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, [j] can be inferred from the spelling i. 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.

A final d before consonant is not pronounced, but does act like a consonant and shifts stress to the last syllable. If a vowel follows in the following words, a [d] is usually pronounced (though not consistently for all speakers and in all sitations).

In some monosyllabic words where it is not needed to indicate stress, particularly very frequent and unstressed words (e.g., i u), the final d is not written unless actually pronounced when a vowel follows.

In sequences of nasal + r or l, a voiced homorganic plosive is regular pronounced, but not written.

The same holds true for s + r or l and l plus r. The spelling is completely regular, and it is even applied to clusters where the plosive was historically phonemic, and even in loan words, 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 or -s ends up right before r- or l-.

Epenthetic plosives are either voiced [b d ɡ] or unvoiced [p t k] depending on phonological context.

Plosives remain even if the r is lost to sound shifts , which is frequently the case at ends of words.^[4] The following table shows the correspondance between phonemic spelling and phonetic pronunciation:^[5]

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(e) < in- and s(e) < -ex. The resulting combinations are already considered above – many combinations starting with s, n, m are not found in a lemma in the lexicon.

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

Phonemically, all the clusters are single syllable, e.g. for determining stress. But phonetically, a pre-consonantal n or m is pronounced syllabic, and a postconsonantal final or interconsonantal r is also syllabic, or alternatively realised with a short schwa, i.e., [r̩] or [ə̆r]. Final l in that position is resolved by suffixing schwa, like any other consonant cluster, and this schwa is shown in spelling as e.

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

The final ng represents the phoneme /ŋ/ (but ngr is /ŋɡr/).

To avoid other final consonant clusters, an epenthetic e pronounced /ə/ has emerged. 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 five phonemic vowels in stressed syllables: /a e i o u/ and four in unstressed syllables: /a i u ə/.

The following table shows how vowels are spelled and how they are pronounced in different phonological context, determined by stress and closedness of the syllable.

Length is not phonemic.

In unstressed position, only vowels /a i u ə/ occur. /ə/, the schwa, occurs mostly at ends of words. If due to stress change (when suffixing a morpheme), a stressed /e/ moves to an unstressed syllable, it becomes /i/ and similarly /o/ becomes /u/. This is always shown in spelling, i.e., o never occurs in unstressed position in spelling and e is used for a /ə/ in unstressed position.

Some dialects do not distinguish between [a] and [ɐ], but only use [a].

A few dialects go in the opposite direction and have [ə] instead of [ɐ] for unstressed /a/, which means these dialects only have only three vowels in unstressed position: /ə i u/, e.g., atar [əˈtar] 'steel'. The dialects sometimes use a modified spelling where a is used for /ə/, e.g., canta 'to sing', pronounced [ˈkantə], instead of the standard cante.

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 quality 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. For simplicity and to ease reading, [j w] are used for showing diphtongs, e.g., /ai iu/ are shown as [aj iw]. Strictly speaking, glides are not parts of diphthongs and /ai iu/ are more like [a͡ɪ̯ i͡ʊ̯].

Due to phonetic diphthongs, even stress shifts occur is colloquial pronunciation, when a falling diphthong can be used by shifting stress to the first vowel of such a diphthong, e.g. in staiu [stɐˈiw], which may become pronounced [ˈstrajʊ] in fast speech.

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 row] 'egg white'.

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 grapheme 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.

In some languages, e.g., with a three vowel system /a i u/, a [ɜ] may be /i/ and may be spelled ë instead of ä, e.g., in Greenlandic or Inuktitut, because it corresponds more closely with the phonology of the source language.

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 [m̩ˈbre] 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. In some phonological contexts, a vowel that is usually dropped is reduced to a schwa instead, and this happens most frequently at the end of words. This schwa is written as e.

Some final a survives, 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.

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

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) is u in Tirkunan. 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 salvad 'saved, healed', in contrast to beu 'to drink' with the participle biud 'drunken'. Similarly lua '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.

A final e (i.e., a schwa) 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. This happens most often to verbs where an a was thematic. Further, this happens more often for a voiceless than a voiced plosive, e.g., tente 'to tempt' where there is tintad 'temptation' that may have prevented the loss of knowledge that the stem is in -nt. Similarly, there is cante 'to sing' and conte 'to count'. This preservation does not always happen, e.g., in fun 'to found' and the derived word funamen 'foundation' did not prevent the collapse, but instead got remodelled itself and has no d anymore. There is also the possibility that the thematic a is conserved, like it is usually the case with other thematic vowels, like in crea 'to create', and this most often found in more recent loans or re-learned verbs from Latin.

Final schwa also occurs if a -r dropped and a voiced plosive would remain, like cumbe from earlier cumr.

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. Derivation happens by prefixing or suffixing.

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

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

Prefixes that modify verbs can also be used to denominalise nouns, e.g., ne- '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.

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.

Final unstressed e /ə/ is always a weak vowel and drops in the presense of another vowel, even across word boundaries.

Final unstressed a (in multi-syllable words) is also a weak vowel that drops if suffixes starting with a vowel are appended, but it does not drop at the end of a word if the next word starts with a vowel.

In suffixes, any initial vowel may drop. Which one drops and which one does not is marked in the lexicon.

A final u in words ending in a two-vowel sequence ai, ei, iu, ou 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. The rules for this u are bit involved and listed separately below.

Adjacent same vowels merge into a single vowel. This is shown in spelling only inside words, but spoken language also exhibits this across words. Note that ee may represent /ˈeə/ if the first of these e happens to be stressed, and in this case, this will not merge into a single vowel.

The following is an overview of the most important rules of elision:

Final /ə/ is dropped	before vowels	all words and affixes	sof + -ad > sufad	'blown'
Final /a/ is dropped	before vowels	within words	leva + -it > livit	'to float'
Final /u/ becomes /v/	between vowels	many words	ou + -um > uvum	'somewhere'
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 /əˌ a/ is dropped	after vowel	all suffixes	solu + -amen > sulumen	'solution'
Vowel is dropped	after/before same vowel	all affixes	flui + -iu > fluiu	'liquid'

The general 'final /əˌ 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 + umbe > d'umbe	'of the shadow'
Initial phoneme is dropped	special circumstances	individual affixes	greu + -itad > griutad	'gravity'

Between prepositions that end in a vowel and article li, there is a special rule for elision: the article drops its final vowel: di + un = d'un, di + li = di'l, cu + li = cu'l.

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.

In special cases for the words is and li, an apostrophe is placed before the word if it is shortened. In this case, there is no white space between the apostrophe and the previous word, and this apostrophe does not indicate a stress shift in the previous word, although the apostrophe is at the end of that word, i.e., an apostrophe with no space on both sides belongs to the next word.

A dropped vowel is indicated on the side that causes the drop (i.e., not the side that drops the phoneme): ni'l ring '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 [ˈdimbrɪ].

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:

Tirkunan word classes are much less clear cut than Latin word classes, because the endings were reduced so much. Adjectives could already be used as nouns in Latin and in Tirkunan, the difference has been reduced much more to the point that many noun-adjective conversion endings have not survived in Tirkunan.

Furthermore, adjectives can function as adverbs in Tirkunan, without any change (or ending). This means that also noun phrases can function as adverbials if semantics permit it, similar to phrases like 'one day'.

Adverbs that have some kind of point of reference often allow to be used as prepositions, to set that point of reference, e.g., avan 'before' can be used as a preposition meaning 'before'. A few prepositions can also be used as conjunctions, e.g., ca 'than', minde 'during, while', or car 'because of, because'. More often, a preposition can be made a conjunction by adding ci, like pui ci 'after'.

Single word prepositions, articles, and conjunctions are always unstressed. If they derive from another word class that is stressed, the form may change, e.g., the adjective prons 'next' may be used as a preposition pruns 'next to'. This stress change is a very common pattern which is shown in spelling in Tirkunan.

There are many more relative, interrogative, and indefinite pronouns, listed in the lexicon, and also in the list of correlatives.

Morphological case and mostly also number have disappeared, also for pronouns. The pronouns are not distinguished by case, i.e., subject and object pronouns, as well as oblique pronouns and possessive pronouns are the same, and none of the pronouns usually reduces, i.e., no vowel or consonant may drop.^[12]

Note how articles li, lur can be interpreted as unstressed versions of the third person pronouns le, lur.

Pronouns retains the number distinction, with the 3rd person equivalent to the definite article, but stressed.

With the gender distinction lost, Tirkunan often uses noun phrases or demonstrative plus noun instead of a pronoun.

Plural can be shown on demonstratives or indefinite pronouns by supplying a modifier that indicates number. This can be either a number like dui, or sul 'alone', tut 'all', or casci 'each'. Other constructions may also indicate number implicitly. This can also be applied to personal pronouns to be more specific than just expressing number.

A relative pronoun can be marked also by prefixing a definite article, also for a possessive relative pronoun:

Reflexive se and reciprocal prapal cannot be in subject position. The impersonal pronoun se if no specific entity is expressed. This pronoun is exclusively used as a subject, complementing the reflexive pronoun.

Pronouns are not mandatory in Tirkunan, in particular subject pronouns, but also object pronouns. This can lead to ambiguities that have to be resolved by context, because the verb does not carry any information about person.

Note that how le behaves differently after a vowel than the article: e.g. di + le (pronoun) is di le while di + li (article) is di'l.

The pronoun tro is used in conversations, often formal, to refer to the organisation, company, or group the interlocutor is part of, often in order to avoid direct addressing of the interlocutor. It derives from tua horda 'your gang'. Since tro is 2nd person, for its reflexive, also tro is used, not se.

The 3rd person pronoun se is strictly reflexive (refering to the subject of the same clause). This pronoun is only used to refer to a 3rd person, i.e., reflexive 1st and 2nd persons are referred to using 1st and 2nd person pronouns, resp. The reflexive pronoun is also used as a possessive.

Possessives follow the noun and are formed with d(i) + noun/pronoun:

Pronouns follows the same pattern.^[13] Also note that di may drop, particularly for lexicalised phrases and also for pronouns, so shorter possessives exist:

Dependent possessive pronouns are either equal to the plain pronoun or are composed of the preposition di plus the plain pronoun. For the relative and interrogative possessive pronoun, the di is not optional.

The independent possessive pronouns are composed of a definite article plus a pronoun, with an optional di preposition in between.

Again, a pronominal possessive relative pronoun is either equal to an attributive relative pronoun, or can be marked by prefixing a definite article:

Preferably, possessive pronouns follow a complex noun phrase they modify if it is sufficiently lexicalised. I.e., if a phrase is lexicalised 'enough', then inserting into the phrase is less likely. E.g. spastur nuc 'nutcracker' and spastur nuc mi 'my nutcracker'. Strictly speaking, this phrase is ambiguous and could be understood as 'the cracker of my nuts', but because it is sufficiently lexicalised, the ambiguity is resolvable. On the other hand, spastur mi nuc or even spastur mi di nuc are not wrong, just not preferred.^[14] Similarly, possessives generally follow adjectives: spastur nuc vril mi 'my green nutcracker'.

As can be seen, the difference many other Romance languages draw between 'person' and 'thing' has been lost in all the pronouns and determiners, e.g. nin and nimbe and nici can all mean 'nothing' as well as 'nobody'.

Number is indicated on the article. There are the following articles:

The articles are optional and are used only rarely, in constrast to many other Romance languages.

Note that there is no plural indefinite article. 'Some' can be expressed using racan.

In some situations, Tirkunan does use the definite articles, e.g.: in pronoun-like phrases from numbers: lur dui 'the two'.

Nouns have only one form and are not inflected.

Since number is marked only on pronouns and articles, noun phrases are often also underspecified for number: gen filic 'a/the/some happy man/men/woman/women/person/people. The previous example also shows that the loss of the thematic gender vowel from Latin is continued in the lexicon, which often has gender-neutral nouns where English would be more specific, e.g. the very frequent gen 'man/woman/person' and many kin terms like tiul 'aunt/uncle' and fil 'son/daughter'.

Attributive adjectives always follow the noun they modify.

The comparative is formed with pru 'more'. The superlative is formed by using mans 'most'. For smaller comparison, min 'less' is used, which has the superlative mimbe. The ending -is < -issimus is used for an absolute superlative of adjectives, and this can also be used on mrut and poc.

These words to modify adjectives can also used to modify sentences (to specify a frequency or excessiveness), nouns, and adverbs, and can be used in isolation, i.e., they are adverbs and indefinite pronouns.

Some examples:

Also note the usage of gran 'big/famous one' and santus 'saint': they are adjectives and follow the noun, unlike other Romance languages.

There are no synthetic comparatives or superlatives in Tirkunan, not even of the very common irregular comparatives 'best' and worst' like in other Romance languages.^[15]

Adverbial that are single words are called 'adverbs'. Adverbials are placed before the word or sentence they modify, or can also follow the verb.

In elliptic sentences without verb, when the adverb logically modifies the whole phase, they usually follow the subject and precede the object: mi ncui 'me, too', mi ncui ti '?I also you.'

Adjectives can be used as adverbs without modification.^[19]

As this list shows, several adverbs contain the preposition a. In regular adverbial phrases, a space is placed after it, but for more idiomatic usages, no space is used, e.g., if the same word is also a preposition (like acas and abas).

There is a strong relationship between prefixes, prepositions, and adverbs. Prepositions and prefixes are often the same, although semantics may shift a bit. Some adverbs are also derived from a prefix, and are themselves prefixed with a preposition a (which stays unstressed) to become an adverb.^[21] Despite being stressed, these adverbs are derived from the unstressed prefix/preposition form, not an original unstressed version (e.g., it is a ri and not *a re). But the same happens when stress shifts to prefixes on monosyllabic verbs that end in a vowel: di mre = dimri not *demri. The following table gives a few examples, with forms in parenthesis if they are derived differently.

Conjunctions are generally unstressed, so all the vowels collapse to unstressed a, i, u. Conjunctions that are derived from other words (e.g., from relative pronouns or adverbs) also lose stress and, therefore, may have different vowels.

Note that ce has merged and means both 'who' and 'what'. This merger has not happened in other Romance language, so it may be confusing.^[22] More confusion may arise because the conjunction ci is derived from ce, but is not equal to it due to being unstressed. In many other Romance languages, the word for the conjunction 'that' is the same as the word that means 'what'. Furthermore, pronunciations [ki] and [ke] may be reversed compared to other Romance languages. Also, 'than' is different from conj. 'that', which may not be the case in other Romance languages.

Prepositions are a closed lexical class in Tirkunan, i.e., no new prepositions can be formed from other words by adding affixes or by reinterpreting them. A few prepositions are listed in the following table.

Note that usage of prepositions is largely lexicalised, i.e., translating does not mean to map prepositions 1:1 according to the this table. One typical source of confusion is the preposition 'on', which has no direct equivalent in Tirkunan. In other Romance languages, a preposition may have been formed, either by reinterpreting sta or by using a derivative of sūrsum. In Tirkunan, usually ni is used. E.g. ni tabe 'on the table', ni lun 'on the moon'.

Tirkunan has a strong tendency to drop the di preposition entirely when expressing a kind of genetive construction or in ordinals, particularly in lexicalised and very common phrases. This phonenomenon contributes to weakening the difference between nouns and adjectives in Tirkunan.

The distinction between adverbs and adjectives is also very weak in Tirkunan, which does not require adverbs to be marked explicitly, e.g. with some reflex of mentem, but instead, adjectives can be used as adverbs directly.^[26] .

Some adverbs may also function as prepositions. Syntactically, in this case, the phrase of preposition + noun could also be interpreted as an adverb + noun used as an adverb. Semantics determine what can be used how, and the lexicon often lists such words as both adverb and preposition. Examples for adverbs that can function also as prepositions are avan, furi, abas, nrat, dest, srul, i.e., adverbs that define a spatial relation. These often require prepositions in other Romance languages, particularly the equivalent of di, but that preposition is easily dropped in Tirkunan, and semantics also does not block their usage as a preposition, so they can be. Examples:

For time/space words, the distinction of motion/direction vs. location is typically not made in the function word itself, but is expressed or implied by context.

The following is a list of a few examples.

Verbs in Tirkunan only have one morphological form, i.e., they are not inflected, but some derivational forms exist, most notably -(a)d, which forms a past passive participle. This is used in analytic verb forms. It is also often used to enrich the lexicon with adjectives (e.g. sicad 'dry') and nouns (e.g. ripud 'receipt').

When an ending is added to a verb, like in any derivation, a possible final -a drops if the ending starts with a vowel. Likewise, an initial a- of endings drops after a possible i or u vowel on the verb stem.

A verb can be nominaliser, i.e., made into nouns or adjectives, in many ways using suffixes. The gerund or noun of process is formed by suffixing -(a)d (which has collapsed with the supine suffix). Verbs are also used as nouns without nominalising ending, which is more frequent today. Mangad e bon (obsolete) or Mang e bon 'Eating is good', Mi am mangad (obsolete) or Mi am mang 'I like to eat' (lit. 'I love eating'), Mi e cansad di mang 'I am tired of eating.'

Plain verb forms carry no tense information, i.e., the plain verb can be used for present, past, or future tense, and the meaning will be inferred from context. There are auxiliary verbs for marking tense. The only exception is the verb e, which expresses present tense and which has irregular past and future tense forms fu and eri, resp.