My philosophy on how to create family auxlangs

[Content note: this essay detailing my opinions on how to go about a certain type of artificial language construction also turned into an incomplete exposition of two of my conlangs-in-progress.  It therefore became quite long.  Many of the invented words I use in these examples were decided on the fly and are nowhere near set in stone.]

I. What I mean by “family auxlangs”

Today I want to switch from my usual topics to conlanging, which has long been a hobby of mine.  My favorite constructed languages to work on are the sort which I imagine to be naturally spoken languages in some alternate universe.  I particularly enjoy creating families of languages which have their share of illogical quirks, partly explained through their own histories, in the way that real languages do; these are often called “artlangs”.  However, I’m also very interested in the idea of “auxlangs” — that is, languages constructed with no pretense at looking natural, but for the purpose of easing international communication.  The most famous auxlang of all time is, of course, Esperanto, but many more have been created in an effort to improve upon its effectiveness.  Creating an world auxlang seems like a rather daunting task, but I’ve always thought it would be fun to create what I call a “family auxlang”: an auxlang intended to facilitate communication between speakers of languages in a particular family (Romance, Germanic, Slavic, Bantu etc).  I couldn’t find a name for this type of conlang.  I did recently run across the term “zonal auxlang”, but that seems to imply slightly different intentions: that the language be learned by everyone in a certain geographic area, regardless of all languages spoken there belong to a particular family.  I would hate to have to invent a zonal auxlang for Scandinavia — imagine having to accomadate characteristics and vocabulary of both North Germanic languages and Finnish!)

In this endeavor, the two language families I’ve focused on are Romance and Germanic, because they are the only families of languages about which I have extensive knowledge.  Creating a Romance auxlang seems somewhat easier, given that the Romance languages are quite close-knit and have a common ancestor which is not prehistoric; we have an extremely clear description of Latin and don’t need to rely on much guesswork there.  The most obvious strategy for creating a Romance auxlang is to concoct a much simpler and easier-to-learn version of Latin, and this is indeed more or less what has been done over and over.  Many Romance auxlangs have been put forth, the most famous of which is perhaps Interlingua.  I always had some disagreements with how things are done in Interlingua, particularly the lack of grammatical gender and verb conjugation, although I understand the thinking behind this lack of grammatical complexity (to be fair, Interlingua might not really fall under my definition of “family auxlang”, since it was designed to be very accessible to speakers of English, which is not a Romance language and lacks grammatical gender and complicated verb conjugation).

I find the art of creating Germanic auxlangs to be more interesting and less clear-cut.  Years and years ago (I can’t remember how many), I ran across a Germanic auxlang for the first time.  I’m pretty sure it was Volkspraak or something with a very similar name.  Anyway, I just remember looking through the proposed grammar and vocabulary and thinking over and over that there were so many decisions in making this language that I disagreed with and would do differently.

(To be clear, whenever I refer to something conlang-related that I “disagree with”, I don’t mean that I’m actually opposed to someone else doing that thing or that my overall attitude is one of looking down upon it.  In the particular realm of auxlangs, it may well be that a decision a creator makes that I “disagree with” is objectively more helpful to their goal, and/or that in the first place their goal is different from my mine would be.  Our disagreement is only in the simple sense that I would do something differently, perhaps mainly for aesthetic reasons.)

Anyway, I remember thinking, back when I first looked over Volkspraak, that if I ever got Project Get Online going and had my own conlang website, I would write a little essay outlining the guidelines I would follow in creating family auxlangs, using my ideas-in-progress for a Romance auxlang and a Germanic auxlang as examples.  As it is, Project Get Online only gradually began to gain steam in the last couple of years, culminating in this blog.  While I still don’t have a website for my conlangs (which have mostly fallen into disrepair anyway), I thought I would write down my ideas on family auxlangs anyway.

I should mention that in preparing to do this, I decided to look up Germanic auxlangs again to see what was new.  I am surprised at how few of them can be found online, given that the internet continues to grow and the hobby of conlanging is booming.  I did find Volkspraak again, except now it’s broken up into a number of separate projects (“dialects”), because apparently it was always a group effort but there are (unsurprisingly) a lot of disagreements over how things should be done.  I was intrigued that they do tend to follow a lot of the guidelines I’m about to suggest, and I don’t think any one of these “dialects” is as “bad” as I remember Volkspraak being when I first discovered it long ago.  The only other Germanic auxlang I found which really intrigued me is Frenkisch, mainly because it’s well-developed with a nicely laid-out grammar which you can access here.  While Frenkisch looks beautiful in its own way, I’m pointing to it as an interesting example of a conlang that definitely does not do things the way I would do them, on several fronts (starting with the relatively obscure vowel sounds, French-inspired orthography, and relatively high level of morphological complexity).

With that out of the way, let me dive into my several basic guiding principles with examples from my not-yet-fully-invented Romance and Germanic auxlangs.  These conlangs are nowhere near fully fleshed out, and every example from them is extremely tentative and only reflects the general ideas floating around in my head.

II. Principles of family auxlanging

1) Avoid complexity if it is not shared by all (most) languages in the family.

At first glance, this is a no-brainer: if part of our goal is to create a language which is easy for people to learn, we should avoid needless complication.  Nearly all auxlangers make at least some substantial effort in this direction even if it isn’t always their top priority.  But it becomes a bit unclear how far this principle should be taken as it begins to compete with the other principles below.

Grammatical gender is a feature not shared by all Germanic languages (English and Afrikaans don’t have it).  So there is no grammatical gender in my Germanic auxlang.  Although Latin had many noun cases, none of the modern Romance languages decline their nouns for case at all (except Romanian, the “dark horse” of the Romance family, and the only one which I don’t know very well), instead using prepositions (something like de “of” to indicate possession, for instance).  So there is no declension for noun cases in my Romance auxlang.

Most of the Germanic languages, meanwhile, have lost their case declensions, except for the possessive case which is usually marked by an -s ending.  As I understand it, even this is falling out use in some of the Low Germanic languages like Dutch, but probably even speakers of those languages are somewhat familiar with it and wouldn’t have too much difficulty with learning a simple rule for possessive endings.  Thus, in my Germanic auxlang, there is no case inflection for nouns except to indicate the possessive case, which I’ll explain under (2) below.

Furthermore, while verbs conjugate (a little) for person and number in many of the modern Germanic languages, some of the Scandinavian languages as well as Afrikaans don’t have it at all.  Moreover, even the Germanic languages which do conjugate their verbs extensively don’t have the habit of dropping subject pronouns, and I don’t think they would particularly miss verb conjugation if it disappeared.  Therefore, I see absolutely no reason to include it in my Germanic auxlang.  Several Scandinavian languages (like Swedish and Danish) tend to end their present tense indicative verbs in -r, so why not end such verbs this way in my conlang as well?  It’s a simple enough rule that it won’t be too difficult for non-Scandinavian-language speakers to learn.  If verken is the infinitive for “work”, then we have

ek verker, du verker, he/se verker, vi verker, ji verker, di verker
— I work, you (sing.) work, he/she works, we work, you (pl.) work, they work

This idea extends to phonetics as well.  Latin had long and short versions of the vowels a, e, i, o, and u.  The modern Romance languages generally lack this distinction, either by having lost it altogether (in the case of long vs. short a) or by transferring the distinction to diphthongs.  As far as I know, Portuguese may have simply lost the long/short contrast of a, e, and o without diphthongization.  Spanish has greatly simplified the vowel system to just /a e i o u/ with no contrast of long/short quality, and Italian and Portuguese have nearly done the same (with a little added subtlety).  Therefore, it makes sense to me that a Romance auxlang should have the simple vowel system /a e i o u/ as in Spanish, with a few diphthongs allowed but kept to a minimum.

With the Germanic languages, things are again more complicated — all of them have a lot more than five vowel phonemes, and they mostly do have some kind of long/short distinction — so I feel that it’s necessary to have long and short forms of a, e, i, o, and u, and allow some common diphthongs such as /ai/ (e.g. “stone” and “home” might be stain and haim.  By the way, there are several ways to indicate vowel length orthographically that are seen in Germanic languages, but for the moment my favorite is through acute accents on the long vowels; at least it’s easy to do this on WordPress, so it’s what I’m sticking with for now: “house” is hús.)

I see no need in either auxlang to include phonemic distinction of consonant length, e.g. Italian’s double consonants.  It just isn’t shared by most of these languages.

2) Include complexity if it is shared by all (most) languages in the family.

This idea is a little more tricky and controversial, and more likely to lead to disagreements between individual conlangers who are trying to follow it.

All languages of both the Romance and Germanic families inflect nouns for number — that is, they distinguish singular from plural.  So the respective family auxlangs should also have this feature.  We just need to make sure that it’s done with a form of suffixing which is easy to implement.  For the Romance auxlang, this is obvious: Spanish, Portuguese, and French all pluralize most of their nouns using -s.  So let’s make sure that all nouns in our Romance auxlang end in vowels (easy to do, given Romance phonology — if we get stuck, we look to Italian, which really does end all of its native nouns in a vowel).  This is a super easy rule to learn, even for speakers of languages like Italian which don’t pluralize their nouns this way.  Thus we have amico “friend”, amicos “friends”, casa “house”, casas “houses”, etc.  So far, nothing surprising.

Now you’ll notice that I said there would be no grammatical gender for my Germanic auxlang without saying what I would do for the Romance one.  That’s because I want it to have grammatical genders (masculine and feminine, not neuter of course, which didn’t survive much beyond classical Latin!)  Since all of the modern Romance languages divide their nouns into masculine and feminine, the principle of “include complexity if shared” applies here, and we should have a easily-identifiable distinction (based on Italian, Spanish, and Portuguese) between masculine and feminine nouns in the conlang.  One will usually be able to tell a noun’s gender by the vowel it ends in, unless that vowel is e; e.g. amico, libro, and cane are masculine, while amica, casa, and voce are feminine.  Moreover, I propose a system of definite (and indefinite) articles which reflect the gender and number of the noun they modify (unlike for the Germanic auxlang, where in view of English and Afrikaans it makes sense to have only one form de), as well as adjective inflections which match the noun inflections.  Thus, lo cane bono, los canes bonos for “the good dog”, “the good dogs” and la casa bona, las casas bonas for “the good house”, “the good houses”.

Now I get the impression that a lot of auxlangers disagree with this, and indeed this is definitely not how things work in prominent auxlangs like Esperanto or Interlingua.  Indeed, a much bigger component of what we might call “the traditional auxlang philosophy” is to keep things as simple as possible in order to facilitate learning.  My reasons for (often) wanting to go against this principle when all languages in a family have a shared feature boil down to two things.  One: if the auxlang is only meant to be learned by speakers of languages in a particular family, then what’s the harm in including a feature already present in their native tongue?  (I admit this is quite a weak argument, but still claim that there’s not much harm if it’s executed properly.)  Two: some things, like uttering a Romance-sounding sentence without article-noun-adjective agreement, just feel wrong.  At least I strongly feel this way, even though I’m not a native speaker of any Romance languages.  I remember squirming at the way the definite article as well as the adjectives in Esperanto end in -a while the nouns all end in -o; it just doesn’t feel right to say la bona libro for “the good book” instead of something like lo bono libro.  And information-theoretically, a rule saying that articles and adjectives must change for gender and number with endings that reflect the noun endings doesn’t add much more complexity.  It would probably be quite difficult for, say, a native Spanish speaker to get used to not changing them, even though such a system might be a tiny bit simpler.

So, for similar reasons, I’m going to go a step further and proclaim that in the Romance auxlang that we’re developing, verbs should be conjugated for person and number.  Without verb conjugation, it just wouldn’t feel like a Romance language, and we wouldn’t be able to drop subject pronouns (which, to be fair, French can’t do, but French is an outlier in this respect).  Full verb conjugation is much trickier to pull off in an auxlang, and I probably wouldn’t attempt it if it weren’t for the fact that at least in Italian, Spanish, and Portuguese, the inflections look very similar.  We end up with something like this (in a similar vein, I’ve included two verb classes whose infinitives end in –are and –ere, a simplification of the three or more classes in actual Romance languages).

amare — “to love”                                                        vendere — “to sell”

amo — I love                                                                   vendo — I sell
amas — you (sing.) love                                              vendes — you (sing.) sell
ama — loves                                                                   vende — sells
amamos — we love                                                       vendemos — we sell
amates — you (pl.) love                                               vendetes — you (pl.) sell
aman — they love                                                         venden — they sell

I would of course try to minimize irregularities in verb inflection, but I’d make exceptions for cases where regular conjugation would feel bizarre to a speaker of any of the Romance languages (e.g. “to be”).  There will of course be other tenses and full conjugations for them as well, but again they can be set up to be much more regular and easier to learn than in any of the real languages.  In particular, I notice that Interlingua lacks a subjunctive, which again I would prefer to include as I believe it would be very strange not to be able to use subjunctive constructions to express certain things in a Romance language.

Now  we return to our Germanic auxlang: how do we inflect nouns here?  Well, it’s not quite so clear what a good plural suffix should be, but I’d go with –e (pronounced as a schwa); most nouns will end in consonants, but this can be suffixed even to vowel stems.  And, even though the possessive case is being phased out of a few modern Germanic languages, it just feels wrong not to include it, as long as it’s done using a very easy rule.  My choice at the moment is –es (or –‘s) for the singular possessive, and –er for the plural possessive.  Thus, we end up with something like de hús for “the house”, but de dór de húses for “the door of the house / the house’s door” and de dóre de húser for “the doors of the houses / the houses’ doors”.

As for inflecting adjectives to agree with nouns, I’m not so sure, but I’m leaning towards probably not.  If we do, we should definitely use the same endings as the nouns have, but I’m not sure this feels any more natural to a Germanic language speaker than no adjective inflection at all.

3) Allow multiple options if they are understandable to all

Sometimes it’s hard to know whether to err on the side of (1) (in the sense of avoiding needless complications) or of (2) (including special features if they are common enough).  For example, all Germanic languages have a long list of strong verbs which change a vowel in the stem rather than using dental suffixes to indicate the past tense or form the past participle (e.g. sing, sang, sung in English).  It would feel unnatural to most Germanic language speakers to use dental suffixes for many of these words — for instance, singed for “sang”.  So following (2), it seems as though we should put a class of strong verbs in our Germanic auxlang, where past tense and past participles are marked with certain vowel changes and must be learned separately.

But there’s a lot of inconsistency among the Germanic languages as to which verbs are strong or weak (or mixed) as well as exactly what the vowel changes are (which is natural, given that vowels are particularly volatile under phonological evolution).  Afrikaans doesn’t even have strong verbs; there’s not enough verb inflection to make for much irregularity at all outside of wees “to be”.  So to include strong verbs in the auxlang doesn’t really seem to be in the spirit of (1): it would imply substantial information for each Germanic language speaker to learn, and the phenomenon isn’t even shared by all the languages in the family.

In cases like this, I say we shouldn’t be afraid to allow two possibilities: regular inflection, and a special strong verb inflection.  After all, in no natural language is it the case that there is never more than one correct grammatical construction for something.  And it should create less difficulty to err on the side of allowing speakers to choose between several options what they’re most comfortable with and requiring the listener to be able to understand more things, rather than forcing speakers to remember how to use a construction they’re not comfortable with.

So in this case I would want to include the option to inflect any (or almost any) verb regularly, say ek singer, ek singde, ek har gesingt for “I sing”, “I sang”, “I have sung”, while also including alternate forms like ek sáng for “I sang” and ek har gesong for “I have sung”.  Here are several more examples.

  • The Romance languages have a lot irregular past participle forms of verbs, which many would feel uncomfortable avoiding and which help to form a lot of Latin-based nouns which are familiar to English speakers as well as Romance language speakers.  Still, it seems like an awful lot of new forms to force everyone to learn, especially when not all Romance languages always agree on which ones should be regular.  So, as with the case of strong Germanic verbs, we should allow the option to either inflect regularly or to use certain natural-looking irregular forms.  For example the past participles of vedere “to see”, facere “to do/make”, trahere “to draw”, corere “to run”, and morere “to die” could be regular vedito, facito, trahito, corito, and morito or the more natural visto, facto, tracto, corso, and morto.
  • All Romance languages have cognate words for “sea” descending from Latin mare.  But some treat it as a masculine noun (Italian, Spanish) while others treat it as a feminine noun (French, Rumanian).  So we should allow it to be either gender in our conlang (lo mare and la mare both correct).
  • There’s some variation among the Romance languages, or even between dialects of a single Romance language (Spanish) as to how to pronounce the consonants c and g before front vowels.  So I would allow anything within the range of /t∫ ∼ ts ∼ s/ for c before e and i, and anything within the range of /ʒ ∼ dʒ ∼ dz/ for g before e and i.  This is done in Interlingua, and is one of the features there that I do agree with.  We can allow similar phonological flexibility in our Germanic auxlang — for instance, maybe final g‘s in words like dag “day” could be pronounced as either /g/ (as in Norweigen) or /x/ (as in Dutch), or post-vocalic h‘s in words like naht “night” could be pronounced as /x/ (as in the non-English western Germanic languages) or indicated only through aspiration or drawing out the previous vowel.
  • Some Romance languages and most Germanic languages have a simple past tense, which is formed by inflecting the main verb without using any auxiliary verb, while all Romance and Germanic languages have a present perfect tense which requires an auxiliary verb.  English has both (“we worked” and “we have worked”), which are used to convey different senses of the time at which the action took place.  Several other languages in these families (German, French, northern Italian) don’t use this tense except in a literary context or to describe events from the distant past, and use only the present perfect in everyday speech.  I propose that we include a construction for both contexts in our family auxlangs, and if someone is uncomfortable with the simple past, they don’t have to use it: vi verkde, vi har geverkt for “we worked”, “we worked / we have worked”, and parlaron, han parlato for “they spoke”, “they spoke / they have spoken”.
  • A variant on that last idea: some Romance languages (like Spanish) use only the auxiliary verb meaning “to have” to form the present perfect, while others (like Italian) only use it for certain types of verbs and for other types use “to be” as the auxiliary verb (e.g. Italian for “she has come” is è venuta “she is come”, while in Spanish it would be ha venido “she has come”).  So let’s allow either construction for such verbs in our Romance auxlang: “she has come” could be é venita or ha venito.  I can tell you, as someone who learned Spanish and then switched to Italian, that it was quite difficult to get used to using “to be” in certain present perfect constructions, but on the other hand quite easy to understand when others used it.

4)  Start with the most recent ancestor language as a default source.

This is probably more controversial than the other things I’ve suggested.

Every auxlanger has to face the question of how to come up with the roots used in the vocabulary of their languages (as well as the actual morphemes used for grammatical inflections, but this is a less overwhelming task).  There are several ways to go about this.  One is to base each root on a word in any particular language which you like the sound of or which seems easy to pronounce or remember (this is kind of what was done in Esperanto).  Another is to consider a set of languages and, for each word, form a root by sort of taking an average of the translations of that word in your set of languages, perhaps weighted towards the languages with the most speakers.  I get the impression that this has often been the tactic of family auxlangers.  It seems like a reasonable idea: the vocabulary should be equally easy for speakers of each language in the family to learn.

But what I like better is to stay as close to the most recent common ancestor language as possible when constructing word roots.  For the Romance family, this is obviously Vulgar Latin, while for the Germanic family, this is some prehistoric ancestor language which many linguists have tried to reconstruct.  The House Carpenter has been kind enough to refer me to a long manuscript by Ringe and Taylor which thoroughly outlines what the vocabulary of this proto-Germanic language might have looked like and what sound changes have taken place in its transition to Old English.  I still haven’t gotten around to studying it, though, so a lot of the words I show in examples here are my wild guesses of what best reflects the hypothetical vocabulary of the proto-Germanic toungue.

But anyway, to give an example in the Romance conlang, I would put the word nocte for “night”.  This is despite the fact that none of the actual Romance languages, as far as I know, have retained the /k/ sound in their word for “night”: Italian, Spanish, Portuguese, French, and Romanian have notte, noche, noite, nuit, and noapte respectively.  If I took an “average” over the set of words for “night” in the major Romance languages, I’m not sure what it would come out to be — perhaps something like note — but it certainly wouldn’t have a c in it.  Yet I’m choosing to stick with the root of Latin noctis.

In fact, there are fairly regular sound changes among the Romance languages that apply to the /kt/ combination in the root, all of which (unfortunately) happen to result in losing the /k/.  There are many other regular sound changes like this.  For instance, intervocalic /t/’s in Latin have softened to /d/ or /δ/ in Spanish and Portuguese and have disappeared altogether in French; e.g. Latin vita “life” has remained the same in Italian but become vida in Spanish and Portuguese and vie in French.  I’m proposing to go with vita even though clearly “on average” these languages don’t preserve the hard /t/ sound (this goes for morphemes like -ato which form past participles).  Similarly, in many Germanic languages (though not the Scandinavian ones), there has been a lot of palatalization of k before front vowels; e.g. /sk/ became a soft /∫/ in Old English.  I prefer to be conservative and keep the non-palatalized k‘s: the words for “church” and “should” might be kirk and skolde.  (Of course, we shouldn’t get ridiculous about it to the point of clearly violating our other rules.  Almost all Germanic languages, with English and Icelandic as exceptions, have shifted the old /þ/ sound to /d/, and in fact many Germanic language speakers seem to have difficulty pronouncing /þ/ at all.  So we go with the common sound change in this case.)

I have several arguments in defense of looking toward the common ancestor, or at least against other algorithms.

One: it often seems tricky to determine the “average” of all the cognates for a particular word: in the above example, we might arrive at note, but it could also be something like noite.  We will never agree on how to determine which combination of sounds is the most equidistant from the words in each language of the family, but a common ancestor language gives us a natural instantiation of equidistance: its word roots are separated from the modern ones by an equal amount of time, even if some members of the family tended to evolve more quickly than others.

Two: if we try to take an “average” of all the cognates to determine each word root, or worse, just randomly determine which cognate to choose our root, we end up with a lot of phonetic (and occasionally even morphological) inconsistency.  If, in considering what our word for “dog” should be, we look at the cognates in Italian, Portuguese, and French (cane, cãe, and chien respectively; note that Spanish doesn’t have a cognate), then we may well end up wanting to use a nasal vowel.  Yet probably many other roots we end up with will end in n without nasalizing the vowel (e.g. from the cognate words for “wool” we might end up with lana).  At least if we stick with drawing vocabulary from a single ancestor language, we’re unlikely to end up with such inconsistencies or any confusing homophones.

Three: One of the main objections to this paradigm might be that choosing roots and morphemes in this way creates an unfair burden for speakers of the less conservative, more innovative languages in the family.  My response is that, while this complaint is more or less valid, the unfairly-distributed difficulties would be almost exactly the same if there were no family auxlang and the speaker of the less conservative language were trying to learn how to communicate in other languages of that family.  Yes, monolingual speakers of more innovative languages like English and French are a bit disadvantaged under my suggested algorithm, but they would have to learn a lot of the same difficult features anyway by studying, let’s say, German and Italian respectively.

Four: Building on that last point, in the case of the Romance family at least (this may apply more subtly for the Germanic case, but it’s less obvious in the absence of a heavily-borrowed-from ancestor language), some of the ancestral roots may be familiar to speakers of each language anyway.  To take the example of nocte (from Latin noctis) for “night”, that /kt/ still shows up in words like Spanish nocturno and French nocturne “nocturnal”.  Similarly, the /t/ in vita “life” still shows up in Spanish, French, and Portuguese vital “vital”, and so on.  Being somewhat familiar with a lot of the ancestor roots via Latin borrowings makes it a lot less difficult to learn to use them all the time, and might even provide the opportunity to make connections between fancy Latin words and the everyday words from one’s native Romance language.

And five: I just strongly prefer drawing from the ancestor language on an aesthetic level, and that’s enough reason for me to make any conlanging decision, even if it’s regarding an international auxiliary language.

III. Our Father…

As is traditional in any good exposition of constructed languages (even the ones that are only halfway constructed), I want to close with a prayer.

Our Father who art in heaven,
Hallowed be Thy name;
Thy kingdom come;
Thy will be done on earth, as it is in heaven.
Give us this day our daily bread.
And forgive us of our debts, as we forgive our debtors.
And lead us not into temptation, but deliver us from evil.
For thine is the kingdom and the power and the glory forever.

The Romance auxlang version:

Nostro Patre che es en lo celo,
Tuo nome sia sanctificato;
Tuo regno vena a Te;
Tuo volontate sia facto en la tera, como é facto en lo celo.
Danos hodje nostro pane djornale.
E perdónanos nostros débitos, como nos perdonamos nostros debitores.
E no nos enducas en temptatjone, ma libéranos de lo male.
Perche lo Tuo é lo regno e la potentja e la glória per sempre.

And the Germanic auxlang version:

Ons Fader dat er in de himel,
Dín nám vese gehailigt;
Dín kuningrík kome;
Dín vil vese gemákt an de érd, sva er gemákt in de himel.
Gife ons dis dag ons daglig brod.
And forgife ons ons skolde, sva vi forgifer dem dat skolde ons.
And ne léde ons in forsóking, már frie ons fra úfel.
For dín er de kuningrík and de máht and de hérlighód in evighód.

5 thoughts on “My philosophy on how to create family auxlangs

  1. Andreu Massana


    I don’t know if you’re going to read this, well, anyway. I have been working for a long time in a Romance standard language, that would, in a parallel world, work as the language of culture and trade among Latin peoples. As of today, my lexicon is at about 7000 words, and I have the skeleton of syntax and morphology more or less worked out.

    I once took a look at Interlingua, and I didn’t like it much, but since it intends to be a “world conlang” rather than a “Romance conlang” I guess this is understandable. Anyway, I basically agree with your criteria, I can’t imagine a Romance language without genders, or verbal inflexions, it would sound totally alien! My proposal is generally based on Spanish, French and Italian; taking into account as well regularity and conservativeness, although not as strongly as you (e.g. I use lenited consonants between vowels, i.e. t > d “vida”, p > b “trobar”, b > v “scriver”). As you can see, the infinitive also drops the -e, as do all words ending in -re, -le, -se, -ne (e.g. “normal”). My goal is to find the “gravitational centre”. This aspect approaches the language towards Portuguese/Spanish, but this is countered with mainly French/Italian syntax and lexicon. Sometimes, I allow myself to choose a word which is only common in one of the three languages because it is the original one, like “dia” and “lavorar”.

    Also, having Catalan as my native language really helps me because it acts like a “bridge” language between the Iberian sphere and French/Italian (e.g. “voler”, “menjar”, “arribar”… but “però”, “més”, “també”, “dia”…).

    Finally, I will translate the Our Father to my language:

    Nostro Padre que es en el celo,
    Sia santificado tuo nome,
    Vegna a nós tuo regno,
    Sia faita tua volontà en la terra, come é faita en el celo,
    Da-nos hoi nostro pan de cada dia,
    E perdona-nos nostros dettos,
    Come nós perdonamos nostros dettores,
    E no nos induzas a tentación,
    Mas libera-nos del Mal,

    Liked by 1 person

    1. Hi Andreu, thanks for commenting! It is very cool to receive a comment like this from someone else interested in Romance auxlangs, especially someone whose native language is Catalan!

      I think I’d consider being flexible on leaving off the vowels at the end of infinitives etc. as you have. I basically only wanted to keep them in order to make phonetic rules more streamlined, but I can see how they may seem a bit unnatural and cumbersome to native speakers of Spanish/French/Portuguese (or any language except Italian, actually). On the other hand, I do feel strongly about being conservative with things like intervocalic voiceless stops (e.g. vita). The idea for me is generally to keep roots of words as phonetically close to Latin as possible, but adjusted for a much simpler phonetic inventory.

      It’s interesting to see just how similar our renditions of the Lord’s Prayer come out. If you ever get your Romance conlang up online, please link me to your website.


  2. Roger / Rodrigo


    I’ve read all your post and I should say that I agree with you 100%. I’ve been reading about pseudo Romance conlangs for quite a while. I have started my own quite a few times but I always end up back on page one. I’m an Argentine (Spanish native), who studied TEFL at College, I also speak, Italian, French, a little Portuguese, some Chinese (Mandarin) and a little bit of Argentine Sign Language.
    It is true that most Romance Auxlangs go away with verb conjugations and final gendered-vowels. I have studied Chinese and tried to mix up the highly analytic form this language has, but then when I have a quick look at what I’ve done I see nothing but a Frankenlanguage.
    I have been dealing with the idea of a “Romance common divisor” which could work as a bridge language for Romance speakers.
    If you are interested in developing a “Standard Romance”, please let me know. I think our starting points are close enough in order to go on with a common start.
    I also do have some personal minor innovations, for this possible bridge language such as:
    -the spelling “qe qi” instead of only-Italian “che chi” and useless-silent-vowel “que qui”.
    -Singular First Person present tense of -ir verbs with -u: dormir >> dormu (that way not interlocking with -ar verbs)
    -Using J/V as semiconsonants. And forcing stress in second to last vowel, except when that would break a dipthong. The final -e could be silent in long words or not, that could be left up to the speaker, but that letter would be kept to indicate stress: informatjOne.

    Liked by 1 person

    1. Hi, Rodrigo, thank you for your comment and your ideas! I very much like the idea of doing an official auxlang project with someone else, and I think for a Romance auxlang it is best to have a native Romance language speaker (which I am not) on the team. Unfortunately I’m way behind on all the online projects I wanted to do this year, so I don’t know when I’d get around to it. But I will keep your comment in mind.

      This may sound stubborn or difficult, but I don’t think I would go with either of your first two suggestions (using Q without U in that way, and ending first-person singular forms in -u). I’d be happy to fully explain my justification but I don’t want to make this comment too long. I agree that using CH seems very Italian-biased and might consider using K instead (which looks very un-Romance but at least it’s not as alien as Q and I have seen it in casual online Spanish conversation). I would also ask if you’d prefer to stick with the spelling for hard G’s that come before front vowels, or if you have a different suggestion.

      I think I mostly like your final suggestion. I think I already suggested using J as a semiconsonant? (I should reread my own post, because I don’t remember exactly what I said!) It also seems reasonable to allow V to optionally be a semiconsonant — the speaker is free to pronounce it as /v/ or /w/ as they wish. I like your idea of making final -e‘s optionally silent. Fixing the stress always on the penultimate vowel worries me because it seems unnatural to pronounce many words, such as for example clasico, with that stress, but maybe at least French speakers would find it normal.


      1. I like the idea of using J as a semiconsonant, but it sits oddly with the examples you gave in the Lord’s Prayer. Specifically, /tj/ and /dj/ palatalized pretty early in every Romance language I know of. My intuitive reading of the relevant words would be /odʒe/ for hodje, /dʒoɾnale/ for djornale, and /temptaʃjone ~ temptasjone/ for temptatjone. I’d suggest spelling these and similar words with ⟨cj⟩ and ⟨gj⟩ and applying the rules for C/G before I and E, allowing variation on whether to pronounce the /j/. This would mean e.g. hogje, gjornale, temptacjone.

        Also related to hodje, what are your plans for the letter H? I assumed it was silent; is that correct? If so, what are your criteria for when to use it?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s