Why There Can't be a Perfect Auxlang

March 5, 2024 - Published

In my other article discussing my search for the "perfect" auxlang, I mention that the perfect auxlang is inherently impossible. While I touch on it in that article, I wanted to go into more detail here. I do feel that the concept is pretty obvious, if only because the "perfect" anything is pretty much always impossible, but I think it's a very interesting topic since it reveals a lot of details about how a lot of conlangs are designed. Of course, this article will be tainted with my personal opinion, as someone that's researched a lot of auxlangs (and even dipped toes into multiple natlangs) but is ultimately just a hobbyist.

To start, as always, a definition: international auxiliary languages (auxlangs for short) are languages meant to facilitate worldwide communication between people who don't share a common first language. This definition seems straightforward, but there are two important things that people seem to throw into the definition which aren't actually there: first is that the language does NOT need to be able to communicate everything that possibly exists. If there are important things that the language can't communicate, then there might be an issue with how well it can communicate, but that doesn't mean it's not an auxlang. A badly written book is still a book.

The second thing, slightly related to the first, is that auxlangs aren't supposed to replace any current language. While English (the current de facto auxlang) may be overtaking a lot of countries and possibly overwriting some of their culture, that's obviously not the goal of any auxlang, nor should it be. These two points of mine also highlight a strong personal belief of mine: I actually think that it's better for an auxlang to be an extra, simplified language that anyone can learn, rather than being a complex language that could replace any existing language. That's entirely personal opinion, but when Toki Pona is under consideration for being the next possible big auxlang, it's important to keep in mind that it doesn't contradict the definition of what an auxlang is, even though some might argue against it. Any language that allows people to communicate, no matter how loosely or simplified, is still an auxlang, and that's very valuable, in my opinion.

With that out of the way, let's look into some areas of possible contention for auxlang design.

Vocabulary

Vocabulary is the easiest to understand, so we can start here. If you want a language that everyone in the entire world can use, what words do you use? One solution is to use a priori vocabulary -- in other words, vocabulary that is not based on existing languages. This is the "it's equally hard for everyone" approach, but it's not very common for auxlangs for the obvious reason that it's... equally hard for everyone. Solresol is one of the more well-known a priori auxlangs, so I can pull some vocabulary from it as an example. Do you want to learn a language where "food" is "dolafare" and "April" is "dosolsol"? While the language has its own internal logic, it's a logic that has to be learned independently and has no basis on any other language that exists in the modern day. If one of these gets pushed as an auxlang, it'll be VERY hard to convince people to switch over since it's easy for nobody.

Most auxlangs are a posteriori, meaning that they take vocabulary from existing languages. My opposite example to Solresol would be Elefen, which takes its vocabulary from exactly 5 closely-related languages and tries to find a middle point between all of them. Most Romance language speakers would probably understand the phrase "me ama tu" even if they've never heard of Elefen. However, that phrase would probably mean nothing to people who only speak languages like German, Russian, Chinese, Japanese, Arabic, or any other language that isn't directly related to the Romance languages, which accounts for millions if not billions of people.

The final main solution that a lot of languages are trying to do nowadays is to find an even spread of vocabulary from existing languages. Globasa is one of my favorite examples of this. An example sentence it has in its beginner's guide is "Hin lala sen meli," which means "This song is beautiful." From that sentence alone, it has influences including (but not limited to) Indonesian, Swahili, English, Italian, Mandarin, and Vietnamese, according to their dictionary. However, as you may have noticed... this still has a very similar issue to having an a priori vocabulary! As an English speaker, I would not be able to guess the meaning of that sentence at all, and I doubt Swahili or Mandarin speakers would be able to understand it, either. The obvious plus is that basically every speaker would be able to understand at least 1% of the language's vocabulary, but in the long run that probably wouldn't be significantly better than an a priori language considering the thousands of words necessary to learn. (EDIT: English speakers should be able understand over 50% of Globasa's vocabulary and Chinese speakers over 20%, so this claim might be dubious depending on the conlang)

I think that the last solution is still gaining traction due to ideals, and it makes sense to believe in a universal language having pieces of every language in it. However, I personally still prefer the second solution on paper, since the existence of a language that 5+ language speakers can immediately understand and interface with feels a lot more practical to me. Either way, for the designers, it's all based on what they're willing to sacrifice and what they value the most. I think most people can agree that for the goals of an auxiliary language, there's already no perfect solution when looking at vocabulary alone.

**Before moving onto the next section, I want to highlight something I feel is important, and it's that for auxlang learners, I don't think universality actually matters as much as people think! I hear the term "Eurocentrism" thrown around a lot for languages like Esperanto, but for what little data there is, none of it points to the idea that non-Europeans inherently dislike Euro-related languages, or that they prefer not to learn them. This is why I say that I think it's mainly due to ideals rather than any practicality, but I would also like more data to prove this point, since it's basically just Wikipedia and Wordpress right now.

Phonology

Phonology refers to the sounds of a language, and we already run into some very important issues very quickly. There are actually very few sounds that EVERY language has. Some versions of Arabic don't have /e/ and /o/, Hawaiian is missing a ton of consonants including /s/ and /f/, and French doesn't have /h/. Even though pretty much every auxlang uses the Latin alphabet since it's probably the only objectively correct choice in the entire auxlang design decision department (it's the closest to a universally known writing system), there's also the issue of how every letter will be pronounced.

Most auxlangs will allow some wiggle room for pronunciation. Toki Pona is the biggest example of this where basically every single letter allows for alternate pronunciations since there's so few letters -- "v" doesn't exist, but if you have trouble pronouncing "w", you can just pronounce it as /v/ and still be understood. Even for other auxlangs with more letters, this is commonly done for the 5 vowel system (which is nice because English speakers are used to much more than 5 vowel sounds but also usually have trouble differentiating them from my experience, e.g. Americans trying to naturally pronounce the pure /o/ sound in foreign words). However, even if you have that sorted out, some languages treat letters differently. English pronounces the letter "j" different from basically every other language, German pronounces "s" like the English "z", and Italian will often pronounce "c" like the English "ch." 

Anyway, let's say you make every sound correlate exactly with their IPA equivalents, meaning that it's very easy to point people to correct pronunciation. You narrowed down the phonology to only the sounds that most people can pronounce. You have the 5 vowel system, you don't have H, you removed similar-sounding pairs like T/D, and difficult consonant clusters don't exist. That's Toki Pona! Now we run into a different issue. 

Can you guess what the Toki Pona word "tawa" means? ...Did you guess that it was based on English? It's based on the word "towards," and can also mean "move" or "go." However, it had to be massively simplified in order to fit Toki Pona's phonetic rules. While it's recognizable once you learn it, it makes it a lot harder to include "recognizable vocabulary" as a plus for the language. We see the balancing act again, between having simple phonology versus recognizable vocabulary.

Complexity

I could keep going into more areas of languages, but I wanted to wrap things up in a broader category of complexity. A very interesting side effect I found out from Toki Pona (and possibly my own minimalistic conlang) was that the simpler a language is designed... the harder it usually is to express things! If the grammar is simpler, and if the vocabulary is smaller, then usually it means you need more words to express things that are even slightly complex.

Probably my favorite example from the grammar angle is conjugation in most Romance languages. Most students hate learning conjugation in Spanish class because it requires memorizing multiple different forms of every verb, and I'm no different. However, once you memorize them, it feels so nice to express things faster. The phrase "I ate" in English is only one word in Spanish, "comí." By comparison, a conlang like Toki Pona that has immensely simplified grammar rules, would have to phrase it something like "tenpo pini la mi moku" (more literally, meaning something like "during past time, I eat") since there is no past tense or conjugation.

Smaller vocabulary also means that certain things will be harder to express. Toki Pona doesn't have a word for "anger," so you'll have to express it through other ways, like "pilin utala," or "feeling of struggle/battle." Having to add a few extra words might not matter in a lot of cases, but when specificity matters, or you need to express more complex ideas, it can be a lot more difficult if your language doesn't support it. This is one of the primary reasons people don't think Toki Pona works well as a language for things like news, detailed stories, and official documents, though I will emphasize again that this doesn't make it a bad auxlang, it just may not be suited for everything.

Most auxlangs will try to have to strike a balance -- while it's possible to make a language in 50 words or less (which would make it very easy to memorize the vocabulary), it'll be very ill-suited for much more than basic conversations. Toki Pona has convinced a lot of people of its capabilities with only ~120-140 words, though it's generally accepted that around 1,000 words or more is the bare minimum for a functional auxlang. As for grammar... that's a much more complex topic that can definitely be delved into by someone, but I don't think I currently have the qualifications to do a deep dive into it.

But yeah, that's about the gist. Obviously, I don't mean to bash on any of the conlangs here, especially Toki Pona, which I actually do advocate the use of as an auxlang (for reasons I talk about here and might re-write sometime to update my exact feelings). I bring up these examples to explain how every design decision in a conlang comes with balancing factors, because every decision comes with pros and cons. I love Toki Pona for its simplicity and accessibility, but I think most people can agree that it isn't a perfect auxlang, because the perfect auxlang doesn't exist.

I hope this article helped you learn or think about something new, and explained my criteria for my first article.

• Vecderg