Brief reviews of various conlangs

These reviews were written for the conlang_learners mailing list, and thus the criteria by which I review the conlangs are not the same as I would use in other contexts. I was commenting primarily on whether they would be suitable for use in the conlang_learners project, so what kind of web presentation they have, how large the corpus is, how easy their orthography is to type on a standard keyboard layout, and so forth were just as important as how interesting or original the languages themselves are, perhaps a bit more so.


Voksigid was an engelang created by Bruce Gilson, Jim Carter, John Ross and others in 1991-1992. It remained unfinished; the vocabulary comprises 256 words and affixes, and the grammar pages, though the syntax and morphology are described thoroughly at a high level of abstraction, lack any example sentences. The phonology is clearly defined except for stress, though the presentation uses multiple natlang examples for each phoneme and its allophones rather than IPA. It uses a purely ASCII phonetic orthography, with 24 letters representing 19 consonants and 5 vowels. The vocabulary comes from a variety of natlangs, mostly European.

Basically, Voksigid is a verb-initial active language with free ordering of the phrases following the verb, each phrase consisting of a preposition, a noun, and potentially one or more adjectives and/or modifying particles. Fairly specific case tags (prepositions) mark the subject as well as the objects and complements. Content roots are always verbs (e.g., "katse" is defined as "to be a cat"); nouns and adjectives are derived from root verbs with case tags used as nominalizing suffixes. I infer that "katselen" would mean "a cat" or "pertaining to (a) cat(s)" (depending on context), although use of specific nominalizers is not discussed. ("len" is defined as "subject of a non-quantifiable statal predicate" and seems to be the most appropriate of the case tags for this kind of nominalization. "tor" is "active subject", i.e. agent I suppose, so "katsetor" would seem to mean "one who is deliberately being a cat".)

Some details of the grammar are underspecified and would have to be (re)created based on the clues in the lexicon and grammar documents. In particular, the definitions of the case tags aren't always specific enough to give the reader a clear idea of how they're used. However, I think it would be quite possible for us to study the Voksigid documents, discuss them, and come up with a more ample reference grammar (with lots of example sentences) and lessons. It would also be necessary to coin more words based on the principles described in the morphology document. Voksigid was clearly not intended to be a lexically minimalist language, given the kinds of concepts to which root words are devoted in the extant lexicon; the small vocabulary is due to the project having ended prematurely.

Figuring out the grammar from the terse documents on the Voksigid website would take some time and effort, but once someone has done that and written an expanded reference grammar and/or lessons, I think Voksigid would be relatively easy to learn. The only hard part would be getting used to the way case tags and nominalizers are used (and perhaps the way relative clauses are formed); there's no inflection, much less irregularity thereof, and the phonology should be easy for anyone who speaks a European language (and many who don't).


Vabungula by Bill Price is an a priori artlang with a relatively simple but not perhaps unnaturally regular phonology and grammar. It's definitely learnable, since the creator is fluent in Vabungula. It's SVO, prepositional, with adjectives preceding nouns and genitive phrases following them.

Phonology and orthography

The native orthography is pleasing. The romanization requires a handful of non-ASCII characters, e and u with circumflex and s and z with hacek. (Acute accents are used to show stress on the phonology page but not apparently elsewhere.) The explanation of the phonology is reasonably clear in spite of not using IPA; several of the phonemes have examples from more than one natlang, and only a few of the vowels' pronunciation is unclear because of the reference to American English without specifying what dialect thereof is intended. The phoneme inventory looks to be a subset of that of English, except for having distinct /e/ and /E/, pure vowels, and a tap rhotic /r/. The phonology/alphabet page doesn't say much if anything about phonotactics, but I haven't noticed so far any difficult clusters.

Stress is irregular and often lexically significant. Many nominalizations of verbs have the same segmental phonemes as the verb but stress a different syllable; the phonology page doesn't make it clear whether moving the stress is a productive derivational pattern.


Nouns are apparently inflected only for number, not for case, definiteness, etc. Noun stems ending in a consonant pluralize with -e, stems ending in a vowel with -n, and there's one irregular plural.

The pronoun system is even more simple and regular than that of Esperanto; three basic singular pronouns, marked as plural with -l. There's also an animate/inanimate distinction in the third person, which may be optional to some degree.

There is one class of modifiers, no distinction being made between adjectives and adverbs. Some adjectives are root words, others derived from nouns or verbs with the suffix -ke.

Comparison works much as in Esperanto, although the comparative particle "ne" is used prepositionally for some spatial relations as well.

Verbs are inflected only for voice (-ke marking passive, o- impersonal). Preposed adverbial particles and auxiliary verbs mark tense, aspect and mood. There are productive derivational processes forming gerunds and participles as well.

There is an extensive derivational morphology, but many of the derivational patterns are irregular and it's not clear how productive they are. The derivational patterns appear to at least have a strong mnemonic value for learning derived words, and for deducing the meaning of some unfamiliar words, even if the derivational morphology is irregular enough that one can't coin many new words on the fly as one does in an agglutinative auxlang such as Esperanto.


The example sentences and phrases are generally helpful, but would be improved by adding interlinear glosses, especially early on. Some things said in later documents should be mentioned earlier, e.g. the fact that adjectives precede nouns should be mentioned in the adjectives section as well as in the word order section. Given the unpredictable lexical stress, it seems that it would be helpful to show stress in all the example words, phrases and sentences, not just in those on the phonology page.

At least one link (the "Roots" section, 4a) is broken.

Vabungula has a wonderfully impressive corpus online — sixteen texts, some of them fairly long (including full translations of two of the Gospels and the Apocalypse). If we decide to learn Vabungula we'll have no shortage of reading matter to practice on.

Qakwan and Ilomi

Qakwan /tʃak.wan/, by Larry Sulky, is a little hard to classify. It's clearly an artlang, with a fictional setting if not (as far as I can tell) a thoroughly developed one, but in spite of its fictional setting it doesn't seem that a high degree of naturalism is one of Larry's goals; it has some engelangy features like self-segregating morphology, part-of-speech marking, and particles to demarcate and classify proper names; and my impression is that it's too regular to be plausible as a naturally evolved language. The numerous apparent or actual cognates to English and other European languages suggest a possible origin as a pidgin or creole, or possibly a language isolate that massively replaced much of its vocabulary from European source languages, but the grammar as a whole doesn't seem entirely consistent with that theory.

The phonology is only slightly more complex than that of Toki Pona, with eleven consonants, six vowels and a few diphthongs. It apparently allows few consonant clusters, though if the set of allowed consonant clusters is set out explicitly anywhere I haven't seen it yet. Stress is not phonemic. The orthography is nearly one-to-one phonemic, exept that "w" can sometimes represent /w/ and sometimes /ə/.

The syntax is generally SVO, with interesting features like verb chains and a conflation of relationship and spatial verbs wtih prepositions. What I've seen so far of the derivational morphology includes an engelangy part-of-speech marking vaguely similar to Esperanto or Ido, though simpler. Adverbial phrases are formed from adjectives and nouns with a prepositional particle "wi". Yes/no questions are marked with a sentence-final particle — if I recall correctly, that's unusual for an SVO language, most such having sentence-initial particles if they don't use word order or intonation or verb morphology to mark questions.

A discussion group for Qakwan exists, but has only two members so far and has seen little traffic since its inception (16 messages in about 3 months).

The corpus is of an impressive size for a conlang as new as Qakwan, but not huge; the Babel Text, a few fables or jokes, and about a hundred or two hundred sample sentences and phrases mostly organized like a travel phrasebook.

Qakwan looks, based on the little study I've had time to devote to it, as though it would be more expressive and precise than Toki Pona but similarly easy and fun to use.

Larry's earlier conlang Ilomi is fairly similar to Qakwan in phonology and grammar, and has a more diverse corpus although possibly not quite as large. For a few months after its first publication it was learned and used by several people, some of whom wrote poetry and wrote or translated short stories and jokes into it. Ilomi was proposed as an auxlang at the time, and didn't have the fictional context that Qakwan has. Qakwan seems to have a more varied sound than Ilomi, allowing a few consonant clusters and diphthongs which Ilomi lacked, while still being very easy to pronounce. I suspect that there are a variety of other differences which I would notice if I had studied Ilomi more recently or Qakwan more deeply.

I don't know whether Larry intends to preserve the Ilomi website elsewhere when Geocities goes down. I have an offline copy of the Ilomi and Konya websites which I will email to anyone who requests them.


Taruven, by Kaleissin (his other handle, on some lists and fora, is "taliesin the storyteller"), is an unusual but still, in my judgement, learnable and speakable language. It has a high-tech fictional setting (with words like "ksānyélla", "Dyson sphere"; lit. star-house), and a history as the lingua franca of a very old multi-species empire. The nominal morphology is really cool, with categories for number, case, locativity and possession (both orthogonal to case and to each other), and quality (a handful of adjectivial affixes, not just augmentative and diminutive but e.g. male, female, bad, good, same, other, wild, tame...). The case system generally marks agent and patient rather than abstract subject and object. Passivation deletes the agent without changing the case marking of the patient.

Verbs have affixing inflection not only for voice, tense and aspect, but for a variety of categories that IE languages tend to mark with auxiliary verbs, adverbs or other paraphrastic constructions; e.g. validationality, evidentiality, mirativity, mood, and intensity. Besides transitive/intransitive, there is a class of "experiencer verbs", whose nominal arguments take different case marking than those of other verbs; the experiencer or agent is marked with the complemented agent case, while the patient or focus is marked with the benefactive case.

Statives, corresponding to adjectives and adverbs in other languages, share some of the grammatical categories of nouns and some of those of verbs, with a few unique categories of their own (e.g. permanence/ephemerality); they can also be turned into verbs or nouns. Statives in an attributive role agree with their head noun in case and number unless they're immediately to its left.

Inflection is agglutinative, with a certain amount of sandhi at morpheme boundaries and at the ends of words; e.g. an aspirated consonant being de-aspirated when not followed by a vowel, a nasal consonant assimilating to the following consonant, or a cluster of vowels simpifying in one of several ways based on phonological and morphological criteria. Besides the sheer number of categories words inflect for, the sandhi rules seem to me the only difficult aspect of the language.

The phonology is moderately complex but not unusually so; it has vowel length distinctions and geminate consonants, and a few moderately rare phonemes like /θ/ and /y/.

The romanized orthography is fairly phonemic, though not all predictable phonological processes are marked. It includes vowels with macron, acute and grave accents, plus ř (r with hacek) and thorn (þ) and edh (ð); nothing that can't be easily typed on a standard U.S. International keyboard layout.

There are a few texts available on the main Taruven website, plus at least four slightly archaic texts on the various conlang relay websites, and Kaleissin sent me a utf-8 text file of other unfinished texts for me to study while preparing for the inverse relay. The available corpus is not nearly as large as that of, say, Vabungula, but I think it's adequate to make Taruven worth considering for this project.


Alurhsa by Tony Harris is an artlang with a fictional setting; conhistorically, it's an auxlang based on a reconstructed protolanguage, but because of the culture of its speakers, more complex and irregular than most or all Terran auxlangs. Tony is fluent in Alurhsa — I've heard him speak in it off-the-cuff. He's been using it for over thirty years.

The phonology is complex, with 25 vowels (including nasal/oral distinction and three degrees of length) and 46 consonants (including distinctive bilabial vs. labiodental fricatives, dental vs. alveolar stops, and postalveolar vs. retroflex fricatives). The romanized writing system requires acute, grave, circumflex and diaresis on vowels, and a few other extended characters for consonants — nothing that isn't found in the Latin-1 character set or easily typed on a U.S. International keyboard. Many of the consonants are represented by digraphs or even trigraphs. The phonology description is generally good, but would benefit by adding IPA symbols to the descriptions using similar phonemes in various natlangs and instructions for how to pronounce the more uncommon sounds. Punctuation is well-defined and not terribly complex or unusual.

Nouns inflect for six cases and two numbers, although plural inflection is only used when no other indication of number appears. There is an archaic, non-productive dual. There are five noun classes or genders, but only pronouns show agreement with their anaphors in gender; adjectives, verbs etc. don't show gender agreement.

Pronouns inflect for person, number, gender, case, and formality; they have gender agreement in first and second person as well as third person, and three degrees of formality in the second person.

Verbs are marked for tense, aspect, voice, and mood, and agree with their subject in person, number and formality. They have a variety of infinitives and participles for different tenses, voices and aspects. The terminology used on the verbs page is a little nonstandard (e.g., referring to an "imperative tense" rather than "imperative mood"); I'm not sure yet whether or to what extent this reflects a real difference between Alurhsa and the familiar languages for which the standard terminology was devised.

Positionals are the equivalents of prepositions and locative/temporal adverbs in other languages; they interact with the case of their object noun in interesting ways.

The Alurhsa articles show definiteness, but don't agree with their heads in case, number, gender etc.

Numbers and other quantifier particles are invariant.

There are a number of productive but not always predictable derivation processes, prefixes and suffixes and compounds.

A few pages on the Alurhsa website give 404 errors or are marked as "under construction", but Tony says he can email us some newer and fuller documentation than what is on the website.

There are a number of texts on the website, both original and translated, concultural and Terran, plus short sample phrases and sentences in the grammar documents, and a page of folk sayings and proverbs.

Tony has also written a bilingual (Alurhsa and English) blog. And there are seven MP3 files of Tony reading seven of the texts; beautiful! The site would be easier to use, however, if the sound files and texts pages were combined, with links to the corresponding MP3 files next to each of the texts Tony has read aloud for us.

Alurhsa is a complex language, perhaps one of the more difficult ones nominated, but there are certainly plenty of materials to practice with. I haven't counted up the total size of files on each site, but I think it's roughly comparable to Vabungula and maybe exceeds most or all of the other artlangs nominated in the size of its publicly available corpus.

Amanda Furrow, who translated into Alurhsa for the second Inverse Relay, reports that Tony sent her an updated copy of the grammar and a more easily searchable version of the lexicon.


Feayran by David Edwards is an artlang with a non-Terran fictional setting. There is a fair amount of concultural information on the website in addition to the reference grammar and lessons; my brief comments here are based mainly on the reference grammar. The name "Feayran" itself is a designation of the language in yet another conlang set in the same conworld, used on the website for brevity; the native name is "Laaoshte Kimoiatuaianoi".

The phonology is pretty well documented, with X-SAMPA notation; the phoneme inventory is about average, with twenty consonants and five vowels. I'm not sure the phonotactics are as well documented as the phoneme inventory, allophony and tone — most languages with syllables as complex as this have more restrictions on what consonants can cluster than the restrictions enumerated here. Feayran is a tonal language — I think perhaps the only one nominated for the project besides Ithkuil/Ilaksh. There are three tones, and moderately complex rules for when particular vowels have particular tones.

The inflection system of Feayran is interesting, unique as far as I know; instead of affixes having an inherent prefixing, suffixing, or infixing behavior, roots instead have "inflection points" which determine where their inflectional affixes go. The same affix can apparently go at the beginning of one root, the middle of another, and the end of a third. (Alex Fink's Sabasasaj has a similar system, but not quite the same.)

"Stance" is a Feayran category roughly equivalent to politeness of verbs and formality of pronouns in Japanese; it manifests as inflection on nouns, qualifiers and verbs. Verbs also inflect for person, number, aspect, mood, mode, voice, and polarity. Some subsets of categories are marked fusionally; for instance, one set of fusional markers show stance of speaker toward the subject, stance of speaker toward listener, person, and number. Nouns, in addition to stance, are marked for ten cases and two numbers; qualifiers are only marked for stance. The case system, though it's basically nominative/accusative, has interesting nooks and crannies and seems well documented.

With all these inflectional categories, word order is fairly free.

There are six texts on the website, totalling around 800 words; most of them have smooth English translations, but only a couple have interlinear glosses. David says he's going to add interlinear glosses for the others. That's a decent corpus for a relatively new conlang, though not nearly as impressive as Vabungula's or Alurhsa's.

The website design is generally very visually pleasing, although I'm not fond of the choice to devote about a fourth of the screen real estate to a marginal background image. The reference grammar is generally very clear, well-organized and well-written; there are many example phrases and sentences, with interlinear glosses (though the glosses aren't formatted as clearly as they could be). The navigational linking is a bit lacking, with some subpages not having links back to the top-level page, but for the most part it's an easy site to find one's way around.

Other conlang review sites:

My home page
My conlang page
Last updated September 2009