Trukva: Conveying Meaning With Adverbials


The Name

‘Trukva’ simply means ‘the Trukva language’ in Trukva. It is a lexicon entry with no further interpretation, but just denotes this conlang.



The basic structure in Trukva are adverbials. The structure was motivated and presented at the Fourth Language Creation Conference in Groningen in 2011. The idea was lying around for a long time, and I had already documented it in Tesяfkǝm, but I had not continued to any state of usefulness.

‘Adverbial’ will be used in a very broad sense, and probably a little differently from what is typically seen to be an English adverbial.

In English, adverbials are all structures that can modify a complete clause to yield a new clause. Adverbials are optional components in English for modifying clauses. In the following sentence, ‘with chopsticks’ is an adverbial:

I eat →I eat with chopsticks

The simplest form of adverbials are adverbs, which single words that modify clauses. The following lists some adverbials in English:

often quickly with chopsticks using chopsticks when I am hungry

In Trukva, adverbials are constructed similar to ‘using chopsticks’, i.e., using a verbal part and a nominal part together. In English, adverbials are optional and modify sentences: they add to the meaning. Another idea for Trukva was to leave out everything but the adverbials and to see how clauses composed only of adverbials could convey all the intended meaning.

Serial Verb Construction

The idea of using adverbials comes mainly from serial verb construction (SVC) in Mandarin and other natural languages. Here, a string of verb-object phrases and adverbs is used to form a longer sentence. E.g. starting from two sentences, a new sentence can be formed where the verb-object phrases are stringed together in something that is translated as a adverbial phrase in English:

yòng kuàizi.
he use chopstick
He uses chopsticks.
chī yángyúpiàn.
he eat crisps
He eats crisps.
筷子 洋芋片。
yòng kuàizi chī yángyúpiàn
he use chopstick eat crisps
He eats crisps with chopsticks.

Trukva uses basic phrases like 用筷子 (yòng kuàizi) ‘using chopsticks’ and 吃洋芋片 (chī yángyúpiàn) ‘eating crisps’ to form sentences.

Note that although the Mandarin sentence translated as ‘he eats crisps’ looks superficially similar to the English equivalent, the structure is different, since ‘he’ is not a subject but a topic. So a more literal translation may be: ‘Concerning him, crisps are eaten, chopsticks are used.’ We're closing in on the structure of Trukva here.

Compared to the above Mandarin examples, the Trukva sentence contains no topic in front of the serial verb construction. Instead, it uses an additional adverbial. Thus in designing the lexicon, each transitive verb has to be broken down into its parts until there are only intransitves. Of course, we might collapse intransitives, e.g. the object of ‘to eat’ and the object of ‘to drink’ would identify on their own whether it's eating or drinking that is going on, so we can use the intransitive verbal ‘to be ingested’ for both of them. And we do: the above sentence reads as follows in Trukva:

Ta kengilvifa postifoksisku pongihohkisku tru.
[I see/saw, that] He eats crisps with chopsticks.

Literally, this is ‘he eats, crisps being ingested, chopsticks being the tool.’ Because we split the transitive verb into two substantives here, in contrast to Mandarin, there is no topic.

As you can also see from the Trukva example, a pair of verbal and nominal is one word in Trukva and the initial consonant of the nominal mutates:

crisps being eating / crisps being (the) food

Other conlangs, e.g. AllNoun, analyse this structure of combining two words, one being the role marker, one being the argument, as only one part of speech. In my view, the different positions of a word in such a pair (which I call ‘adverbial’) justify the view of having two parts of speech. But because there are no constraints on which part of speech a substantive may be used in, there is only one open lexical class of words.

Yet other conlangs, e.g. Kēlen (or my own Qþyng|ài), try harder to have only one part of speech by using a closed class of case markers plus a closed class of verbal-like morphemes (relation markers, valence markers, whatever).


Calling the two parts of an adverbial verbial and nominal is purely arbitrary. Several conlangs have used the same concept with a different name. And in natlangs, such structures correspond almost 1-to-1 to things that are called differently. The following table lists a few possible synonyms that might be used:

verbial nominal e.g. ‘leaving Germany’ this is what this text uses
preposition noun e.g. ‘from Germany’
case noun e.g. fi. ‘Saksasta
role argument e.g. AllNoun ‘Germany : source’
foo bar

So what you call it really doesn't matter, especially since there is only one open lexical class.

Please note that even though AllNoun looks very similar in structure, by using the notion of ‘whole’, most role markers are actually implicitly transitive relations and a sentence is more like Mandarin in structure, when identifying ‘whole’ as the topic. This is unlike Trukva, where there is no such notion, but simply a sequence of adverbs without identifying a topic. All transitive concepts have to by split into two substantives here. The connector is the serial adverb construction. Of course, somewhere, semantically, the concepts do connect in Trukva, but not at the syntactic level.

If you look closer, AllNoun is actually very different from Trukva although it looks superficially similar due to pairing of substantives. Mandarin, I would say, is much closer in structure to Trukva (except for the notion of a topic, which is closer to AllNoun).

This was a short overview of the basic grammar concepts. The following sections describe the details.

Short Classification

Trukva is a typical engelang of mine. I am fond of a few things I have always done in my engelangs so far, and this is no exception. So what I always do is:

  • Only one open lexical class
  • Mandatory evidentials
  • Other categories optional (including number, tense, etc.)
  • No articles
  • Completely regular
  • Some degree of surdéclinaison (especially for tense and number)

Some ideas from previous conlangs will be used as ‘modules’ in Trukva, maybe sometimes slightly changed. This includes:

Trukva's main contribution is the sentense structure briefly described above. I feel this is progress compared to the case marking system of Qþyng|ài.

A neglected topic in Trukva is the phonotactics. I could spend months on this without ever really liking the outcome. So I decided to use some off-the-shelf phonotactics and not think about it too long, in order to not distract myself from the main design goal: the new adverbial sentence structure using adverb serialisation. The phonotactics might not look like off-the-shell now, but it is really only a modification of a very simple design. I could spend many more months on this, but I won't, for this language.

Further, Lexicon design is not one of my favourites. I am never satisfied with my lexicons, and due to enormous amounts of individual work (i.e., for each word), I find it tedious. So the lexicon design will be kind of ad-hoc, although I will try to use only states and derive events by derivational suffixes. I will allow lexicalised compounds and lexicalised derivation to have some freedom of design.

Finally, grammatical aspect is not well developed and needs more work.



The following phonemes exist in Trukva.

labial alveolar velar
voiceless voiced voiceless voiced voiceless voiced
stop p t k
nasal m n
fricative f v s x
trill r
liquid l

Varying in context and from speaker to speaker, the following allophones exist:

labial alveolar retroflex velar uvular glottal
voiceless voiced voiceless voiced voiceless voiced voiceless voiced voiceless voiced voiceless voiced
stop p b t d k ɡ
nasal m n ŋ
fricative f v s ʂ x χ h
trill r
liquid ɬ l

[x], [χ] and [h] are allophones of /x/. The pronunciation varies from speaker to speaker. The phoneme is written as ‘h’.

The other allophones occur in consonant clusters. For more information, see the section about sandhi.


The following table lists the spelling of the vowels of Trukva. It is a typical three-height vowel system.

front central back
close i u
mid e o
open a

Unaccented syllables only use a, i and u.


In a string of phonemes, there are certain constraints on and changes to pronunciation. There is a general tendency for voicelessness to spread from stops, and voicedness to spread from nasals. Further, epenthetic consonants pop up. Variations are given in the following table:

pr → [pr] ~ [pr̥]
tr → [tr] ~ [tr̥]
kr → [kr] ~ [kr̥]
mr → [mbr]
nr → [ndr]
pl → [pl] ~ []
fl → [fl]
ng → [ŋɡ]
rs → [rs] ~ [ʂː] ~ [ʂ]
rp → [rp] ~ [r̥p]
rt → [rt] ~ [r̥t]
rk → [rk] ~ [r̥k]
ls → [ls] ~ [lts]
lp → [lp] ~ [ɬp]
lt → [lt] ~ [ɬt]
lk → [lk] ~ [ɬk]
ns → [ns] ~ [nts]
tl → []
hr → [χr] ~ [χr̥] ~ [hr] ~ [xr] ~ etc.
hl → [χl] ~ [χɬ] ~ [hl] ~ [xl] ~ etc.
rh → [] ~ [r̥χ] ~ [rh] ~ [rx] ~ etc.
lh → [] ~ [ɬχ] ~ [lh] ~ [lx] ~ etc.
kv → [kv] ~ []


There are two types of words in Trukva: those that begin with p, t, k and those that begin with l,r.

Those that start with p, t, k are called substantives. This is explained in another section.

The group of words starting with l, r is called special suffixes.

For substantives, the initial consonant mutates depending on how the word is used in a sentence. Trukva distinguishes three mutation grades for all substantives: ‘verbal grade’, ‘nominal grade’ and ‘suffix grade’. They are marked by the initial consonant of the morpheme: ‘stop mutation’, ‘fricative mutation’, and ‘nasal mutation’, resp.

Verbial Grade Nominal Grade Suffix Grade Lexical Suffix
Stop Mutation Fricative Mutation Nasal Mutation
p f m
t s n
k h v

As an example, pitsa ~ fitsa ~ mitsa means ‘pizza’. Which form is used will be explained in the section about syntax.

As you can see, some initials exist only in suffixes. Such lexicon entries cannot be used as verbals or nominals.

There were some questions as to why k mutates into v when the other stops become a nasal. Well, the imaginative conhistory would be that it originally was a nasal, but then changed. Maybe like this: ŋ > ɣ > ɣʷ > w > v. Finnish is a language that has some k→v mutations, namely when k between u undergoes consonant gradation, e.g. pukuapuvun (genitive case). And Kisuahili has kideovideo, although neither of the two prefixes ki- and vi- is likely to be derived from the other.

Anyway, in the lexicon, the verbal grade will be listed, except for suffix-only morpheme, which will be listed in suffix grade.

Each morpheme in Trukva has the following phonological structure:

Initial Part Remaining Part(s) [Repeatable]
Consonant Vowel Consonant(s) Vowel
Verbial Grade/
Stop Mutation
p t k
pr tr kr
pl kl
a e i o u
mb nd ng
ps ts ks
sp st sk
hp ht hk
a i u
Nominal Grade/
Fricative Mutation
f s h
fr sr hr
fl hl
Suffix Grade/
Nasal Mutation
m n v
mr nr vr
ml vl
Lexical Suffix
Normal Scope
l ls lv
r rs rv
a i u
Lexical Suffix
Morpheme Scope
ln lp lt lk
rn rp rt rk

The initial mutation is used in Trukva as the single means to mark syntactic roles and parts of speech. There are no other affixes for this, nor are there any particles for this – the three grades suffice under all circumstances.

Some suffixes modify only the morpheme they are attached to. Only lexical suffixes can do that, all other modification has ‘broad scope’, modifying the whole structure they are attached to. Such suffixes can be identified by their initial consonant cluster, as show in the above table.

All suffixes and non-stressed syllable use only three vowels: a, i or u.


Morphemes in verbal grade start words.

Morphemes in nominal grade act like enclitics and thus attach to whatever is in front of them. Typically, this is a verbal, but the clitic nature can be seen when a verbal is modified by a relative clause: the nominal then attaches to the final evidence/mood marker.

Morphemes in suffix grade act like, well, suffixes. They attach to the word that they are modifying.


Only mutable morphemes are stressed, lexical suffixes are not. Stress falls onto the first syllable of a stressed morpheme. In multimorphemic words, the last stressed morpheme has primary stress, all preceding ones have secondary stress.

Phonetically, stress is lengthening of the vowel by about 30%, a slightly raised base frequency, and somewhat more tensely pronounced vowels and consonants, e.g., stressed vowels are slightly more off center: unstressed /a i u/ are pronounced [ɐ ɪ ʊ] (/e o/ are never unstressed) while stressed /a e i o u/ are [a e i o u].


The lexicon is stored in a data base and can be accessed online.

Phonetic Distance

In some cateogories of suffixes, morphemes are constructed in such a way that they are

  1. quite different in order to not get mixed up easily,
  2. easy to derive, for example by traversing tables in a regular way.

Achieving (2) is easy: I will just state a rule:

Traversal Rule For each syllable of a morpheme, find that syllable in the corresponding syllable table (in the appendix). Then move one step down and one step right, with wrap-around, in that table. This is how you find the next syllable. For polysyllabic morphemes, do this in parallel for all syllables (probably in different table, though).

OK, that's easy. Be aware that this is a rule with exceptions: sometimes for frequently used morphemes, I decided to go for another, usually shorter, alternative.

Now for making the morphemes easy to distinguish. By the table traversal rule above, we achieve that both consonant and vowel change in each step. We will order the vowels in such a way that single steps change much.

To measure contrast of vowels, the vowels are classified in three dimensions: height (0..2), frontedness (0..2) and roundedness (0..1):

Vowel Height Frontedness Roundedness
a 0 1 0
e 1 2 0
i 2 2 0
o 1 0 1
u 2 0 1

The distance of two phonemes is the sum of the absolute difference in each dimension, i.e. a=(0,1,0) and u=(2,0,1), so the distance is 2+1+1=4. It was tried to find an order this difference is large in each step. For vowels, the order is: a, u, e, o, i.

For the consonants, no particular order was created, because we also need large differences for large numbers of morphemes in the same category. For the quick change, the vowels are used.

The starting point of a morpheme category is chosen by random. The order within the category is the one by which the morphemes are listed in their respective table in the grammar sections.

One Open Lexical Class

Trukva has only one open lexical class, words of which are called substantives. In contrast to that, English has three: verbs, nouns, advectives. Many languages have only verbs and nouns, so that English adjectives end up being either verbs or nouns in such a language.

Some natlangs, like Nootka, have been claimed to have only one open lexical class, but I am not aware that a majority of the linguist community could be convinced yet that it is really true.

Yet, this is a conlang, and I am fond of having only one open lexical class, so Trukva is among those conlangs that have this property.

What does it mean? It basically means that all but a small number of words in the lexicon can function both as the predicate or an argument in an adverbial. I.e., each one can be the first or the second word of an adverbial. In English categories: each one can be either an intransitive predicate (a verb), or an object (a noun). The lexicon does not restrict words to be used as either.

Of course, the world consists of natural nouns (i.e., objects) and of natural verbs (i.e., actions), but this does not mean these need to map to different categories in the grammar. You can coerce them in English from one to the other category, too, usually by appending suffixes, e.g. ‘to eat’ is verb but ‘the eating’ is a noun. Some English words can work as noun and verb without any change, e.g. ‘party’. In Trukva this is basically the case for each word in the lexicon (except a finite number of special words, i.e, particles, which have special grammatical functions).

For example, the word pongi means ‘tool’. It denotes a class of objects. It can be used nominally to refer to such an object, or it can be used verbally to mean ‘to be a tool’ or ‘to be the tool’, the latter of which you would probably translate as ‘using’ or in a serial verb construction simply as ‘with’. Examples:

Ta topsafongi tru.
I see a/the tool.
Ta pongihohkisku prahku.
Chopsticks are tools.
Ta kengilviso pongihohkisku tru.
I eat with chopsticks. (Lit: I eat, chopsticks being the tool.)

These three sentences translate pongi differently in English, but in Trukva, it is always the same thing, the closest translation being ‘being a tool’. In Trukva, it is the same word with the same meaning. Only the relation to other words in the sentence is different, so you may analyse that there are two parts of speech in Trukva, but there is only one open lexical, which can be used freely as both parts of speech.

One typical pair of translations in English for the same word in Trukva is a past perfect participle, e.g. ‘ingested’, and the object of the same verb, e.g. ‘food’. This is the same concept in Trukva: ‘ingested’ is a short translation of a more precise ‘being ingested’. And that's what ‘food’ is. The ‘used’ and ‘tool’ pair in English is of the same kind.

Lexical Aspect

In order to design a lexicon full of words of only one class, one might simply rephrase verbs into nouns (‘to eat’ → ‘eating’) or the other way around (‘the shoe’ → ‘being a shoe’).

Many conlangs, including Trukva, have a more restricted lexicon design. The Trukva lexicon will only contain states. To do this, the notion of state will be very broad. States will include properties (‘blue’) and objects (‘shoe’). It will also allow continuous movement/change to be lexical (i.e. ‘being fast’).

As an example, the translation of ‘to eat’ is based on ‘the state of being satisfied wrt. food’. This is an adjective in English, and you could also call it ‘full (wrt. food)’.

To explain exactly what Trukva views as states, we need to look closer at inherent aspect of verbs. This lexical aspect can be used to classify verbs.

State to know No internal change, or at least not on human scales, or the time is not of interest and disregarded.
Activity to walk The process of constant state change without transition.
Accomplishment to drown The process preparing for a state transition.
Achievement to realise The event of a single state transition.
Semelfactive to knock The event of a double state transition, forth and back.

For Trukva grammar, it is sufficient to distinguish states, processes (activities, accomplishments) and events (achievements, semelfactives).

Further, Trukva classifies objects as states, i.e., being that object is the state. In a similar way, functions are states, i.e., being president is a state. This categorisation is simply a result of not distinguishing verbs, nouns, and adjectiv.

Lexical substantives in Trukva are usually states and sometimes processes, but never events. Allowing processes means that state change per se is not forbidden in a lexicon entry, as long as the word does not focus on the event of the state change. States and processes are not distinguished in the lexical design because there are many cases where the distinction feels very artificial to me (e.g. ‘falling’).

For the English reader, it may be noteworthy that lexical entries are underspecified wrt. whether an action is habitual (‘he eats’) or durative (‘he is eating’). This must either be inferred from context, or can be marked explicitly (and optionally) by aspectual suffixes. Nevertheless, we often gloss substantives with an ‘-ing’ ending here, simply to indicate whether the active or the passive is denoted. This ‘-ing’ gloss does not imply a progressive or continuous aspect.

Grammatical aspect in Trukva also allows the aspect of ‘general truth’ like ‘he often reads’ (habitual aspect) or ‘boys will be boys’ (gnomic aspect). These are counted as states, too, because they have no internal change (at least, it is not the focus of the proposition). As far as I know, there are no lexical entries in English that directly convey such an aspect.

Whether a process has an end or not (cf. telicity) is not important to Trukva lexicon entries, and is not implicitly marked.

Events have to be marked in Trukva, otherwise they cannot be expressed. For example, putting a lexicon entry like ‘to drown’ in the past tense: ‘He drowned’, does not focus in any way on the state transition finalising the process.

Tane klivuhkamufa plo.
He was drowning. / *He drowned.

In this example, the substantive ‘drowning’ in the past tense cannot be translated as ‘he drowned’, because this English translation focuses on the resulting event. The Trukva sentence simply puts the process of drowning into the past, so the translation could be ‘he was drowning’.


The basic structure of an Trukva sentence is the adverbial. Some concepts are naturally adverbial in nature so pressing them into the substantive category felt clumsy to me. For such concepts, Trukva has a closed class of lexical adverbs. These can be used as normal adverbials, but are in fact atomic lexicon entries and not pairs of two substantives.

The most notable grammatical category that is expressed by adverbs in Trukva is aspect. You could view the polarity and mood markers as lexical adverbs, too, but their restricted position at the beginning and end of a sentence resp., would indicate that they are in fact a different category of words.

No Absence

In the Trukva lexicon, there are not entries denoting ‘lack of ...’. This is expressed with degrees.

Some lexicon entries in English may not obviously fall into this category, but at a close look, they do. For example the adjective ‘dry’ is such a lexicon entry. ‘dry’ means ‘for liquid being absent’, i.e., you need the notion of ‘liquid’ and the absence thereof to define what ‘dry’ means. Such lexicon entries do not exist in Trukva.


Scales and the corresponding lexicon design are taken from Qþyng|ài.


I want to mark both ends and beginnings of clauses in order to unambiguously mark clause embedding. To do that, we'll simple put two markers around the clause. There markers will conveniently mark clausal categories so they serve an additional purpose.

My earlier conlang Qþyng|ài uses mandatory evidentials and mood markers. I liked that and will copy it one-to-one to Trukva. These markers will go to the end of the sentence.

At the beginning of each clause, there'll be a polarity marker: either positive or negative polarity will be marked there. The polarity markers will also optionally carry endings to mark other clausal categories like tense and aspect and possibly more.


The following polarity markers exist in Trukva.

Trukva Gloss Meaning
ta true, positive it is true that...
ku false, negative it is false that...
pre either true, either false it is either true or false that...; (Used in yes/no questions (together with interrogative mood) or in citations of questions (often to translate `whether'): `Did he tell you whether he will come?'.)

Evidence and Mood

The following evidence and mood markers exist in Trukva.

Trukva Gloss Mood Category Meaning
prahku fact evidential it is an accepted fact that...
tru perception evidential I saw/smelled/tasted/felt/heard/witnessed/... myself that...
kretsa experience evidential I am experiencing/experience(d) myself that...
plo hearsay evidential they say that..., someone told me that...
klispi belief evidential It is my (religious) belief that...; From acquainted knowledged, conclusions, and debates, I am deeply convinced that...; It is like a fact to me that...
pasta intuition, instinct, internal state evidential Looking inside myself, ...; Although I cannot justify it, I assume, that...
tusku conclusion, inference, assumption evidential I conclude/infer/assume due to facts, knowledge, experience, perception, that...
kekvi conditional conditional if ... (For use in ‘if’ clauses (apodosis), not in the dependent clause (protasis))
pronsa presumptive conditional Let ε > 0; Let's assume that ..., (This is used for large scope conditionals that are given as a main clause, presuming something for several sentences.)
trimbu interrogative epistemic Is it so that...? (for yes/no questions)
plunga hypothetical epistemic Under different circumstances (but not in this universe/situation), ...; it might have been that... (but it is not)
kra subjunctive epistemic ..., that ... (for use in subordinate, dependent clauses)
pluhpu declarative deontic I hereby...
klehti suggestive deontic I suggest to ...
pohka permissive deontic I allow that ...
tipsu deliberative deontic Should I ...? (Expresses a request for instruction. Contrasts with the interrogative, but that mood queries a truth value (and is thus epistemic).)
katsi optative deontic It is my wish that,...; I hope that ...
pruksa precative deontic Would it please happen, that... (used for polite requests)
pra imperative deontic I command that,...; Let it be true that,...

Note that there are differences compared to Qþyng|ài:

  • The ‘consequential’ has been removed. Instead, the consequence after a condition must be given in some other evidence/mood.
  • The ‘opinion’ evidential has been removed, because the evidentials are about giving the source of information, so instead of simply stating an opinion, the source of that opinion shall be stated instead: belief, intuition, conclusion.
  • The interrogative is only for yes/no questions in Trukva, because it queries the truth of the proposition. Other questions query other things, so the proposition itself is marked normally with a another evidence/mood marker.
  • The ‘irrealis’ has been renamed to ‘hypothetical’ for a more specific term.
  • The ‘optative’ has been split into ‘optative’ and ‘precative’.
  • The ‘subjunctive’ has been introduced for dependent clauses where the view is not the speaker's.
  • The ‘presumptive’, the ‘deliberative’ and the ‘permissive’ have been added. They were missing, as there was no clean way to express.
  • The ‘assumption’ evidential was removed because its usage was unclear (to the author of this conlang). Use ‘intuition’ or ‘conclusion’ instead.

These changes may be incorporated into Qþyng|ài in the future, too, because I feel it is general progress wrt. the common design goals of the two languages.

FAQ: Evidentials

  • Why is there no first-hand hearsay vs. second-hand hearsay, or something for a credible source vs. dubious source? To prevent politicians from using the credible source evidential. If the speaker has no first-hand knowledge, then hearsay has to be used.
  • Why is there no quotative vs. hearsay? For the very same reason.
  • Why is the conclusive not split depending on source of conclusion? For the very same reason: if the speaker has only indirect information and concludes, then the conclusive has to be used.
  • Why is there no mirative to mark surprising new facts? Surprise will probably be expressed with a different (optional) category, not with an evidential.
  • What about commissive (I will...) and volitive (I want...) moods? Use the ‘declarative’ for ‘I will’ and ‘optative’ for ‘I want’ moods. This is not distinguished further.
  • What about ability or duty? Whether something is possible or someone is able to do something or someone is obliged to so something can be true or false in itself and shall thus be marked with mood itself: ‘[hearsay] she can sing’ vs. ‘[perception] she can sing’. Therefore, ability and duty are not expressed using mood in Trukva.
  • What about necessitative/obligative modality, especially for first person (I must..., I am forced to...)? I am still thinking about that. Maybe the same holds as for ‘duty’.
  • The evaluation from speakers viewpoint, is this expressed? Nes, yo, perhaps. There's currently the ‘belief’ and ‘intuition’ evidentials, which do some evaluation. These may be replaced or supported by a new category in the future, which expresses evaluation by the speaker, separately from evidence (expressing source of information) and modality (expressing how truth relates to other propositions). (‘[fact][but I doubt that] Grass is green’) But I am still thinking about that, too. E.g. I am not sure how to split ‘belief’ into source and evaluation of truth. Maybe the evaluation can only be used for certain evidentials/modalities so there is no clear line.


Let us begin this section with a category that is a little different from other categories in that it comprises morpheme modifiers, instead of stem modifiers. This means that degrees always modifiy the immediately preceding morpheme, even if that is a suffix, and not the entire structure before that. Trukva marks this kind of morpheme modifying morphemes with a special initial consonant cluster.

Degrees are an essential structure in Trukva as they can be used in virtually any other substantive or suffix or even particles, if that makes sense. What degree does is to modify the amount of particularity of the preceding morpheme: imagine each morpheme expressing a particular idea. If this idea can be sensibly quantified, then degrees can be added to change the quantity. For example, a morpheme meaning ‘mountain’ could be modified to change the degree of the maintain: which is probably mainly its size.

Trukva Gloss Meaning
lpu coarse-low
lti coarse-mid
lka coarse-high
lnista fine-zeroest
lni fine-zero
rpu fine-low
lkatsi fine-mid-low
rti fine-mid
lputsi fine-mid-high
rka fine-high
rnu fine-full
rnusta fine-fullest
rna opposite This swaps around a scale and focuses on the opposite. Note that in contrast to the zero degree, this does not focus on a smaller piece of the scale, but turns around the scale. So while appending ‘zero’ and then ‘full’ degree does not make much sense, appending ‘opposite’ and then ‘full’ degree is perfectly ok. E.g. ‘up’ + ‘opposite’ is ‘down’ while ‘up’ + ‘zero’ is ‘flat on the ground’.
lnangu expected
rkingu surprising

Like many endings, degrees can be stacked if the result makes sense.


In Trukva, counting and grammatical number are very closely related. For small numbers, the singular is the same as saying ‘one thing’. This is similar in many natural languages (think Romance or German for example).

The absolute number suffixes are formed from the monomorphemic numbers 1, 2, 3 plus a morpheme extension. The two parts are analysed as one morpheme and thus lexically. These morpheme extensions can be analysed to mean +3, +6, +9, +12. 16 is an independent word.

Notice how the numbers are substantives so they can be used in isolation.

Trukva Value/Gloss Latin Name Explanation
po 1 Singular
ti 2 Dual
ka 3 Trial
ponda 4
tinda 5
kanda 6
postu 7
tistu 8
kastu 9
poksi 10
tiksi 11
kaksi 12
pomba 13
timba 14
kamba 15
tresti 16
pru several Plural The normal additive plural of a homogeneous number of entities.
tre group Collective 'a group/herd/flock/... of X'. Note that this denotes a single thing again, although it is commonly called a ‘collective plural’.
trahpi similar Similative ‘X and such’: Without further number suffixes, this is understood as a plural. But you may add a singular ending to denote ‘an X or something like that’.
kundi fellow Associative ‘X and his buddies/fellows/associates’: usually used on proper nouns, but not formally restricted to that. This is quite a generic suffix that may denote all kinds of fellows.
prindi family Familiative ‘X and his family’: usually used on proper nouns, but not formally restricted to that. This is more specialised than the associative and only refers to family members.
kensi friend Amicative ‘X and his friends’: usually used on proper nouns, but not formally restricted to that. This is more specialised than the associative and only refers to friends.
pengu horde Representative ‘the people you are a representative of’: usually used to refer to representatives of companies/(peer) groups etc., often in order to make insulting claims about the company impersonal to not upset the representative personally. In the 1st person, it denotes the same kind of distance.

Note that although all of the suffixes can be used as normal substantives (verbals or nominals), their use as a number marker is grammaticalised. It is transparent, but still fulfils a very specific grammatical function each. Furthermore, while normal derivational suffixation is head-first, the grammaticalised usage of ‘fellow’, etc., as number markers is head-last. Further, you can append all of the number suffixes to pronouns.

my people (but not necessarily me)
you (representative) / Lit. ~the company/group you represent
Peter and his family
Peter's family

Appending explicit numbers to a normal plural ending will be understood as an implicit collective plural, such that ‘three several dog’ would be understood as ‘three groups of several dogs each’. Using the collective redundantly is an option, though.

More numbers can be formed by applying degree shaders.

0 Nullar
few Paucal
some Oligal
all Omnial
all ~three groups, each of a few of Peter with his friends

This is just a selection of possibilities. Any degree may be used, and degree may also be applied to the other base numbers (e.g. for ‘not 3 people’ etc.).


Distribution suffixes are morpheme modifiers, i.e., shaders, which can modify several other types of suffixes or morphemes that contain notions of repetition or punctuality. E.g. they can be suffixed to number markers to signify distribution of multiple entities, or to aspect markers to clarify distribution of events in time.

Trukva Gloss Meaning
sequentially, one at a time
all at once

Distributive shaders can be stacked to expressed something like ‘one at a time, regularly distributed’.

Derivation, Compounding, Suffixing

Adding suffixes is another structure in Trukva. Suffixing is used for derivation, compounding, and possessives.

Suffixes are morphemes in suffix grade, which is marked by initial nasal mutation. There is a closed class of derivational suffixes that can only be used in suffix form, never in nominal or verbal form.

Compounding and derivation are usually head-first. In compounding, sometimes two parts form a new head together. For derivation, some suffixes may become the head of the whole structure.

Possessives are head-last, i.e., the possessed is suffixed to the pronoun. Directly, only pronouns may be prefixed to a substantive to indicate possession. Possessives with proper substantives are formed by first suffixing a pronoun to the possessor, then the possessed to the pronoun.

The suffixing modification in Trukva is always strictly left-to-right, i.e., a suffix always modifies the whole structure it is suffixed to.

We have already seen an example for derivation, where a special derivational suffix is used.


Here is an example for compounding, where the second component is used in suffix grade (‘nasal mutation’).

red wine
Suffixing is always restrictive by default in Trukva, not descriptive, i.e., constructions like the above always mean ‘wine that is red’, and never ‘wine, which is red’. See the section about restrictiveness for information on how to change the default. Note how the nominal grade translates differently, as an adverbial:
Ta konduhleksa tru.
The wine is red.

Finally, here is an example for a possessive. If you know Semitic languages, you may notice that this is similar to construct state, since the possessed is marked (by suffix grade), instead of the possessor.

my food
And with a proper substantive as the possessor, we get:
Peter's food


All modification (suffixing, relative clauses) is restrictive by default. Consider the following sentence.

Ta kahkalviFetsu krembihispivreksima plo.
Peter came with his tall son/daughter. / ?Peter came with his son/daughter that is tall.

In Trukva, this sentence implicitly expresses: (1) that Peter has several sons, (2) he came with the tallest one, because the modification ‘tall’ applied to ‘offspring’ restricts the set of individuals, instead of describing the individual. (In English, the relative pronoun ‘that’ conveys a similar restrictiveness, but using it for people may be questionable.)

To convey the meaning: ‘Peter came with his son. The son is tall.’, i.e., to get the descriptive meaning, you have to use a particle to mark the modifier as non-restrictive/descriptive. The marker is a morpheme shader, so technically, it changes the modifyer's restrictiveness to ‘false’ (from the default ‘true’).

Ta kahkalviFetsu krembihispivreksirkima plo.
Peter came with his tall son/daughter. / Peter have with his son/daughter, who is tall.

The particle can be used in relative clauses, too. It is then appended to the subordinate polarity marker:

Ta kahkalviFetsu krembihispimana kreksi kra plo.
?Peter came with his son/daughter that is tall.
Ta kahkalviFetsu krembihispimanarki kreksi kra plo.
Peter came with his son/daughter, who is tall.

As a side node: German also has a particle ‘ja’ that can convey the descriptive meaning. Forcing restrictive meaning is also possible, but needs some restructuring and the use of the determiner ‘derjenige’ (‘that one’).

Peter kam mit seinem Sohn, der groß ist. restrictive/descriptive (ambiguous)
Peter kam mit seinem Sohn, der ja groß ist. descriptive
Peter kam mit demjenigen Sohn, der groß ist. restrictive

Further note that indeed, using both ‘demjenigen’ and ‘ja’ at the same time is ungrammatical in German.



Adverbials consist of two substantives. The first one is in verbal form, the second in nominal form. The nominal is a clitic, so it fuses into one word with whatever word is in front of it. In the simplest case, this is the verbal.

wine being red

Adverbials are the main building blocks of clauses.

Simple Clauses

Simple clauses where introduced at the beginning of this document already. They always begin with a polarity marker, followed by a sequence of adverbial, which are each pairs of a verbal followed by a nominal, and finally followed by a mood marker.

Ta kanduFetsu plandaFansi tusku.
[So I conclude that] Peter likes Mary.

The simplest clauses contain no adverbials at all. This can be translated as ‘yes’ or ‘no’.

Ta pasta.
[My intuition is] Yes.

Clause Sequences

The simplest of all clause compositions is clause serialisation, where two clauses are simply put in a sequence without additional morphological marking. The orthography prescribes a semicolon between clauses.

Although there is no morphological marking that the clauses belong together, they usually do and cannot be uttered in isolation without making no sense. For example, conditional clauses are formed in this way.

Ta tustaso tru; ta kahkaso tusku.
I think, therefore I am.

Clause serialisation is used even if one clause is strictly a subordinate clause of the other (like in the above example), i.e., the evidence/mood marker of the first clause precedes the second clause. One could expect it to follow at the very end, but that is not so. This means that for full clauses combined using serialisation, there is no syntactic difference between a subordinate and a coordinate clause.


Coordination particles like ‘and’ and ‘or’ are a closed class of words that look like substantives. Just like in English, there is only one type of coordation particle for all parts of speech.

Coordination particle in Trukva can coordinate substantive (both in verbal and in nominal function), adverbials, and clauses.

When coordinating two nominals, the coordination particle will fuse with the following word (because nominals are enclitics):

Ta pepsalvifa postihleksa tufu tru.
[I see that] He drinks wine and water.
When coordinating two verbals, the nominal fuses with the last of the coordinated verbals:
Ta tasku tu tetsuFetsu tru.
[I see that] Peter is reading and writing.
When coordinating two clauses or two adverbials, not much interesting happens, except that coordinated clauses require a semicolon before the coordination particle:
Ta taskuFetsu tru; tu ta tetsuFansi tru.
[I see that] Peter is reading and Mary is writing.
Ta kastuhospa tu pongihohkisku kengilviFetsu tru.
Peter eats quickly and with chopsticks.

Nominal Clauses

Nominal clauses can replace nominals. Thus together with a verbal, they act like adverbial clauses in English.

Nominal clauses have the same structure as main clauses. The only morphological difference is that the initial polarity marker is used in nominal form (=fricative mutation).

Pre pispulaFetsu krombafre kahkafa kra trimbu?
Did Peter say whether he'll be there?

Here, the nominal clause is paired with the verbal kromba 'word', indicating that the nominal clause expresses what was said (‘being said’ and ‘word’ are the same concept in Trukva).

Relative Clauses

Relative clauses modify substantives. In Trukva, they are very similar to normal clauses, as they also include polarity and evidence/mood markers.

Relative clauses are externally headed, meaning that the modified head is outside of the subordinate clause, just like in English, where the relative clause follows the head. The relative clauses in Trukva begin with a polarity marker in suffix form (=nasal mutation) and are suffixed to the head they modify (again, order is like in English).

The first component after the polarity marker is different compared to a main clause: it is either a plain verbal or a plain nominal (which will be appended to the polarity marker due to its clitic nature). This single substantive is the modifying component of the modified head. By this construction, no relative pronoun is required, and in fact, Trukva has no such pronouns. Instead of a pronoun, there is a gap.

The following example shows a relative clause that modifies a nominal, and where that substantive is also the nominal in the subordinate clause.

Ta topsahena kengilvi pongihohkisku kra tru.
I see the person who is eating with chopsticks.

In this example, ke 'person' is modified with a relative clause. It is the head of that clause and thus outside of the clause. Inside the relative clause, it is modified with the verbal kengilvi 'eating'.

As you can also see, in the subordinate relative clause, a different mood is used, namely the subjunctive kra. In this case, being used inside a clause with an evidential, is means the subordinate uses the same evidential. So we could have repeated the tru for the same meaning, however, using the subjunctive here is more elegant.

The following is a relative clause modifying a verbal. Here, the nominal paired with the modified verbal is suffixed to the mood marker of the relative clause.

Ta konduna planda pastahleksa tru.
The wine is red in a way that I like it.

This example is harder to translate since English does not have relative clauses modifying adjectives or verbs. The subordinate clause alone translates as ‘I like the red.’ And this red is what the wine is.

As you can see kondu 'red' is used as a verbal in the matrix clause, but as a nominal in the relative clause. Trukva poses no restrictions on whether the head of a relative clause is a verbal or nominal, neither in the matrix nor in the subordinate clause. So we can also form another sentence, where it is a verbal in both clauses.

Ta kondunahospi krahleksa tru.
The wine is red as a rose.

Again, the translation is difficult. The relative clause alone is ‘The/a rose is red’, and the matrix clause is ‘(The) wine is red’, so the wine is the same red that the rose is.

Relative clauses, as all suffixing modification, are restrictive by default in Trukva. See the section about restrictiveness for information on how to change the default.


Polarity, Modality, and Evidentiality

In Trukva, polarity, evidence and modality are marked mandatorily. Polarity is a category that is probably mandatory in every language. However, the set of polarity markers in Trukva may still raise questions, e.g. when to use pre, glossed as ‘either’.

Evidentials and mood markers, OTOH, are not set common, especially not in Indo-European languages, and especially not with such a large set of markers.

For these reasons, this section will give some ideas of how to use these categories.

Evidentials and Mood Markers

In Trukva (just like in my other conlang Qþyng|ài where the concept was first used), evidentials and mood markers are one single category, i.e., two linguistic categories are merged into one set of morphemes.

This was done because for many moods, adding an evidential makes no sense, because the speaker does not express a truth. There are several reasons for this:

  • When the speaker asks a yes/no question (i.e., uses interrogative mood), the clause's truth is queried, so it makes no sense to add information about how the speaker knows that the clause is true: he doesn't.
  • When the speaker uses irrealis moods of any kind, e.g. expresses wishes, commands, suggestions, opinions, etc., the truth of the clause is explicitly not the topic, so again, it makes no sense to add evidentiality information.
  • When the speaker uses dependent clauses, the truth may be relative to the matrix clause (marked in Trukva by the ‘subjunctive’), so again, it is not the speaker who reasons about truth. No evidentiality is useful here, either.

In fact, I noticed that for most moods, using an evidential makes no sense. In Trukva, all evidentials imply ‘indicative’ or ‘evidential’ mood, which is, therefore, not explicitly marked. In the lexicon, evidentials are marked as ‘mood.evid.’ while the other moods are marked as either ‘mood.deo.’ (deontic) or ‘mood.epi.’ (epistemic).

Selected Examples


Yes-no questions are marked with the interrogative mood in Trukva. They query the truth of the proposition from the listener. The neutral yes-no question uses the unbiased ‘either true or false’ polarity.

Pre pustufumeksi trimbu?
Is it raining?

Different polarities can be used to achieve slightly different shades of the question. The ‘true’ and ‘false’ polarities each indicate a slight bias in the speaker wrt. presumed information, e.g., the ‘true’ polarity may signal that the speaker has heard or fears that the proposition is true, and wants to confirm.

Ta pustufumeksi trimbu?
It is raining, isn't it?
Ku pustufumeksi trimbu?
It is not raining, is it?

Questions that are requests for instruction use the deliberative mood in Trukva.

Ta kahkaso tipsu?
Should I be there? / Is my presence necessary?

Questions that query parts of the proposition, but not the truth value itself, are in any normal non-interrogative mood. The mood used is determined in the same way as if there was no query.

Ta postisrehtu pasta?
What do you eat? [and I assume you do eat]
If-Then Clauses

If-then clauses (conditional clauses) in Trukva use the conditional mood in the if-clause (the conditional or apodosis), while some other evidence/mood marker is used in the then-clause (the consequence or protasis). There is no consequential mood in Trukva, so the speaker needs to specify evidence/mood in the protasis just like in an independent sentence.

Ta pustufumeksi kekvi; ku kahkaso pluhpu.
If it is raining, I will not come. (Lit.: If rain is falling, I declare false that I am present.)
Ta pustufumeksi kekvi; ta pulombulvifunsu prahku.
If it is raining, the earth becomes wet. (Lit.: If rain is falling, it is a fact that the ground becomes wet.)
Because-Thus Clauses

Because-thus clauses in Trukva ...

Ta tustaso tru; ta kahkaso tusku.
I think, therefore I am.
The Subjunctive Mood

The subjunctive mood is used in subordinate clauses. Its purpose is to relate the truth value of the subordinate clause to the matrix clause. Because of this, it cannot be used in main clauses.

Subjunctive mood relative clauses express that the same mood is valid for the subordinate clause.

Subjunctive mood nominal clauses relate the truth value to its verbal. E.g. if the verbal implies some source of information (e.g. ‘spoken’), the truth value is relative to that source.



Monosyllabic Morphemes

Since there are not too many of them, this section lists all monosyllabic morphemes.

Suffixes from the same category are tried to be kept at a great distance. To achieve this, the order of vowels and consonants was

a u e o i
p 3p, he, she, it, they; his, her, their water alert; attention; caution one
t true and past 1p, me, I, we; my, our two
k three false person present 2p, you, thou; your, thy, thine
pr imperative several either future
tr non-future perception group, collective
kr subjunctive non-present
pl hearsay
kl state, time, location, spacetime configuration dying
Suffixes (Normal Broad Scope)
a u i
l momentane definite
lv adventive, become, changing towards _
lh abientive, changing away from _ habitual
r generaliser resultative continuous
rs perspective cessative
rv durative experiential
Suffixes (Tight Morpheme Scope)
a u i
ln unison fine zero
lp regular coarse low
lt irregular coarse mid
lk coarse high sequential
rn opposite fine full
rp fine low
rt fine mid
rk fine high descriptive
Inner Syllables (for reference)
a u i
mb mba mbu mbi
nd nda ndu ndi
ng nga ngu ngi
hp hpa hpu hpi
ht hta htu hti
hk hka hku hki
ps psa psu psi
ts tsa tsu tsi
ks ksa ksu ksi
sp spa spu spi
st sta stu sti
sk ska sku ski
kv kva kvu kvi
ns nsa nsu nsi



July 31st, 2011
Comments? Suggestions? Corrections? You can drop me a line.