Tirkunan: A Romance Language

Spelling of Tirkunan is completely phonemic. All phonemes are represented by a single grapheme, i.e., there are no digraphs.

Stress and syllable boundaries are mostly implicit and not specifically marked, but the full stress pattern can be derived regularly from the spelling. The apostrophe is used to indicate a vowel or consonant that was dropped due to a elision, and at the end of a word, this apostrophe also marks that stress is retained on the last syllable.

In the following, the phonemic representation is usually not given except in special cases, because the spelling of Tirkunan is phonemic. When pronunciation is discussed, the phonetics are given in IPA.

The following consonant graphemes exist, with the phonemes and possible phones also listed:

Tirkunan allows any sequence of vowels anywhere in a word, and each vowel is counted phonemically as a separate syllable. Some vowel sequences tend to be phonetically realised as diphthongs, as the section on vowels explains. Also, sequences of u + vowel emerging from derivation are usually resolved as v + vowel.

The j is used in a few words only, because usually, there is no phonemic /j/, but the vowel i is often pronounced as [j]. Some words, however, particularly monosyllabic words and loan words, have a phonemic /j/.

Most consonant phonemes have only one phone attached, but n adjusts its point of articulation to the subsequent consonant, if there is one. m also adjust a bit, but is always labial.

Further, g at the end and before consonant is pronounced as a nasal.

The related ng is a single nasal at the end and before consonants, but acts as a sequence n+g intervocalically.

In sequences of nasal + r or l, a voiced homorganic plosive is regular pronounced, but not written. Depending on dialect, it is either voiced [b d ɡ] or unvoiced [p t k] and is shown here as [b̥ d̥ ɡ̊]. The spelling is completely regular, and it is even applied to clusters where the plosive was historically phonemic, and any old phonemic voice distinction of the plosive is lost: it is deleted in modern spelling, so these plosives are completely regularly non-phonemic, but always pronounced. This also holds true for derivation and even across word boundaries, e.g. when -n is prefixed to r- or l-. The following table shows the correspondance between phonemic spelling and phonetic pronunciation:^[4]

Note that a velar nasal only occurs in front of g and k where it is spelled n. The spelling ng is pronounced [ŋɡ] intervocalically, but otherwise, it is generally [ŋ], particularly at ends of words. Following the above rule, it is also [ŋɡ] before liquids.

The language allows the following initial consonants or internal clusters:

Note that this tables only shows the phonetic clusters, i.e., with the epenthetic plosives included. They are not written this way, but without the plosive between nasal and liquid.

Inner clusters are usually simpler, but prefixes may cause more complex initial clusters, e.g., n(i) < in- and s(a) < -ex. The resulting combinations are already considered above – many combinations starting with s, n, m are not found in a lemma in the lexicon.

Phonemically, all the clusters are single syllable, e.g. for determining stress, but phonetically a pre-consonantal n or m is pronounced syllabic.

If a consonant cluster is not feasible, the epenthetic vowel surfaces from the prefixes: this is before nasal for n(i) and before s for s(a). In the case of s, before voiced plosives, it causes the plosive to become phonetically unvoiced, e.g., sba [spa]. More epenthetic consonants additional to what is listed above may emerge for some speakers, e.g. in sr, an epenthetic t may be heard.

The language allows the following final consonant clusters (spelling is shown here):

The final ng represents the phoneme /ŋ/.

To avoid other final consonant clusters, an epenthetic a has emerged historically, and is also written, because it has become phonemic, i.e., modern loan words may violate this principle. Some consonants usually do not occur in stems but are derived from Proto-Romance in different ways, e.g. final d usually appears as r and final b and v usually appear as u.

Tirkunan has the following vowels, which are pronounced slightly differently in different phonological context, determined by length, stress, and closedness of the syllable. The following is the typical average expected pronunciation of vowels in different contexts.

Length is not phonemic.

In unstressed position, only vowels a i u may occur: e is pronounced like i and o is pronounced like u. This is always shown in spelling, i.e., e o never occur in unstressed position in spelling nor in pronunciation.

Some dialects do not distinguish between [a] and [ɐ], but only use [a]. A few dialects that do distinguish the two may have [ə] instead of [ɐ] for unstressed /a/.

In general, unstressed vowels /a i u/, have a laxer pronunciation than the stressed variants, become something like [ɐ ɪ ʊ] for most speakers. This is not universal, however, and some dialect are more like [a i u].

Closedness of the syllable only affects the vowel qualify marginally, if at all. The vowel system has three heights, so /e o/ are usually pronounced roughly half-open, tending more to closed ([e o]) than to open ([ɛ ɔ]), but these specifics are non-phonemic. Also, vowel length is influenced to a small degree by syllable closedness, with a tendency that open syllable vowels are slightly longer.

Tirkunan does not have phonemic diphthongs. Phonemically, adjacent vowels are separate syllables, and there is no theoretical limit of vowel sequences. In most dialects, some vowels merge into phonetic diphthongs, though. The phonetic descriptions in this text show the typical diphthongs that occur.

Spelling does not mark the phonetic syllables intuitively, but is focussed on indicating phonemic syllables and stress. This means that, for example, phones may belong to an adjacent phonological word instead of the one they are written in: clar ou [kla ro͡ʊ̯] 'egg white'.

Vowel length is non-phonemic. Yet vowels have different quantity in pronunciation. In stressed, open syllables, a vowel is long. Otherwise, it is short. The quantity changes only slightly, so we do not indicate it in this document except for stressing the slight difference. The quality, however, changes as the above table shows.

Foreign names are usually spelled in Tirkunan-style phonetic spelling, even if the source language uses the Latin alphabet. This includes names of cites, countries, places, but also foreign names of people.

Some Latin consonant graphemes are not used by Tirkunan natively. Some of these are used in transscribing foreign names. There is an educated pronunciation that is closer to the transscribed language, and a common one that just uses the 'closest' Tirkunan phoneme.

Additional to the plain Latin letters, some letters with diacritics are used in educated transscriptions to more closely render the foreign pronunciation. This is by no means as close as it could get, but only a few sounds are rendered. The following tables gives a list of phonetic and typical educated spelling. It also shows how some sounds are not spelled distinctively. In common spelling, or when the word becomes more native, the diacritics are left out.

Furthermore, Tirkunan uses some digraphs to represent foreign phonemes in loans and names. In common spelling, or when the word becomes more native, the digraphs are simplified as shown, to represent the native, common Tirkunan pronunciation.

The graphme k is sometimes used like c if then transscribed language uses that grapheme. There is no difference in pronunciation. The grapheme x is used also in some recent loans into Tirkunan, but is usually quickly replaced by s as the word becomes native.

The spelling x for /ʃ/ is preferred if the source language uses it that way and in loans into Tirkunan, otherwise, š is used.

Stress in Tirkunan is clearly present, but it is mostly not phonemic. Similarly to Classical Latin, stress can be determined from the structure of the end of a word, and it also influences vowel quality.

Because Tirkunan spelling represents the phonemic word structure quite regularly, stress can be determined directly from spelling. Words are stressed

The apostrophe rule looks like stress needs to be marked, so it would seem phonemic, but really what is marked is a dropped consonant in pronunciation. One could discuss whether this is a process of stress becoming phonemic. Also, foreign loans may have such irregular stress, which might be, again, indicating phonemic stress. But those are still exceptions.

For determining stress, each vowel is counted as a separate syllable, i.e., phonetic diphthongs like ia ie io iu ai au ou still count as two syllables and the word aoi is thress phonemic syllables.

A nasal prefixed to a consonant cluster is phonetically articulated as a syllabic nasal, but phonemically, it is not a separate syllable, i.e., mre [m̩b̥re] is a single phonemic syllable and not stressed on the nasal, but it is two phonetic syllables, with the nasal a separate phonetic syllable.

Exceptionally, if a final consonant is dropped by phonotactic phenomena, or because it is lexicalised that way, a stress on the word final vowel is indicated by an apostrophe. An apostrophe would also be used if the stress stays on the last syllable when adding a prefix, but in Tirkunan, stress may move backwards, so this does not happen.

Finally, because stress does influence vowel quality, if the phonemic structure of a word is shown in this document, stress is marked for convenience, as is the actual vowel quality.

Most endings have been lost compared to Latin, and a has become the least thematic vowel in Tirkunan which drops readily. In older Tirkunan, schwas merged with a and weakened it even more so today, words ending in a are relatively rare, except when they represent a surfaced schwa to avoid an intolerable final consonant (cluster). Some final a does survive, however, and this section gives a brief overview when a thematic a is inherited or a older language schwa (which may have been any vowel before) is retained.

As alread mentioned, final a may represent an older schwa that protects a final vowel or vowel cluster from being exposed to the end of a word, e.g., in apra < aquila 'eagle'.

Secondly, in some loans, a final a is kept, e.g., in candela 'candela (SI unit)', burata 'burrata (cheese)', cinema 'cinema'.

More interesting examples are verbs with stem in v, where a final a may be inherited. Stem final v (from Latin v and also from d) surfaces as u in Tirkunan, and the derived partiple will turn that into -vut. This is different from other consonantal (non-glide) stems, were the participle ending is -at. Depending on the thematic vowel of the Latin verb (a or something else), verb stems that ended in v will either end in a or in u in Tirkunan, so that the participle comes out similar to Latin or other Romance languages, i.e., so that a thematic a is preserved in the participle. E.g., salva 'to save, to heal' with the participle salvat 'saved, healed', in contrast to beu 'to drink' with the participle bivut 'drunken'. Similarly lava 'to wash' and ariva 'to arive', but cau 'to fall' (with u < d) and mou 'to move'. For nouns, no such 'a protection' happened, e.g., compare selu < silva 'forest' for a very similar stem that is not a verb, and ended up in u.

Another final a occurs when a stem final cluster of nasal + plosive refuses to simplify into the simple nasal, usually because the derivations of that stem are more prominent than the stem itself, so the plosive is remembered. Although this is really the manifestation of an older schwa, not an inherited thematic vowel from Latin, this most often happens to verbs where the a was indeed thematic, so maybe, again, this is a reflexion of the participle vowel. This happens more often for a voiceless than a voiced plosive, e.g., tenta 'to tempt' where there is tintar 'temptation' that may have prevented the loss of knowledge that the stem is in -nt. Similarly, there is canta 'to sing' and conta 'to count'. This preservation of a does not always happen, e.g., fun 'to found' is without a and the derived word funamen 'foundation' did not prevent the collapse, but instead got remodelled itself and has no d anymore.

The morphology has greatly simplified compared to Latin. Tirkunan is grammatically isolating with agglutinative derivation.

The verb is usually quite complex morphologically in Romance languages, but it is undeclinable in Tirkunan, except maybe for a past participle, which could also be counted as derivation.

All tenses, aspects, and moods are formed analytically with auxiliaries. The analytical forms that have completely replaced the old system, are still quite typical for Romance. There is only one irregular verb (e 'to be') with a few irregular verb forms, the rest of the verbal system is analytical and regular.

Tirkunan has dropped all grammatical gender distinctions.

There is no trace of Latin case. This is just like in many Romance languages. Tirkunan has thoroughly dropped case from pronouns, too.

The derivational system of Tirkunan is quite productive. Stems never change irregularly when affixes are added. When Latin allows several ways to form a word, often only one has survived and was generalised to be used in many more cases, e.g. endings -icus and -īvus have merged into -iu.

As an example, from the verb am 'to love', we can derive amat 'loved', amatur 'lover', amar 'love'.

Derivation happens by prefixing or suffixing. Homophonous affixes may be different in meaning depending on which kind of stem the are attached to, e.g. ni is 'not' when attached to nouns or adjectives, and is 'in-' when attached to verbs.

Nouns and adjectives are even more weakly distinguished than in Latin, so most noun affixes work for adjectives, too, and vice versa.

Word class changes are a frequent use for affixes, e.g., -(a)t when attached to a verb will make an adjective.

Prefixes that modify verbs can also be used to denominalise nouns, e.g., ni- 'in-' that usually modifes a verb, will create a verb when attached to a noun or adjective. This is a very productive way of denominalisation used more extensively in Tirkunan than in other Romance language, presumably because Tirkunan has lost all endings to distinguish a verb from a noun, in particular because the thematic vowels of verbs have dropped (except for many -īre verbs). E.g., rot 'wheel', a- + rot = arot 'to rotate'.

Tirkunan exhibits most prominently dropping of vowels or consonants when two morphemes are jointed in derivation or between words next to each other in a sentence. Some small and frequent words have alternative forms, depending on the preceding or following word. The following list shows the main rules, roughly from most widespread rule to most restricted/special.

Firstly, final 'a' in words and prefixes and initial 'a' in suffixes is always a very weak vowel that is dropped readily, because it derives from an epenthetic [ə].

A final u in words ending in a two-vowel sequence or in V(l,r)u, regularly becomes v when another vowel (except u) follows, e.g., when a derivational ending is suffixed that does not readily drop its initial vowel.

Adjacent same vowels merge into a single vowel. This is shown in spelling only inside words, but spoken language also exhibits this across words.

Final 'a' is dropped	before vowels	all words and affixes	omra + umla > omr' umla	'humble man'
Final u becomes v	between vowels	many words	deu + -at > divut	'debt'
Final u becomes v	after vowel+(l,r)	many words	selu + -ac > silvac	'savage'
Final u stays u	in other endings	many words	nascu + -i > nascui	'birth'
Initial 'a' is dropped	after vowel	all suffixes	solu + -amen > sulumen	'solution'
Vowel is dropped	after/before same vowel	all affixes	nreu + -us > nrivus	'nervous

The general 'final "a" drops' rule is not shown in the lexicon, because it is universal. Still, as usual, there are exceptions to the rules even if the above table claims they apply to 'all' words. E.g. loan words ending in 'a' may not drop that 'a'. Such exceptions are clarified in the lexicon.

Many small words, particularly those that are often unstressed in a sentence, as well as affixes have additional rules of dropping vowels and consonants. These are not universally applied, but the lexicon lists this with each word and affix. The patterns repeat, i.e., this is not completely arbitrary, but follow the general rules of which consonant clusters and vowel sequences are acceptable – typical those that occur already in stems. Typical phenomena are shown in the following table:

Final vowel is dropped	before vowels	small words and affixes	di + umra > d'umra	'of the shadow'
Initial phoneme is dropped	special circumstances	individual affixes	supril + -itat > supriltat	'superiority'

Special elision rules apply between prepositions that end in a single vowel and articles un and li (but not lur). Here, before un, all such prepositions except a loose their final vowel, and after such prepositions (including a), li looses its vowel. If only consonants remain on either side, no white space is used but an apostrophe is placed between the preposition and the article, otherwise, one whitespace is added. For some of these prepositions, these special rules are actual the normal rules if a vowel follows, shown also in the following table using ap 'water'.

The elision of preposition and article is preferred in spelling over elision with surrounding words: ni'l ap 'in the water' instead of *ni l'ap.

Article usage has become mostly optional in Tirkunan. Therefore, these forms are not that frequent as in other Romance languages.

The apostrophe is used to indicate vowels and consonants that have been dropped due to elision and sandhi, and also to indicate the case of a word whose last syllable is stressed if it ends in a vowel. No apostrophe is used in compounds, i.e., before -.

For multisyllabic words, this apostrophe at the end also indicates that stress is still on the last syllable if a consonant is dropped. There are currently no words where this happens; in older Tirkunan texts, this may be encountered.

A dropped vowel is indicated on the side that causes the drop (i.e., not the side that drops the phoneme): ni'l rin 'in the kingdom', although the i drops on the right side of li. If both sides could trigger the drop, then the right side carries the apostrophe, because the only way this can happen is if the first word is be a preposition: ni'l ap 'in the water' instead of *ni l'ap.

Theoretically, when prefixing a syllable to a mono-syllabic word that ends in a vowel, to indicate that stress stays on the last syllable, an apostrophe could be written at the end of the resulting word: an imagined *di + xe would become *dixe'. This apostrophe would then be removed if a suffix was added: *dixe' + (a)t would become *dixet. The lexicon can mark such stems with the 'stressed' feature, but there are no such stems in the current lexicon, i.e., what actually happens is that stress moves backwards in such situations in Tirkunan (like already in Classical Latin), so no apostrophe is needed, e.g., di + mre becomes dimri [ˈdimb̥rɪ].

The drop of a single vowel after a consonant at the end of a stem does not change stress. However, if a ends in two vowels and is at least trisyllabic, dropping the final vowel might, a similar situation could theoretically occur that would require a stress marker at the end, e.g., in an imagined *axia + axa, the result could theoretically be axi' axa. However, Tirkunan does not have stems that behave like that – the stress would shift backwards in this case, and no apostrophe would be needed. Also, there is no word in the current Tirkunan lexicon that drops a final vowel if the word ends in two vowels.

White space around an apostrophe is handled as follows. Firstly, no additional whitespace is used when adding an apostrophe. If only a single letter remains in the word that dropped a letter, whitespace is removed between an apostrophe on the right side and the following word: l'ap instead of *l' ap. No whitespace is used either between an apostrophe on the left side and a previous monosyllabic word: di'l instead of di 'l. Otherwise, normal whitespace is used.

The following table shows many elision phenomena for prepositions, articles, and verbs, as well as usage of the apostrophe:

In Tirkunan, the distinction between adjectives and nouns is weak. E.g., the ending -atur is often used to form adjectives while -tor was used exclusively for noun formation from verbs in Latin and adjectives used participle endings, e.g., muritur 'mortal (both adj. and noun)'.^[14]

The boundary is further blurred by Tirkunan's tendency to drop the connective di preposition between a noun and a modifying noun phrase, e.g., sup cou bou vs. sup di cou di bou 'oxtail soup'.

There still is a formative ending -an to derive adjectives from nouns, derived from and similar to -ānus and -ālis, but this gets more and more unimportant, especially in spoken language.

So when the lexicon lists a noun or an adjective, using that word as the respective other word class is likely also possible.

Conjunctions often have an equivalent relative adverb or pronoun, and the only difference is that the conjunction is unstressed, which causes a collapse into the a i u vowel system, so some conjunctions (those with vowels e, o) have a different form than the corresponding adverb or pronoun.

Sentences may be non-ambiguous due to the different stressed vs. unstressed form, and some additional patterns may exist for further disambiguation. E.g., the relative clause can be marked by a cil 'that' demonstrative used cataphorically (i.e., refering forward instead of the usual backward (anaphoric)).^[18] Similarly, the conjunction can sometimes be stressed by using a (cataphoric) demonstrative adverb before it: la, sim, etc..^[19] Also note that u is two conjunctions, meaning both 'or' and 'where'.^[20]

The basic digits 0,..,9, as well as 20,..,90 and the larger numbers 100, 200, 1000, 2000, etc., are typical for a Romance language.

Tirkunan has very simply constructed numbers for 11,..,19, by using deca+1,...,9.^[21] The numbers 20, 30, ..., 90 are regular also, using 1,...,9+deca.^[22]

The number un 'one' also acts as the indefinite article.

Longer numbers are stringed together with the largest exponent first. Powers of thousands are grouped together.

Tirkunan has abandoned an older system of numbers that was similar to Latin (with tens using the suffix -inti < -gintā, -ginti), and also abandoned a confusing system of long and short scale (billion vs. milliard), and instead adopted the SI prefixes as names for numbers, which is effectively the short scale, but internationally harmonised. E.g., '299 792 458' is do etu nou deca nou mega set etu nou deca do cilu patru etu cim dec' ot.

For numbers larger than 1e24, scientific base 10 notation is used, using viril for the decimal point, vic for 'times', and putur 'power' for exponentiation. For example '1.417·10³²' is un viril patru un set vic deca putur tre deca do.

Negative numbers are prefixed with min 'minus', e.g., '6.62607016·10^-34' is read as ses viril ses do ses nul set nul un ses vic deca putur min tre deca patru.

Addition 'plus' is pru used as a conjunction, and 'minus' is min. 'Is equal to' or 'equals' is expressed as e pal a. E.g. '5 + 3 = 8' is read as cim pru patru e pal a nou.

The original synthetic ordinal number of Latin have been lost. Ordinal numbers in Tirkunan are formed analytically by using di + cardinal.^[23]

The use of di is optional and often left out, particularly in colloquial speech, because it sounds like counting and meaning roughly the same, basically like 'third person' vs. 'person no. three': gen di tre vs. gen tre.

There is an irregular ordinal number, marked in the lexicon as word type 'adj.num.':

Fractions in Tirkunan are similar to ordinals, but they usually use a connecting pi. I.e., the formation of fractionals is completely analytic. Fractions are also used to express 'percent' pi etu or just centi (from the SI fraction prefix), and 'per mille' pi cilu or mili, and 'ppm' pi mega or micru, etc. Fractions behave syntactically like normal numbers, i.e., as a quantifier, they precede the noun. If it is a single unit, the un is sometimes dropped in colloquial speech.

There is an irregular fraction, marked in the lexicon simply as word type 'num.' like cardinal numerals:

The prefixes are based on SI numeric prefixes. Following the SI guidance, the prefix abbreviations are exactly as in SI. However, the pronunciation is adapted to Tirkunan phonology.

All SI prefixes are also numerals in Tirkunan and can be used in isolation just like any other numeral.

Numeric prefixes are attached with a hyphen because they retain a secondary stress, i.e., they form compounds with the modified unit noun. Sometimes, the hyphen is dropped because of the prefixes also function as numerals. With SI units, the hyphen is meant to be used, because the prefix and the unit form a new unit in the SI system.

The units are based on SI.

Some derived units:

Participles are adjectives that derive from verbs. They can be used as nouns in Tirkunan with the meaning 'the ... one'. The verb e 'to be' has no participles. All other verbs regularly form one participle as follows, which is used in analytic verb forms, too.

The participle ending (a)t suffixes regularly in the way shown: vowels a, i, u are used depending on the ending of the stem. If the stem ends in vowel + u, then it changes to -vu. The stress may shift, which causes the stem vowel to reduce: e becomes i, and o becomes u.

Tirkunan likes to replace short ce relative clauses (typically V or V+Obj) by participle constructions: gen amat bin 'the man/woman/person very much loved' for gen ce e amat bin.

The participle present, the gerund, and the gerundive of Latin have merged into a single suffix -an, but this is pure derivation and does not survive in verb forms in Tirkunan.

The following table lists the verb forms of Tirkunan.

For formal addressing, the optative is prefered over the imperative, as it is felt less direct and less rude. E.g., instead of Vo va mang!, the optative Vo sa mang! is used. Usually, the formal, polite phrases tend to be longer anyway: Vo sa mang pan, pi pracur! 'Please, try the bread.'

A corollary is that the passive imperative cannot be formed with a non-empty subject, because the passive never has an object. Passive imperatives seem to have limited usefulness anyway.

The progressive form has evolved from an earlier form with a, e.g., mi sta a mang. The preposition in this construction was lost and is not used anymore today. In some cases, the old construction may be easier to understand intuitively by speakers of other Romance languages, e.g., Mi sta a e mangat instead of modern Mi sta e mangat for 'I am being eaten.'

More analytical forms can be combined into longer forms. The following is a selection.

The general order of the verb phrase is as follows:

The order may be different if logic is different, e.g., the negation may be after 'other auxiliaries', as in 'I like not to read.' vs. 'I do not like to read.'. This group also allows multiple auxiliaries and nu to be stringed together.

Also, not the full sequence is possible: a tense marker is not used right before e, fu, eri is used, because that already contains the tense. Combining of analytical forms is further limited as the forms get longer. Especially spoken language does not like long sequences of auxiliaries and may leave things underspecified.

Also, the combination sta sta is usually not used. This will probably not hurt, because sta, as a stative verb, cannot really form the progressive (there is no progress while standing).

The interrogative particle is is not part of the syntactic verb phrase, but starts a YN-question and directly follows a fronted question pronoun or adverb.

Note that no participles of e or au are necessary for any verb forms. The irregular verb e and all its irregular derivatives (fu, eri, sa) have no participles.

The following irregular synthetic verb forms exist of the verb e 'to be'. e has no past passive participle.

Note that tense and aspect are not mandatory categories. They are often derived from context. This is especially true in narratives, which generally use the plain verb instead of au+VERB--at. An exceptional verb is fu, the only synthetical past tense, which is used in narratives frequently.

Verbs in Tirkunan can be either transitive (marked v.t. in the lexicon) or intransitive (v.i.), depending on whether they take an object or not. Further, the object may be se in which case the verb is said to be reflexive, marked (v.refl.) in the lexicon if a separate entry is necessary because the meaning is different from the transitive meaning.

Verbs that have no subject are marked additionally with '0s.' in the lexicon. This can happen both for transitive verbs and for intransitive verbs.

Verbs that make verbal phrases, i.e., stand-ins for intransitive verbs, from a following nouns or adjective are called 'copula' and marked 'v.cop.' in the lexicon. Syntactically, they are identical to transitive verbs if a noun follows, but adjectives may also directly follow, which is not the case for transitive verbs. Also, these verbs have no passive voice and cannot be used intransitively, though an intransitve homonymic verb may exist and will have a separate lexicon entry.

There also exist reflexive verbs that act similar to a normal copula, but do not allow nouns to follow, but only adjectives. These are marked as 'v.refl.cop.'.

Finally, auxiliary verbs are marked 'v.aux.' in the lexicon. These verbs can directly precede another verb to modify it.

In many cases, transitive verbs can be used intransitively, by just dropping the object. E.g. scriu 'to write' can be used both as mi scriu lebra 'I write a book' and as mi scriu 'I write (something'). In this general case, the subject is semantically the same in the transitive and intransitive verbs.

Many reflexive verbs simply express that the object is the same as the subject, i.e., the subject acts upon the object. E.g. Lava mi. 'I wash myself.'

Reflexive verbs may also express an effect on the subject of the verb when the subject is semantically not an agent, but the patient of the homonymic transitive, usually causative, verb. E.g. for a transitive verb that implies causation like mou 'to move s.t.', the reflexive verb mou se means 'to move (by itself)', literally 'to move oneself', i.e., the effect of moving works on the subject, although the subject of mou se is not the semantical agent. These verbs are called uncausative verbs, and they are reflexive in Tirkunan quite consistently.

On the other hand, there are pairs of homonymic transtive and intransitive verbs, where again the subject of the intransitive verb corresponds with the object of the transitive verb, i.e., it is not the agent. If a (causative) effect is missing, then these intransitive verbs are usually not reflexive. E.g. flar v.t. 'to sense the smell of' vs. flar v.i. 'to emit smell'. The transitive verb is not causative here, so the intransitve one is not uncausative, so no reflexive se is used here. The reflexive verb, in these cases, corresponds with the transitive verb, e.g., flar se 'to sense the smell of oneself'.

Tirkunan has a few more words that do not fit the categories in the previous section, but should be mentioned here.

Tirkunan is mainly SVO except in special cases discussed below.

In contrast to other Romance languages, pronouns have no special role in ordering, i.e., object and oblique pronouns are put after the verb where also a noun would go: Mi am ti. 'I love you'.

Tirkunan is pro-drop for subjects, so the subject pronoun is often dropped. Object pronouns cannot usually be dropped except maybe in very short answers to questions or other special circumstances. A subject pronoun is more likely to be dropped on a reflexive verb, because then the object pronoun echos the subject.

Some verbs do not have a subject at all, so none surfaces: Prou. 'It rains.'.

Oblique objects are placed after the direct objects: Mi da lebr' a ti. 'I give the book to you.'. For stylistic reasons or for stressing, they can be fronted: A ti mi da lebra! 'To you I gave the book!'. Oblique objects generally behave like adjuncts (see following section).

Auxilaries precede the verb: Mi ro mang 'I want to eat'.

Most modifiers follow the noun: prepositional constructions: mpiri ni Lustani, adjectives: mpiri gran, participles: pan mangat, relative clauses: mpiri ce e ni Lustani, possesives: pau mi, just like prepositional possessives: pau di mi.

In contrast to other Romance languages, adjectives cannot be fronted.

Adverbs to a verb can go at the beginning of the sentence, or after the verb: Aoi mi va a Lustani 'Today, I go to Lustany' or Mi va aoi a Lustani 'I go to Lustany today'.

Adverbs to adjectives and nouns precede the modified phrase: com cat 'like a cat', mrut gran 'very big'.

Relative clauses are introduced with the relative pronoun ce or with a prepositional phrase with that pronoun: di ce.

Relative clauses follow the modified noun phrase: Gen ce beu crivis e tiul mi. 'The person who drinks bear is my aunt/uncle'.

The relative clause itself uses a modified word order in that the reference pronoun ce is fronted and is not mentioned later in the sentence in the position it would normally appear. I.e., Tirkunan uses a gap construction instead of a resumptive pronoun. In the following sentence, the gap is marked with X (which is not pronounced): Gen ce te vei X e tiul mi 'The person you see X is my aunt/uncle'.

Relative clauses may be extrapositioned across adjuncts or even full clauses: Gen ni camra ce beu crivis e tiul mi 'The person in the room who drinks beer is my aunt/uncle', or in some cases, Gen e ni camra ce e tiul mi 'The person who is my aunt/uncle is in the room'. Note that for extrapositioned relative clauses, it is possible that another noun precedes ce to which ce does not refer: ...camra ce..., while the reference is gen. This leads to syntactic ambiguities which are resolved only by semantics. Such constructions are generally only used if the context and the stressing disambiguates the construction clearly.

Conjunctions are used to join to sentences, either on the same level, or in main clause + subordinate clause relationship. Conjunctions start a conjoined clause: Mi ro svol sinu nu potu. 'I want to fly but can't.'.

Questions basically use the same word order as declarative sentences, possible starting with a question word, a pronoun or an adverb. Interrogative mood is marked in Tirkunan with the particle is. This particle starts yes-no questions, and is used after a question pronoun or adverb before the subject and verb.^[32]

In colloquial Tirkunan, is is often contracted to s.

Usually, to use verbs in noun context,Tirkunan requires verbs to be converted formally to nouns, usually be either appending the generic nominaliser -ar, or the passive suffix -at, or the agent suffix -atur. For constructions like 'something to drink' or 'something to eat' or 'something to read', the preposition di plus the plain verb may be used, and is the default way of expressing this. In this case, di cannot be dropped. This way of using the plain verb after a preposition is generally possible and the default prepositional structure with verbs in Tirkunan.

Nominalisation of the verb is also possible, if the nominalisation is semantically sensible. With a noun, the di may then be dropped as usual in noun + di + noun constructions.^[33]

For 'something to drink', there are several possible ways of expressing this, with cium di beu being the most frequent and the default. The same holds for expressions like 'goodbye', which is literally 'to see again', like in many languages:

The construction "the X the Y" is expressed in Tirkunan as pru X, pru Y. Since pru means more and usually a comperative follows, no additional word to indicate the comperative is used. The comperative is separated in Tirkunan so the normal word order can be used.

The construction "the one who X Y" can be expressed in two ways in Tirkunan: cil ce X, Y or cil Y ce X. The second form can be analysed as a relative clause extraposition, and again, it may cause semantic difficulties if Y ends in a noun phrase.

Surnames in Tirkunan are often combinations of a verb and an object or adject.

Often the fused proper noun will show additional elision or fusion or morpheme/word variants, like Stratinlac from strat+ni+lac. The fused proper noun will bear only one stress at the end (last or penultimate syllabel), resulting in all vowels before that to collapse to a i u as for any unstressed syllables, like Numampan from nu+mang+pan.

Other typical surname formations derive from nouns using -us to express 'the one who is like...'.

Further, -ar to express 'the one who believes in/practices/works with/deals with/likes...'. Apart from names, this ending is often used for the dealer of the suffixed thing. Accordingly, any profession or generic agent may be a surname, too. These often end in the same -ar, or in the simple agent ending -atur. These verb based forms carrying a nominaliser cannot incorporate any object or adject -- only the non-suffixed verbs can do that.

Adjectives may be used as is for surnames. Also for adjectives, a -us ending, which may seem redundant, is frequent.

Tirkunan surnames are often very colourful, even borderline ridiculous to speakers of other languages.