today.

If two elements in the main clause are relativized, they can come in either order after the main clause.

sě'râ kâ-i ĥun- vĭj i žĭlm rjâ-i kâj-kô o ruŋ-zô
Sarah-NAME ATT-at meet-V.ACT time REL at butter quest-at exchange-place to go-V.ACT
tu-i lě'jâ rjâ-i rě'ju-kâj-zô.
person REL AGT-at olive quest-to search-exchange-V.ACT
I met Sarah, who was shopping for olives, when I went to the store for butter.

Note that because a relative clause starts with a noun phrase followed by {lǒ}, not with a conjunction like subordinate clauses, you can't omit a subject case postposition after an overt subject in the main clause.

mâ-zla-pôm rě'ĵy rjâ-i sru-van vǒm
person-COLL-EVD wife quest-at desire-V.STATE indeed
--ĵu rě'ĵy-ta mĭ-i, ʝâr-i
person-male-mature wife-without TOP-at person REL EXP-at
kâj-ha-daj ĝĭ- ðĭ-i wuŋ-van.
exchange-stuff-mass big-ADJ relationship-at own-V.ACT
It is a truth universally acknowledged, that a single man in possession of a good fortune, must be in want of a wife.

The main clause can have a cataphoric {ɱ} or {ŋe} pronoun pointing forward to the relativized element of the relative clause.

jyn- jâ-ř rjâ-i ŋe ŝĭw-i sěl'kâ-gôm ĥy-i ĉârn-ca,
pleasure-OPP state-from quest-at 3.INAN substance-at spine-METONYM PAT-at rub-V.REFL
ŝĭw ĥy-i bîn'zĭ-ĵwa i kâj-zô kyn-ŝâm-la.
substance REL PAT-at gasoline-place at exchange-V.ACT parent-womb-ATD1
I rubbed my back with this stuff Mom bought at a gas station to stop the pain.
ɱ ĥy-i ruŋ-ť- zuň--fwa- mwe, rjuŋ tu-i
3 PAT-at go-2-V.ACT alive-OPP2-CAUS-V.ACT IMP dragon REL AGT-at
Ќ-ƥ wuŋ-i bě'rîn'ku pe byj ĥy-i vâ-oŋ-zô de.
1-3 own-at sheep and cattle PAT-at digestion-into-V.ACT HAB
Go and kill the dragon who's been eating our sheep and cattle.

Conditional clauses

The conjunctions {ĉǒ} and {bǒ} can both be glossed as "if", and they introduce the protasis of a conditional statement. The apodosis can be introduced by any of the "therefore" conjunctions {ŝǒn, ʝǒn, wǒn} as appropriate — most often {wǒn}.

{bǒ} is used for conditions within the speaker's control, referring to uncertain plans and future actions. E.g.,

jork-wam-bâm o ruŋ-źa-zô, wǒn
if York-NAME.P-new to go-AUG-V.ACT then
ħĭn-ta-tǒj slân-i Φâ kâ-i rĭm-van.
restriction-without-NMZ symbol-at statue ATT-at see-V.STATE
If I go to New York, I'll see the Statue of Liberty.

{ĉǒ} is used for conditions beyond the speaker's control, whether unknown past events, hypothetical future events, or counterfactuals.

ĉǒ bly-van pwĭm, wǒn re o ruŋ-zô heŋ gwî'jum.
if fall-V.STATE water then 3.PLACE to go-V.ACT not William
If it rains, William won't go [to the aforementioned place].

The particle {šelm} is used after the verb or the main postpositional phrase in the apodosis, if (a) the protasis is counterfactual, or (b) the protasis is implied and omitted.

ĉǒ čĭm ŋĭn-i pwĭm, wǒn ĝâj-van šelm ĥâ-ŋĭw-zla.
if chocolate CMT-at water then decay-V.STATE HYPOTH cut-organ-COLL
If water were chocolate, all of the teeth would decay.
pwĭm-pul-daj pĭw-zô vlym-ta,
if water-powder-COLL inside play-V.ACT clothes-without
wǒn fĭm-cô jâ-o šelm bî'duň pě'ŝlĭ-i.
then health-OPP2 state-to HYPOTH sinuses focus-at
If I were to play naked in the snow (which I won't do), I'd catch a cold.

The latter use often occurs in a reply to someone's proposal; e.g., if Tim says:

ty-i kĭlm-zô srem.
1 home-at party-V.ACT Q.YN.PLAN
Shall we have a party at my house?

then Manuel might reply:

ce mĭ-i pwĭ-van šelm.
this TOP-at delight-V.ACT HYPOTH
I would be delighted [if we did].

This apodosis with implied protasis is also sometimes used following an imperative, especially a negative imperative, giving a reason for the command or request.

pwĭm-pul pĭw-ť-zô vlym-ta źǒ.
water-powder-COLL inside play-2-V.ACT clothes-without NEG.IMP
fĭm-cô jâ-o šelm.
health-OPP2 state-to HYPOTH
Don't play naked in the snow. You would get sick.
mě'hu ĥy-i vâ-oŋ-ť-zô mwe.
stew behold PAT-at digestion-into-2-V.ACT IMP
te kâ-i jyn-lym-van šelm.
3.INAN ATT-at pleasure-taste-V.STATE HYPOTH
Here, eat this stew! You would find it tasty.

{dlen} introduces an "or else" or "otherwise" clause. Typically it would come after the {wǒn} clause, at the end of a conditional construction:

ĉǒ fu-čyw-da jâ-i purj, wǒn pĭw-kô i mrân-,
if light-sun-full state-at environment then play-place at relaxing.meal-V.ACT
dlen ty ŝě'ĥâ-môj tu-mrân-.
otherwise home inside chess-V.RECP AGT-relaxing.meal-ADJ
If it's sunny, we'll picnic in the park, otherwise we'll play chess while eating at home.

{dlen} can also introduce a clause presenting the alternative if a command or piece of advice is not obeyed.

du--ĵwa i ť wuŋ-i ƴâ-ĉa ĥy-i jĭlm-faj-cô-fwa-zô mwe,
part-person-place this at 2 own-at go-tool PAT-at open-able-OPP2-CAUS-V.ACT IMP
dlen wĭm-ŝum-da jâ-o še te ĥy-i Ł tu-i.
otherwise sack-float-full state-to maybe 3.INAN PAT-at 3.GEN AGT-at
Lock your car in this neighborhood, or else someone might fill it with balloons.

Miscellaneous modifier particles

Temporal particles

de nowadays; lately; (with {mje}) in those days (habitual aspect, extended tense)
gwe already (now in contrast with recent past)
ver still, yet (now as similar to recent past)

{mje} and {ler} were discussed briefly in the Verbs section. They are not used like tense marking in English and other Indo-European languages, obligatory on every finite verb, but tend to be used only when the time of an event is not otherwise clear from context. Often, {mje} can function kind of like a perfect aspect in an Indo-European language, indicating that something happened before the surrounding context (which may already be in the past from the speaker or listener's perspective).

źî'câ-cĭ i gjâ-krĭ krĭ-o byn-gĭn-.
1997-1 at language-create this create-at hack-begin-V.ACT
In 1998 I started creating this language.
gjâ-krĭ ĵyn-fwa-ĵar srǒ krĭ-o byn-zô mje.
language-create intellectual.pleasure-CAUS-less several create-at hack-V.ACT PAST
I had created several less interesting languages (before that).

Here, {mje} indicates that the second sentence takes place in the past relative to the first.

{ler} is used more often than {mje}, but not as often as {gwe} and {ver} (discussed below). Besides the basic meaning of "this verb's action will take place in the future," it can also occur after an event-noun or time-period noun phrase and indicate that it is the next, not the previous, instance that is meant:

měn'θu fy-gla ler i rě'ĵy-tôn jâ-o θě'mâ ke sě'râ.
month seven-ORD.T FUT at wife-GNR state-to Thomas and Sarah
Next July, Thomas and Sarah are getting married.

One could also write the same sentence with {ler} after {jâ-o} instead.

kî'prâ ĥun-i ler heŋ hǒl.
goat meeting-at FUT not totally
I will not, will not with a goat.

{de} is glossed as HAB = habitual aspect; it indicates that something happens (perhaps intermittently, perhaps continually) over an extended period of time. Whether it's translated as "nowadays, lately" or "in those days" depends on the temporal context (set by postpositional phrases, {mje}, or {ler}).

kî'prâ-daj ĥy-i wlâm-fwa- de jě'nu.
goat-mass PAT-at graze-CAUS-V.ACT HAB Joan
Joan is herding goats lately.
ĥwĭl i ewropa-wam i -van de sî'ðyr-źa-cu.
era that at Europe-NAME.P at happen-V.STATE HAB fight-AUG-system
Wars happened continually/intermittently in Europe at that period.

I use {ver} and {gwe} more often than any of the above particles. They indicate change or continuity in an action or process in implicit comparison to a past time.

flyr-da jâ-i ver kěr'nâ.
flower-full STATE-at still dogwood
The dogwood is still in bloom.
vě'ty-rĭm - ĥy-i šyj- gwe.
door-see seventeen-ADJ PAT-at wash-V.ACT already
ʝǒ tu-i mwe vě'ty-rĭm ble ĥy-i šyj-zô.
person other AGT-at IMP door-seeing others PAT-at wash-V.ACT
I've already washed seventeen windows. Someone else will have to wash the rest.

"not yet" and "no longer" are also expressed with {ver} and {gwe} in ways that might be counterintuitive at first.

źĭ kâ-i vy-zô kun-hôw-ca heŋ ver sě'râ. REL ATT-at decide-V.ACT know-CAUS-REFL not still Sarah
Sarah hasn't yet decided what field to study.
ĵyj-fja-van gwe, mǒj mwĭl-ŝra-van heŋ ver.
vigor-minimum-V.STATE already but sleep-soon-V.STATE not still
I'm already tired, but still not yet sleepy.
kyl-jâln- tyn-van heŋ gwe gî'sĭr.
box-warm-OPP2 inside place-V.STATE not already spaghetti
There is no longer any spaghetti in the refrigerator.
Dhalgren-gam kâ-i lju-zô heŋ ver.
Dhalgren-NAME.G ATT-at read-V.ACT not still
I still haven't / haven't yet read Dhalgren.

{ver} and {gwe} are glossed with respect to "now" in the table above, but they can compare a past or future time with a time before the referenced time.

źî'câ-gla i gjâ krĭ-i byn-gĭn- heŋ ver.
1997-ORD.T at language this create-at hack-begin-V.ACT not still
In 1997 I had not yet begun to hack this language.
nu mĭ-i nu kujm-ja šir gwe.
moment DEM2 TOP-at moment goal-suitable long.after already
By the time you read this, it will already be much too late.

Epistemic particles

There are several clitic particles that change the truth-value with which a phrase or clause is intended. Like other modifiers, they're postpositive, typically coming after the main verb of the sentence or the most verblike postpositional phrase, although they can also come after other sentence constituents. They can also stand alone as utterances responding to a question.

vǒm yes, certainly
heŋ no, not
fjǒ yes and no; sort of; to some degree
še maybe [facts]
be maybe [intentions]
le probably
ʝem apparently, seemingly
nen not as far as the speaker knows
tǒlm hyperbole/exaggeration marker
belm irony or sarcasm marker

These express one's certainty about something being true or false, real or unreal. {še} expresses uncertainty whether some statement or description is true or not. {fjǒ} expresses a belief that a sentence or description is valid to some degree, but not totally. {be} indicates an uncertainty of one's own plans or intentions.

bĭŋ-van vǒm.
exist-V.STATE yes

I certainly exist.

mwĭl-van heŋ.
sleep-V.STATE not

I'm not asleep.

kyl râm mĭ-i zuň-cô-bô še ŋĭn-i.
box inside cat TOP-at life-OPP2-ADJ maybe CMT-at

The cat in the box might be dead.

fĭw-câŋ fjǒ ŋĭn-i θuň-lâŋ . FUZZY CMT-at story-long that
That novel is sort of science fiction.

In the above sentences, the particles modify a whole verb or comment phrase. They can also be clitic'd to a word and used within a phrase.

kjĭ ĥy-i vâ-oŋ-zô râm-vǒm tu-i.
mouse PAT-at digestion-into-V.ACT cat-yes AGT-at

The definitely-cat eats a mouse.

{le} following the main verb of the sentence affects the truth-value of the sentence as a whole; following a specific noun phrase, it affects only its referent. For instance,

ť ĥy-i čĭn-ħulŋ-zô ler le vě'sî'pâ.
2 PAT-at prick-damage-V.ACT FUT PROB wasp
A wasp will probably sting you.
Ќ ĥy-i čĭn-ħulŋ-zô mje vě'sî'pâ le.
1 PAT-at prick-damage-V.ACT PAST wasp PROB
Something, probably a wasp, stung me.

{ʝem} likewise can follow the main verb, expressing reservations about the situation described by the sentence as a whole, or follow another sentence constituent and modify only that:

pwĭm-daj θoŋ ruŋ-van ʝem ƴâ-ĉa-ŝum.
water-mass go-V.STATE apparently go-tool-float
Apparently the ship sank.
Ќ ĥy-i čĭn-ħulŋ-zô mje vě'sî'pâ ʝem.
1 PAT-at prick-damage-V.ACT PAST wasp apparently
It was apparently a wasp that stung me.
ĥě'ĵuŋ gân-ř ʝem pwĭm-daj θoŋ ruŋ-van ƴâ-ĉa-ŝum.
storm cause-from apparently water-mass go-V.STATE go-tool-float
It was apparently because of the storm that the ship sank.

{nen} is a reserved, hesitant {heŋ}; it signifies that the speaker doesn't know of any particular evidence for a proposition, but won't assert that it's not true.

fĭm--ƥ-van nen.
She's not sick as far as I know.
?ðu-ť-van zǒn pjân-.
able-2-V.STATE Q.YN piano-V.ACT
Can you play the piano?
nen belm.
Not that I know of.

See also the section on derived validationality adverbs. The derived validational adverb {ĵrĭw-pôm}, from the root {ĵrĭw} "expecting, supposing", is something like a positive equivalent of {nen}: "As far as I know..." or "I suppose but don't have strong evidence that..."

Expressing probability and fuzzy degree of truth

A fractional number between zero and one, marked with the {-bô} adjectival suffix, can modify one of the clitics {be, še, le, fjǒ}.

firence-wam o ruŋ-zô ler be ðe--lwa-.
Firenze-NAME.P to go-V.ACT FUT maybe
There's about a one-third chance I will go to Florence.
bly-van θǒ le --ðe-gâr- pwĭm.
fall-V.STATE immediate probably water
There's a 90% of rain.
vlym fjǒ ĉu-ðe-- tyn-ƥ-ca
clothing into sort.of put-3-V.REFL
nu i lâl-van gju-ŋy-ĉa.
time REL at noise-V.STATE speak-far-tool
She had gotten a third of the way dressed when the phone rang.

Irony and Hyperbole

gjâ-zym-byn does not use tone to mark ironic or sarcastic remarks (or for any other purpose); instead, the particle {belm} is used, typically following the verb (like the yes/no question particles {zǒn} and {srem}) or at the end of a sentence, but sometimes marking an ironically intended noun phrase. The last use could sometimes be translated by the phrase "so-called" or the use of scare quotes.

twâ- Φǒ {pwĭ-van belm.} vĭj--šar
Ќ cim šâ-ŋĭw ĥy-i -- ₣âl- rjuŋ.
1 left.part.of carry-organ PAT-at digestion-into-V.ACT sudden-ADJ dragon
"How delightful!" I said as the dragon chomped off my left arm and hand.
ƥ lĭw-i lĭm belm tu-i ƥ ŝu-i sĭŋ-kě'ĝu -i
3 relation-at friend IRONY certain AGT-at 3 quality-at information-hidden TOP-at
sĭŋ-flu- ki lju-θaj-zô.
information-flow-place throughout read-OPP1-V.ACT
A certain "friend" of hers wrote about her secrets all over the Internet.

{belm} can also be used for ironic understatement, e.g.,

kujm-Ł-ja ŋĭn-i heŋ belm ðǒŋ Ł ĥy-i
purpose-3.GEN-suitable.for CMT-at not IRONY that 3.GEN PAT-at
vâm-plâŋ-zô šelm Φě'ĥu.
crush-foot-V.ACT HYPOTH elephant
It would not be convenient to get trampled by an elephant.

{tǒlm} marks deliberate exaggeration or hyperbole.

gě'dĭm hǒl il tǒlm gjâ mĭ-i mî'ħâ-van.
day whole during HYPERBOLE language DEM1 TOP-at obsess-V.STATE
I've been obsessed with this language all day [but not literally every moment thereof].
kî'pĭ ky- tǒlm ŝĭw-i mě'hu ķĭn-o.
pepper HYPERBOLE material-at stew making-to
I put thirty-seven peppers in this stew [no, not really that many, but a lot].

Deontic particles

mwe imperative, hortative, desiderative particle
źǒ negative version of {mwe}

{mwe} and {źǒ} are glossed as IMP = imperative and IMP.NEG = negative imperative, but in fact their use is broader than the imperative in English. They can express a wish, desire, advice or request as well as a command, and they don't cause verbs to default to having a second person subject. They can occur with any subject or an omitted default subject.

gjâ-zym-byn has a verb {dlâw-van} which can be glossed as "should, must, ought to," but it's used less often than its English equivalents; instead of structuring a sentence like "dlâw-van [second verb] [subject]", gzb would often use "[verb] mwe [subject]".

ť ŋâw-o twâ-zô mwe ce mĭ-i.
2 call-to say-V.ACT IMP that TOP-at

I should tell you about that.

re o ruŋ-ť- źǒ.
Don't go there.
suŋ-kě'ĝu-tla ŋâw-o twâ-zô Φǒ {*Φĭlm-van źǒ.} call-to say-V.ACT QUOTE butterfly-V.STATE IMP.NEG

"I don't want to be a butterfly!" I said to the sorcerer.

kiň θě'ku tu-i twâ-zô Φǒ {bĭŋ-van mwe fu.}, kiň bĭŋ-van fu.
and God AGT-at say-V.ACT QUOTE exist-V.STATE IMP light and exist-V.STATE light

And God said, "Let there be light", and there was light.

brâl-van, hǒŋ twâ- mwe le hǒŋ še vǒm
be.certain-V.STATE that say-V.ACT IMP probably that maybe yes
I am positive that a definite maybe is probably in order.

{mwe} and {źǒ} can occur after a postposition in a verbless sentence:

ť , mwĭl jâ-ř mwe.
2 VOC sleep state-from IMP
Wake up!

{źǒ} can occur following a noun phrase in at least one construction:

ĥâr-tôn sâŋ- źǒ rjâ-i rě'ju-ca.
spider-GNR blood-preferring IMP.NEG quest-at search-V.REFL
I checked myself for ticks.

Here, the {źǒ}-marked noun phrase as an object of {rjâ-i} means one is searching for something one hopes not to find. Similar constructions are used for e.g., checking one's boots for scorpions or someone's blood sample for signs of disease. I'm not sure if {mwe} or {źǒ} could occur following noun phrases otherwise.

{źǒ} is often used in subordinate clauses introduced with {kujm-šar}, "in order to," to negate the sense of the conjunction, "in order to avoid/prevent/stop something."

ƥ ŝu-i gjâ ĥy-i sâr--fwa-,
3 have.quality-at language PAT-at order-OPP2-CAUS-V.ACT
kujm-šar gjâ mĭ-i hî'mâr-môj źǒ ƥ tu-i.
purpose-CONJ language TOP-at understand-RECP IMP.NEG 3 AGT-at
I will confuse their language so that they won't understand each other.

{mwe} occurs less often in these contexts because the {kujm-šar} already implies something that someone desires to happen.

suomi-lam kâ-i suŋ-hôw-ca θě'mâ
Finnish-NAME.L ATT-at Thomas
kujm-šar suomi-wam o ruŋ-źa-.
purpose-CONJ Finland-NAME.P to go-AUG-V.ACT
Thomas is teaching himself Finnish in order to travel to Finland.

Postposition-like adverbs

presentative marker
vocative marker
jej enthusiasm marker

These are anomalous particles which have the form of {jum}, but can act like postpositions; they can mark a noun phrase so as to stand alone as a valid sentence by itself, and sometimes when they mark a constituent of a larger sentence it may not require another case-postposition. {gǒ} is a presentative particle, corresponding to French "voici" or "voilá", archaic English "lo" or "behold", or Esperanto "jen", drawing the listener's attention to something.

te .
3.INAN behold
Here it is.
vĭn .
wine behold
Look, [I brought] wine.

A noun phrase followed by {gǒ}, as a stand-alone sentence, can function like a topic-comment sentence; this structure tends to be more emphatic and informal than the topic-comment or comment-topic form.

kyr-ta ŋĭn-i twâ .
verb-less CMT-at sentence DEM1
This sentence [is] verbless.
twâ kyr-ta .
sentence verb-less behold
Look, a sentence with no verb.

When it marks a constituent of a larger sentence, {gǒ} can be a kind of topicalizer (in a different sense of "topic" than that with which it's used for the gzb "topic postposition" {mĭ-i}).

mwĭl-ŝra-van râm.
sleep-tending-V.STATE behold cat
See how sleepy the cat is.
swyŋ sin twâ-cu-vuj .
desk on sentence-system-concrete behold
There's the book, on the desk.
pân ĥy-i bâm-fwa-.
everything PAT-at behold new-CAUS-V.ACT
Behold, I make all things new.

{hǒ} is the vocative marker, following the second-person pronoun, a person's name, or another appellation when one begins an utterance addressed to them; it's not to be confused with the object-of-communication case postposition {ŋâw-o}, which marks someone's name (etc.) as the addressee of a communication-verb:

θě'mâ , ť wuŋ-i twâ-cu .
Thomas VOC 2 possess-at sentence-system behold
Thomas, here's your book.
mî'rĭj ŋâw-o θĭ rjâ-i -ť- mwe.
Mary call-to help quest-at request-2-V.ACT IMP
Ask Mary for help.

The vocative particle can mark a name or other appellation as a stand-alone utterance, when one's just calling someone to get their attention and not immediately saying something in particular to them:

*sě'râ .
Sara VOC
Hey, Sara!

A conventional greeting uses these two particles together with the first and second person pronouns:

ť , Ќ .
2 VOC 1 behold
Hello. [literally: O you, behold me.]

{jej} is vaguely similar in meaning and use to the English interjection from which its form is borrowed. It can work like {gǒ}, marking a noun phrase which thus stands alone as a valid utterance:

*râm-bâm jej.
cat-new yay
Yay kitten!
*ť jej.
2 yay
Go you!

Or it can mark a constituent within a larger sentence; in this use it overlaps in meaning with the affectionate and respectful attitudinal suffixes, but tends to express a more excited, enthusiastic attitude; it can in fact be combined with an attitudinal suffix.

tyn o ruŋ-zô jej pî'tĭr-la.
place DEM1 to come-V.ACT yay Peter-ATD1
Whoo hoo, Peter's arrived!
*gâm-ʝĭl gân-ř kě'pâ jâ-o jej.
picture-motion DEM3 cause-from happy.bewilderment state-to yay
Wow, that film has put me in a state of happy bewilderment!

These adverbs can also function as interjections, as stand-alone utterances:





Most of these particles describe the degree to which some adjective or verb is applicable. Some can also be used with noun phrases.

θje almost; not quite
fem at least, anyway
ƴeŋ barely, hardly
fe slightly, a little bit
źe very, very much, a lot
ķe too, too much
žǒŋ only, merely, simply, just
mew even, also, too
gem especially, particularly
jǒj again, more, still
ť dâm-ř θuň -i lju- θje ŋĭn-i,
2 authorship-from story TOP-at read-worthy not.quite CMT-at
mǒj te ðim du fem -i ĵyn-fwa-van.
but 3.INAN early.part.of chapter at.least TOP-at interest-CAUS-V.STATE
Your story is not quite worth reading, but its first chapter at least is interesting.
te im du ĉu-pa mew im -fĭw pym-fwa gem.
3.INAN part.of chapter two-ORD also part.of person-fiction amusement-CAUS especially
In the second chapter too there are some especially funny characters.
mǒj du -pa -i hî'mâr-faj ƴeŋ pe zâň- ķe ŋĭn-i.
but chapter three-ORD TOP-at understand-able barely and idiosyncratic too.much CMT-at
But the third chapter is barely comprehensible and too idiosyncratic.

{mew} "also, too" can come after any number of sentence constituents. Depending on the intended meaning, it can come after a noun phrase in subject position with no overt postposition, or after the postposition of a normal postpositional phrase, or after a verb.

kĭlm o ruŋ-zô mew θě'mâ.
party to go-V.ACT too Thomas
Thomas also came to the party (in addition to other things he did).
kĭlm o ruŋ-zô θě'mâ mew.
party to go-V.ACT Thomas too
Thomas, too, came to the party (in addition to other party guests).
kĭlm o mew ruŋ-zô θě'mâ.
party to too go-V.ACT Thomas
Thomas came to the party, too (in addition to other places he went).

If the subject pronoun is incorporated into the verb, it can be ambiguous as to whether {mew} is intended to modify the subject or the verb.

mluŋ-čĭm ĥy-i šâ-ƥ- mew.
bread-chocolate PAT-at carry-3-V.ACT too
He also brought chocolate cookies. / He, too, brought chocolate cookies.


These are typically used as nonspecific quantifiers with noun phrases. When used to modify a verb, they (like number-adjectives in the same context) mean the action is done so many times.

zen only, no more than; no one but; (when not qualifying a number) single, sole, alone
kwǒ some, any
srǒ several
few, little
reŋ much, many, a lot
gle enough, sufficient
jǒj more, extra; again
pen all, every, each
hǒl whole, entire
jǒm most of, the majority of
ðǒl mostly

Note the contrast between {cǒ} (few, little) and {fe} (little, slightly). {cǒ} modifies noun phrases, whether count or noncount: {rî'zĭ cǒ}, a little rice, {θĭl cǒ}, a few potatoes. {fe} modifies verb phrases or adjectives, indicating that the action is done to a lesser than usual degree or in a casual, haphazard way, or for a shorter than usual time, or that the quality is present in a limited degree. {cǒ} used with verbs means that the action is done a few times or for a short time. There can be some overlap in their meaning vis-a-vis verbs. {reŋ} is similarly indifferent to count/mass distinctions: {lî'klâ reŋ}, a lot of milk; {cî'jyr reŋ}, many squirrels. (In gzb there is little or no real distinction between count and noncount nouns as in English, since {cǒ} and {reŋ} translate both "few/little" and "many/much".)

gâm-ʝĭl -i rĭm-van de.
picture-motion few ATT-at see-V.STATE HAB
I see few movies nowadays.

Compare {fe} with the fuzzy logic clitic {fjǒ}:

gâm-ʝil -i rĭm-van fjǒ.
picture-motion DEM1 ATT-at see-V.STATE FUZZY
I am sort of watching this movie.
gâm-ʝil -i rĭm-van fe.
picture-motion DEM1 ATT-at see-V.STATE slightly
I watched part of this movie.

{jǒj}, "more, again" is a bit polysemous, especially with verbs:

lukan-ram dâm-ř ĵĭ ₣urŋ-fwa -i lju- jǒj.
Luke-NAME authorship-from announcement blessed-CAUS ATT-at read-V.ACT more/again

could mean, "I read some more of Luke's Gospel" or "I read Luke's Gospel again"; but probably the latter. For the former sense one would more likely say {lju-zô jǒj fe}.

{kwǒ} is similarly polysemous, but not as often ambiguous.

kwǒ wuŋ-i vjâr-ĉa ĥy-i tru-zô.
person some ownership-at shadow-tool PAT-at find-V.ACT
wuŋ-i zǒn te -i.
2 ownership-at Q.YN 3.INAN TOP-at
I found somebody's parasol. Is it yours?
kwǒ -i, tu-i re o ruŋ-zô, ĵlân-- ŋĭn-i.
person any TOP-at person REL AGT-at there to go-V.ACT wisdom-OPP2-ADJ CMT-at
Anybody who goes there is foolish.

The particles {zen} and {žǒŋ} are both glossed as "only". {zen} is a quantitative "only"; it is used in sentences like "Only Tom came to the party", "We have only four bananas left", and so forth. {žǒŋ} could also be translated "merely" in many cases; for instance, when a door has opened apparently by itself, you might say,

by-flu žǒŋ ŋĭn-i.
air-flow only CMT-at
It's only the wind.

{ķe} describes an excess of a quality or action; to say there are too many of something, use {reŋ ķe}. (I may change this.) "Too little" (quality) and "too few" (quantity) are {ķe-cô} and {cǒ ķe}.

ɱ ŝâj-i râm-bâm -i gju- ķe juri-ram
3 possession-at cat-new TOP-at speak-V.ACT too.much Yuri-NAME
Yuri talks too much about his kitten.
râm reŋ ķe ðĭ-i ŝâj-ť-van;
cat many too.much relationship-at possession-2-V.STATE
ƥ pen ĥy-i ðu-van heŋ šu-
3.AN all PAT-at able-V.STATE not
You have too many cats; you can't take care of them all.
Ќ ŋâw-o nî'šĭm ĝy-i gju-ŋy-ť- reŋ ķe.
1 call-to night middle-at speak-distant-2-V.ACT many too.much
You've telephoned me too many times in the middle of the night.

Some of these nonspecific quantifiers can modify another quantifier or a number-adjective.

ƥ ty-i ty-van râm fy- gle belm.
3 home-at home-V.STATE cat seven-ADJ enough IRONY
Seven cats — enough, I think — live with him.
fy jǒm ʝâr-i kjĭ -i ħum-van.
seven most.of experiencer-at mouse TOP-at fear-V.STATE
Most of the seven are afraid of mice.

(Note that here {fy}, "seven", is the head of a noun-phrase rather than an adjective as {fy-bô} in the previous sentence; in such a context it doesn't represent the mathematical object "seven" but rather a salient, recently mentioned set of seven things.)

{jǒm} and {ðǒl}

The modifiers {jǒm} and {ðǒl} have similar but distinct meanings. {jǒm} indicates a subset, a set consisting of most of the members of a superset denoted by the head noun phrase.

twâ-cu jǒm most of the books
tu-gu jǒm most of the voters, a majority

{ðǒl} on the other hand indicates a superset, mostly consisting of members of the set denoted by the noun phrase modified by {ðǒl}. By contrast, you might use {twâ-cu ðǒl}, "mostly books", to describe a box of heterogeneous things including books, clothes, dishes, and ceramic statues of frogs, or {tu-gu ðǒl}, "mostly voters", to describe the set of people you polled on some political question.

These minimal pair sentences may contrast their meanings more clearly:

ĝâk-ram vâns-šam dâm-ř twâ-cu jǒm kâ-i lju-zô.
Jack-NAME Vance-NAME.F author-from sentence-system most.of ATT-at read-V.ACT
I've read most of Jack Vance's works.
ĝâk-ram vâns-šam dâm-ř twâ-cu ðǒl kâ-i lju-zô.
Jack-NAME Vance-NAME.F author-from sentence-system mostly ATT-at read-V.ACT
I'm mostly reading stuff by Jack Vance.

In the latter sentence, one is not asserting that one has read the majority of Vance's works, but is implying that one is also reading some other works by people other than Vance.

{jǒm} always means at least a 50%+1 majority, and may pragmatically suggest a fairly large majority; {ðǒl} need not indicate a majority, only a plurality — perhaps you might utter the second sentence above if four of the last ten books you'd read were by Jack Vance, and you'd read no more than three by any other author in this {drulm}.

{jǒm} and {ðǒl} can modify postpositional phrases as well.

mwĭl jâ-o jǒm gwe râm.
sleep state-to most.of already cat
The cat has mostly fallen asleep.
ĝâk-ram vâns-šam ₣um-i ðǒl mě'θâj hjuz-šam.
Jack-NAME Vance-NAME.F similar-to mostly Matthew Hughes-NAME.F
Matthew Hughes mostly resembles Jack Vance (= his style resembles Vance's more than that of any other author)

In some cases, {jǒm} or {ðǒl} following a postpostion can be interpreted as modifying an implicit noun phrase between themselves and the postpositional phrase; for instance,

ĝâk-ram vâns-šam dâm-ř jǒm kâ-i lju-zô.
Jack-NAME Vance-NAME.F author-from most.of ATT-at read-V.ACT
I've read most of Jack Vance's (works).

{jǒm} also has a qualitative sense, usually when it's used with one modifier in contrast with another; for instance,

mjyl-rô jǒm ŋĭn-i te, mǒj sâl-tan fe mew ŋĭn-i.
honey-QUAL mostly CMT-at 3.INAN but salt-like slightly also CMT-at
It's mostly sweet, but also a little bit salty.

With units of measure, {jǒm} means "more than half, less than one":

ĉě'θâ jǒm il rî'mâ tyn-van ver.
day most.of through house inside place-V.STATE still
I stayed inside most of the day.
mě'tyr-źa jǒm jĭrn-i ƴâ-cjaj-.
meter-AUG most.of quantity-at motion-SPEC-V.ACT
I walked most of a kilometer.

{ðǒl} with verbs means that a plurality or majority of the speaker's time is spent on the action described by the verb:

gě'dĭm i gjâ ĥy-i byn-zô ðǒl.
day DEM1 at language DEM1 PAT-at tinker-V.ACT mostly
Today I'm mostly tinkering with this language. (= that is how a plurality of my time is spent)
tâŋ i ðurm ðĭ-i šĭm-cu-vuj kâ-i ħĭ- ðǒl de.
life.era DEM3 at work relation-at algorithm-system-concrete ATT-at test-V.ACT mostly HAB
In those days my work mostly involved software testing.

{jǒm} means that the ongoing action of the verb is mostly completed, or that the now-completed action of a verb was mostly done within a certain time-frame. It might be combined with other aspectual adverbs or a temporal complement phrase for clarity.

θuň-lâŋ krĭ-o lju-θaj-zô jǒm.
story-long DEM3 create-at read-OPP1-V.ACT most.of
I'm mostly finished writing that novel.
źě'mâ-ky-gla i θuň-lâŋ krĭ-o lju-θaj-zô jǒm.
1973-37-ORD.T at story-long DEM3 create-at read-OPP1-V.ACT most.of
In 2010 I wrote the bulk of that novel (but perhaps started it earlier and/or finished it later).

{ðǒl} is a relatively late addition to the language; in older texts {jǒm} is sometimes used in the same sense {ðǒl} has now.

Error correction particles

There are two error correction particles, used for correcting an utterance partway through when one realizes one has made a mistake. {Φej} deletes the previous word (tells the listener to please ignore it), allowing one to utter some other word in its place. {če} is a conjunction that reverses the normal word order between its two arguments: so ADJ če NOUN is acceptable though NOUN ADJ would be the normal correct order. Also related is the hesitation particle {hem}, corresponding to "uh, um" in English: "I haven't thought of the next word yet, but I'm not done talking, be patient."

bě'gru Φej bě'jâ řm ʝym mĭ-i šul jyn-fwa ŋĭn-i.
beaver bay.tree out.of leaf TOP-at spice pleasure-CAUS CMT-at
Beaver, I mean bay, leaves are a good spice.

suw-fwa če Φě'ĥu-bâm tu-i
enjoyment.of.cuteness-CAUS REV elephant-new AGT-at
ƴâ-cjaj-zô, hem, flâň-bô.
move.under.control-SPEC-V.ACT um shaky-ADJ

The cute baby elephant walks, um, sort of wobblily.

(I stole the idea for the particle {Φej} from Jeffrey Henning's Fith, a stack-based language that has particles which simply pop the top item, or all items, from the stack and discard them. Unlike the Fith conjunctions frong or skuunh, however, {Φej} can't be expected to make the listener really forget the previous word, since gjâ-zym-byn is a human language and the language center of human brains doesn't work that way. So it's not used to weaken severe insults into mild ones like the corresponding Fith particles, but simply to ask pardon for slips of the tongue, as above.)

I use {Φej} a lot when talking to myself in gzb, and not infrequently in writing. {hem}, not so much.

{hem hem hem} is roughly equivalent to "blah blah blah" in English, filler to represent speech or text whose exact content is unknown or unimportant.

Various other modifier particles

ʝǒ other, another
ble others, rest of, remainder of
θǒ immediate, next, previous
certain, specific
rew apiece, each
mrel discourse marker indicating resumption of a dormant topic
men on the one hand... on the other hand...
ŝe in contrast, by contrast, on the contrary
sem so, therefore, thus

{ʝǒ} following a noun phrase indicates a different instance of the referent of the noun phrase than was previously mentioned, or than is about to be mentioned.

Ќ lĭw-i ƴâw ĥun-i ƴâ-cjaj-.
1 relationship-at dog meet-at go-SPEC-V.ACT
ƥ ŋâw-o ħâň- ƴâw ʝǒ.
3 call-to yell-V.ACT dog another
I went for a walk with my dog. Another dog barked at her.
lju-θaj-tla srǒ ʝǒ ĥun-i θuň-lâŋ
read-OPP-professional several other meet-at story-long
srǒ krĭ-o lju-θaj-zô lě'kjân.
several create-at read-OPP-V.ACT Alex
Alex has written several novels in collaboration with several other authors.

{ble} indicates that, in addition to the referent of the previous noun phrase, all the other entities in the relevant category are intended.

źy ble kâ-i hyw-van heŋ gwe.
dream rest.of ATT-at know-V.STATE not already
I no longer remember the rest of (the night's) dreams.

[From a journal entry, just after an account of one dream I remembered well.]

{θǒ} with a noun phrase denoting a time-period or event can mean either "next" or "previous" depending on context. If context is otherwise insufficient, {θǒ} is often used along with {mje} (past) or {ler} (future). Modifying a verb, {θǒ} means "immediately, right away" or sometimes "suddenly."

ĥun θǒ i, šâ-ť- mwe twâ-cu ĥy-i.
meeting next/prev at carry-2-V.ACT IMP sentence-system DEM3 PAT-at
Bring that book to [our] next meeting.
gě'dĭm mje θǒ i Ќ c-i-m zjâm-ny ĥy-i ĥâ-ca.
day PAST previous at 1 left-in-part.of finger-small PAT-at cut-V.REFL
I cut my left pinkie yesterday.
kyl vĭn-da - ĥy-i -- ƥ,
rigid.container wine-full one-ADJ PAT-at digestion-into-V.ACT 3
nu-šar ŝrun-twâ-gĭn- θǒ.
moment-CONJ music-sentence-begin-V.ACT immediate
He drank one glass of wine, then immediately started singing.

{rew} can be glossed as "apiece," though it has a broader use than "apiece" in English. It indicates a distribution of something to multiple targets in some way. Unlike in English, it marks the targets of distribution, not the thing distributed. (In older texts, its use was inconsistent, sometimes marking the target and sometimes the thing distributed, and sometimes coming after the noun phrase, sometimes after the postposition.)

ty rew o ruŋ-zô pen.
home apiece to go-V.ACT person all
Everyone went to their separate homes.

Here, the things distributed are the agent.

swyŋ rew son krĭ-vuj-ĉa - ĥy-i tyn-ť- mwe.
desk apiece onto create-concrete-tool one-ADJ PAT-at place-2-V.ACT IMP
Put one pencil/pen onto each desk.

Here, the things distributed are the patient. In both the above, the target is the object of a spacetime postposition, but it can be marked by an abstract genitive postposition as well:

lĭm rew ŝâj-o twâ-cu ĉu-bô ĥy-i bwĭl-zô.
friend apiece possess-to sentence-system two-ADJ PAT-at give-V.ACT
I gave two books apiece to each of my friends.

or other abstract postpositions:

ŝrun-twâ rew ĉul-i mâ-ĵĭn pen tu-i.
music-say apiece performance-at person-young all AGT-at
Each of the children sang a (different) song.

Contrast with:

ŝrun-twâ ĉul-i mâ-ĵĭn pen tu-i.
music-say performance-at person-young all AGT-at
All the children sang a song (the same song, in chorus).

{mrel} comes at the beginning of an utterance and indicates that the speaker is resuming a topic that has lain dormant for a while. I use it in my diary when one diary entry continues the narrative of the same events from a previous one. One might begin a conversation with {mrel} to remind the listener of a topic you were talking about earlier. It could be translated as "As I was saying," or "Remember what we were talking about earlier? I have more to say about it."

mrel vĭj-brĭ-ƴu-cô ði jâln-rî'mâ-ĉa
RESUME time-daytime-long-OPP2 before heat-house-tool
ĥy-i grĭ-fwa-zô mwe Ќ-ť tu-i.
PAT-at function-CAUS-V.ACT IMP 1-2 AGT-at
As I was saying, we need to get the heating system fixed before winter.

{men} is repeated after two or more phrases, or after the main verb or postpositional phrase of two or more clauses, to express a series of contrasted alternatives. It can be translated "on the one hand... on the other hand..." though there is no arbitrary limit on how many {men}s one can have in a sentence.

θuň krĭ-o lju-θaj-zô be men, krĭ-gjâ-zô men.
story create-at read-OPP-V.ACT maybe ALT create-language-V.ACT ALT
On one hand I might write a story, on the other hand I might do some conlanging.

{ŝe} indicates that the preceding sentence constituent (noun phrase, postpositional phrase, verb phrase) is in contrast to some other corresponding constituent, either in the same clause or an earlier clause, or perhaps an implicit contrast. Compare to the comparison postposition {dî'fu-i} and the contrastive conjunction {ðe}.

tâŋ srǒ ði tyn reŋ o ruŋ-Ќ- de,
personal.era several ago place many to go-1.V.ACT HAB
mǒj tâŋ i ŝe kě'vĭj gân-ř
but personal.era this during contrast COVID-19 cause-from
ty ř ruŋ-zô ƴeŋ.
home from go-V.ACT barely
In those days I used to go lots of places, but nowadays, because of Covid, I barely leave home.
ĥul jâ-o źǒ, cluŋ-cô-zô mwe ŝe.
anger state-to IMP.NEG repaying.good.for.evil-OPP2-V.ACT IMP by.contrast
Don't get mad, get even.

{sem} is a postpositive clitic with a meaning similar to {ŝǒn, ʝon, wǒn}. It's not used very often.

tě'θru-van sem pě'pâ-daj.
collapse-V.STATE so paper-mass
So the pile of papers fell over and scattered.
kujm o gî'sĭr-tôn ðĭ-i mĭ-i gju-Ł- źǒ sem.
purpose DEM3 to spaghetti-GNR relationship-at event TOP-at speak-3.GEN-V.ACT IMP.NEG so
So that's why we don't talk about the noodle incident.


The suffix {-pôm} derives evidentiality adverbs from root words or compound stems referring to the source of information. Such adverbs can be placed after a verb or a postpositional phrase, or at the beginning of a sentence.

bly-van ku-pôm pwĭm -i.
fall-V.STATE hear-EVD water TOP-at
It's raining (I hear it).
fĭm--van rĭm-pôm ƥ ʝâr-i.
healthy-OPP2-V.STATE see-EVD 3 experiencer-at
He's sick (I saw him).
jě'lĭ -i lju-pôm fĭm-fwa ŋĭn-i.
garlic TOP-at read-EVD healthy-CAUS CMT-at
(I read somewhere that) garlic is good for you.
tam-ram-pôm twâ-cu -i ĵyn-fwa heŋ ŋĭn-i.
Tom-NAME-EVD sentence-system DEM3 TOP-at interest-CAUS not CMT-at
(Tom tells me that) that book isn't very interesting.

Evidentiality marking is optional and actually fairly uncommon in gzb, though I'm trying to make myself use it more often in my writing when appropriate.

Use of {-pôm} to form validational adverbs

{-pôm} can also produce validational adverbs; words expressing the speaker's degree or kind of certainty about what they are saying. Sometimes the difference between a validational and evidential adverb is blurry:

hyw memory, experience
hyw-pôm I know this from experience / I remember this happening
źy dream
źy-pôm This happened in a dream / This didn't really happen, I just dreamed it, but it's interesting
brâl certainty
brâl-pôm I'm certain of this
ĵrĭw supposing, guessing
ĵrĭw-pôm I suppose, I guess that...
frâ-θaj-ƥ- Φǒ: {*vǒm, -mruň-ŋô ðu-gju- ŋĭn-i Ќ hyw-pôm.}.
ask-OPP2-3-V.ACT QUOT yes thing-mountain-element able-speak-ADJ CMT-at 1 experience-EVD
He answered, "Yes, I know from experience that I'm a talking stone!"

(from the LCC2 translation relay text)

źy-ŋa-pôm kyn-la-ĉu ty-vir
dream-ATD.surprise-EVD parent-ATD.affection-two
tyn-van ₣âl-ĉa kwǒ. ~~~
place-V.STATE sudden.change-tool some ...
(I dreamed (astonishingly) that:) Way out in front of my parents' house was a transformation machine.

(From a passage in my journal describing a dream weird even by dream standards.)

Use of {-pôm} to form attitudinal adverbs

When {-pôm} is used with a root or stem for a mental state, it forms an adverb describing the speaker's attitude to the situation described by the sentence or utterance. No ambiguity with the use of {-pôm} to form evidentiality adverbs is likely to result, as a mental state cannot be construed as evidence for anything.

sjum-pôm jĭlm-ĉa ĥy-i tru-zô.
thankfulness-ATT open-tool PAT-at find-V.ACT
Thankfully, I found my keys. (= I found my keys; I feel thankfulness.)
hěl'vĭ-pôm ƥ lĭw-i lĭm ĥy-i ₣ĭ--ƥ-
contempt-ATT 3 relationship-at friend PAT-at trust-violation-3-V.ACT
Contemptibly, he betrayed his friend. (= He betrayed his friend; I feel contempt.)
sru-kun-pôm gjâ-krĭ krĭ-o ƥ tu-i im
desire-knowledge-ATT language-create create-to 3 AGT-at thing REL in-part.of
ĉĭ-twâ-θy tâň-θaj-faj ŋâ-bô.
kind-sentence-element take-OPP1-able 53-ADJ

Interestingly, she has devised a conlang with fifty-three open classes. (= She has devised a conlang with fifty-three open classes; I want to know more.)

It appears, from a look at my online corpus, that I use {-pôm} more often for attitudinals than for validationals, and more often for validationals than for evidentials -- although {-pôm} was originally used for evidentials, and its use was stretched to cover validationals and attitudinals later. This is probably a function of my primary uses of the language: in writing my journal, if gzb were the sort of language where all indicative sentences are obligatorily marked for evidentiality, the vast majority of sentences would be marked as "direct experience" or "visual". But occasions to obliquely mention my attitude to the situation expressed by a sentence are far more frequent than occasions to mention how (other than by direct experience, the pragmatic default in such a context) I know about something. As for validationals, I find that the most common (in)validational use of {-pôm} in my journal is {źy-pôm}, "I dreamed this" (see example above). A common pattern is to begin one paragraph with {źy-pôm} and then, after recounting the dream, to begin another paragraph with {hyw-pôm} "I experienced this".

Main {gjâ-zym-byn} index
Derivational morphology
Abbreviations used in the interlinear glosses
My conlang page
My home page

Last updated September 2023.