General problems in language design

[Ναθια]

Hi everyone. For the past few years, my wife and I have been struggling to design a family conlang. That is, our idea was to design a language, teach it to our children, and have a language unique to our family. Unfortunately, we're both hard scientists, not linguists, and after years of discussion and thinking we still don't have a sense of what a good conlang looks or sounds like. We've read a lot online, and asked a few questions on the conlang email list, but nothing has seemed particularly helpful from either a practical or aesthetic strategic point of view (e.g. we still don't know whether it would be wise to make a distinction between nominative and accusative, or, whether we ought to expand our set of phonemes).

Can anyone offer either A) some general principles regarding language design, or, B) specific help regarding this language in particular? What we do have so far is a lexicon of about 800 words corresponding roughly to the vocabulary contained in Ogden's Basic English, a five page document outlining grammar, a basic an alphabet and number system with a computer font to match, and a simple computer program I coded to help suggesting new words. I'm happy to post whatever you might need, if anyone is kind enough to help!

Edit: language had a distinction between accusative and dative, not nominative and accusative.

[Celinceithir]

I'd be definitely happy to help you, Ναθια. I've spoken our family language for 25+ years and my own heartlang for 10 or more, so I may have a few insights into what it's like to learn a family conlang, the positives and setbacks. An important consideration is how different you want to make this language from your own mother tongue(s) and its level of complexity. I personally would recommend a dative case, but it greatly depends on what your aims are. In any case, I'd be interested in reading your documents so far and lending a hand.

[Ναθια]

Wonderful, thank you! Could you tell me a bit about your own language?

To answer your questions, I wanted the language to be quite different, but noting the difficulty of learning it I threw in a few features to make it similar to English. Also, I wasn't clear in my first post - it has dative, but currently no accusative distinct from nominative.

Anyway I'll start here, and if you need more, just ask.


Phonemes

The written language aims to be nearly phonetic in its spelling; each of these sounds has one associated written letter.

Vowels: /a/ (leaning towards /æ/), /e/, /ɛ/, /i/, /o/, and /u/ (sounding like /y/ if after an /i/)
Consonants: /f , h , k , j , l , m , n , p , ɾ , s , t , v , w, x , θ /

(Consciously absent are /b , d , g , z/ in order to force the language to have a distinct feel. Unfortunately we've been noticing the many liquids and frictatives create some tongue-twisters and make it tough to distinguish a few words; we're thinking about adding /d/.)


Grammar

SOV (for other beginners, that means "Subject, Object, Verb," for constructions like "I apple eat.")
Forms commands with VO word order
agglutinative (i.e. it makes frequent use of affixes and suffixes)
modifiers before nouns
prepositions
no articles (but the word for "this" sometimes works as a definite article would)
no grammatical gender
Two cases: Nominative vs Dative (forms genitive by turning the noun into a modifier)


Vocabulary & Sample Sentences

Turns nouns into modifiers with suffix -iɛ
Turns nouns into verbs with suffix -ɛv
Pluralizes with prefix s-
Negates with prefix θo-
Pronouns are basically as English
Prepositions as English with a couple exceptions (distinguishes between two forms of "with")
Interrogatives as English but differentiates between "Why-what cause?" and "Why-what justification?"


For ease of reading, sentences are written in English characters rather than IPA; read "th" as θ

Lai thyu lea! (I you like!)
Rith iesaisia hea. (That interesting is.)
Lai itai isil. (I hot touch. / I’m hot.)
Me aiya hea! (It heavy is!)

Lahiol thyuie melia hea? (Where your home is?)
Me thyu avir? (It you want? / Do you want it?)
Lataio thyu rith aeo? (How you that do?)

Riak owe. (Come here.)
Ves thyuie piev. (Eat your food.)
Sil rith lir. (Give that to me.)
Thoisil rith! (Not-touch that!)
Psel lai ma rith. (Help me with this.)

[Testyal]

Phonemes

The written language aims to be nearly phonetic; each of these sounds has one associated letter.

Vowels: /a/ (leaning towards /æ/), /e/, /ɛ/, /i/, /o/, and /u/ (sounding like /y/ if after an /i/)
Consonants: /f , h , k , j , l , m , n , p , ɾ , s , t , v , w, x , θ /

(Consciously absent are /b , d , g , z/ in order to force the language to have a distinct feel. Unfortunately we've been noticing the many liquids and frictatives create some tongue-twisters and make it tough to distinguish a few words; we're thinking about adding /d/.)

Only a spoken language, then? Interesting. I always make a written form because that's where the beauty lies, I feel. As for distinguishing words, do you have any stress or other suprasegmental nonsense? (Basically how words and sentences are separate. You'd never say 'I ate the salad. It wasn't tasty, because I hate salad' as /aɪ̯eɪ̯tðəsælədɪtwɒzn̩teɪ̯stibikʌzaɪ̯heɪ̯tsæləd/, because that's frankly a bunch of garbled ridiculous sounds. You'd need pauses and stressed syllables to distinguish stuff.)

Grammar

SOV (for other beginners, that means "Subject, Object, Verb," for constructions like "I apple eat.")
Forms commands with VO word order
agglutinative (i.e. it makes frequent use of affixes and suffixes)
modifiers before nouns <- I assume by 'modifiers' you mean adjectives?
prepositions
no articles (but the word for "this" sometimes works as a definite article would)
no grammatical gender
Two cases: Nominative vs Dative (forms genitive by turning the noun into a modifier) <- I'd change the Nominative case into Direct.


Vocabulary & Sample Sentences

Turns nouns into modifiers with suffix -iɛ
Turns nouns into verbs with suffix -ɛv <- Like "Dogev" means "To turn into a dog", or "To make a dog"?
Pluralizes with prefix s-
Negates with prefix θo-
Pronouns are basically as English <- How do you mean? In those examples you provided you had 'Lai' to mean 'I' and 'Me' to mean 'It'.
Prepositions as English with a couple exceptions (distinguishes between two forms of "we") <- Same goes here. What do you mean by 'distinguishes between two forms of "we"'? Do prepositions and pronouns have some sort of relationship?
Interrogatives as English but differentiates between "Why-what cause?" and "Why-what justification?" <- I don't quite understand what you mean.


For ease of reading, sentences are written in English characters rather than IPA; read "th" as θ

Lai thyu lea! (I you like!)
Rith iesaisia hea. (That interesting is.)
Lai itai isil. (I hot touch. / I’m hot.)
Me aiya hea! (It heavy is!)

Lahiol thyuie melia hea? (Where your home is?)
Me thyu avir? (It you want? / Do you want it?)
Lataio thyu rith aeo? (How you that do?)

Riak owe. (Come here.)
Ves thyuie piev. (Eat your food.)
Sil rith lir. (Give that to me.)
Thoisil rith! (Not-touch that!)
Psel lai ma rith. (Help me with this.)

This is honestly not that bad for a first language. You seem to know the basics of IPA, and you have a fair bit of knowledge on affixes and the like. Hopefully my unhelpful questions can be of some sort of use to you.

[Celinceithir]

I certainly can share something of my own language. I began documenting it on the internet a few months ago (before that, I had stuck to manuscripts!) but there is some information here on phonology, orthography and grammar, and here's my dictionary; I've uploaded around 3,000 words and counting so far. I wouldn't recommend it as a model (even compared to other, albeit less developed, languages I've made), simply because it is designed only to cater for my personal whims and not for others' ease of use, but this brief lunch break summary might make for somewhat interesting reading.

Spoiler:

Phonology of Standard Perisian Celinese:
Vowels: [ä] <a, à>, [ɛ] <e, è>, [e] <é, ê>, [ø] <i>, <í>, [ɪ] <y>, [ɔ] <o, ò>, [o] <ô>, <u>, [ə] <ë, ı>

Consonants:
p b t d k (c) ɡ
m n
r
f θ (th) ð s χ (ch) h
j (i) w l

(Allophony: r = /ʐ/ in coda; g = /ç/ in coda; c = /χ/ in coda; f = /v/ intervocally, s = /ʃ/ before <é>, <í> or after <ı>; n, t and l = /ɲ/, /c/ and /ʎ/ after /ı/)

Other points:
- Three genders, formally determined by their endings: maiðír (mother) would be neuter, garys (boy) would be feminine, gathal (girl) would be masculine simply because of this.
- There are no articles, so the main way gender comes into play is in the plural endings: ím - masculine, ot - feminine, ain - neuter.
- Most tenses are created periphrastically with auxiliary verbs: fyðí for the future, weðí for the conditional, etc. The present and the simple preterite on the other hand are formed by inflections on the infinitive.
- There are initial word mutations, like in Gaelic. Old Celinese had several mutations, modern Celinese dialects usually have just one, the soft mutation, occuring usually after prepositions, personal pronouns, numbers higher than three and adjectives when placed before a noun. E.g. cafel (/ˈkavɛl/) = room; in my room = na mo g-cafel (/ˈgavɛl/)
- Word order is very flexible, with different orders preferred depending on context. The most common is OVS (ané mairí aıs - you love I)
- Past continual and past habitual are split, as are the present and the past-present continuous.
- No cases other than in the pronouns (this is not the case with some dialects though)
- 30+ productive affixes

I like what I've read so far - it seems like a pleasant and distinctive language with some very nice features; I like the overall feel and look of what I've seen of it so far. I think that you should keep up with your idea of not having any voiced fricatives or plosives bar /v/; if you're thinking of adding more phonemes, I'd suggest that they be voiceless too (ʃ/ʂ and/or ç, perhaps) to retain this aesthetic. As for the accusative, if you strictly keep to your specified word order, it may not be necessary because it will nearly always be implicit. If you don't, I'd suggest an accusative ending for the sake of clarity. You might want to consider an direct object particle instead if the latter is the case.

Testyal: Some v. good questions there. Regarding the pronouns, I may be wrong, but I guess s/he means that the system is much the same as English: 1SG 1PL, 2, 3SG, 3PL except that there is an inclusive and exclusive first personal plural pronoun. I think Nathia's written system in Latin is much the same as the IPA, except [θ] is replaced with th.

[Solarius]

Phonemes

The written language aims to be nearly phonetic; each of these sounds has one associated letter.

Vowels: /a/ (leaning towards /æ/), /e/, /ɛ/, /i/, /o/, and /u/ (sounding like /y/ if after an /i/)
Consonants: /f , h , k , j , l , m , n , p , ɾ , s , t , v , w, x , θ /

(Consciously absent are /b , d , g , z/ in order to force the language to have a distinct feel. Unfortunately we've been noticing the many liquids and frictatives create some tongue-twisters and make it tough to distinguish a few words; we're thinking about adding /d/.)

Don't worry about it. Natural languages like Ayacucho Quechua don't have any of those sounds.

[oxlahun]

Does <me> also serve for "he" and "she"? Are the pronouns plural in the same way as other nouns (<slai> "we", <sthyu>, <sme>)?

Is there some facility for pluralizing words that already begin with /s/ (or some other tricky thing like /ps/ or /x/)?

The romanization of your sample doesn't show any distinction of /e/ and /ɛ/. Are all of those <e>s really /e/?

Your samples are not particularly agglutinative (I see a couple of <thyuie>s and one <tho->). Maybe try some longer sentences so we can get an idea of the way these things string together?

Lai itai isil. (I hot touch. / I’m hot.)
Me thyu avir? (It you want? / Do you want it?)

Having an intransitive <isil> (instead of <hea>) to express a physical feeling is an interesting idea. I kinda like it.

In the second, why isn't it <Thyu me avir?>

Are you using a rising tone (as in English) to signal a question?

Riak owe. (Come here.)
Ves thyuie piev. (Eat your food.)
Sil rith lir. (Give that to me.)
Thoisil rith! (Not-touch that!)
Psel lai ma rith. (Help me with this.)

In all of these, it looks like the verb is first. Is this a grammar rule for expressing the imperative (to distinguish from declarative <Thyu thyuie piev ves> "You eat your food.")?

Is there a predictable production rule for <lai> "I, me" becoming <lir> "to me"?

[Khemehekis]

Pronouns are basically as English
Prepositions as English with a couple exceptions (distinguishes between two forms of "we")

"We" is a pronoun, not a preposition. I believe you got pronouns and prepositions mixed up.

So your language has I, we exclusive, we inclusive, you, he, she, it and they?

[Systemzwang]

In what manner do you intend to teach your children the language?

[Micamo]

Unfortunately, we're both hard scientists, not linguists, and after years of discussion and thinking we still don't have a sense of what a good conlang looks or sounds like. We've read a lot online, and asked a few questions on the conlang email list, but nothing has seemed particularly helpful from either a practical or aesthetic strategic point of view (e.g. we still don't know whether it would be wise to make a distinction between accusative and dative, or, whether we ought to expand our set of phonemes).

Well, it looks like you've already avoided (or conquered) Engineer's Syndrome, so you're past the first step. Unfortunately, there are a few concerns that need to be addressed here:

1. People's aesthetic tastes between languages differ greatly. We can help you to reach your goals, but exactly what your goals are can only be chosen by you and your wife. So what are they? If you don't quite know what they are yet, I suggest you don't try to discover them by yourself, a priori; The number of possible languages is so large trying to find what you like independently by simply exploring the conceptual space is like finding a needle in a haystack floating through intergalactic space. A much better way is to explore the spaces that natural languages already have: Read up on as many natlangs as you possibly can, from as wide an area as possible, things strange and new like Warlpiri, Ekari, Tagalog, and Hixkaryana. See what sticks and what doesn't. It's by doing this that I discovered my love of polysynthetic structures from the indigenous languages of the US and Canada, for example Pawnee "iraceruksta:taweku:ki" meaning "The customs we had."

2. What I personally think makes an objectively good conlang (independent of aesthetic preference for one structure over another) conflicts pretty heavily with your goal of it being a language you and your wife can learn and teach to your children. I think a conlang grammar should read out like a math proof: Quite complex, but all the pieces fitting together perfectly. If you don't need ten pages to explain how your case system works, then your language is too simple for my liking. However, this will make your conlang difficult to learn for the same reasons natlangs are difficult to learn. Are you willing to sacrifice interestingness for learnability, or vice-versa?

[Visinoid]

Hi everyone. For the past few years, my wife and I have been struggling to design a family conlang. That is, our idea was to design a language, teach it to our children, and have a language unique to our family.

Lolwhut?

I hope you're not taking this seriously, but rather as a game... Maybe your children won't give a damn about that lang.

[Ναθια]

Wow. Thanks for the thoughtful responses; I'll try to reply to each of you.

Phonemes

Only a spoken language, then?

I believe I must be using term "phonemes" incorrectly. The conlang has a written form, with a font I designed to let us type it out on computers.

As for distinguishing words, do you have any stress or other suprasegmental nonsense?

Usually words stress the first, third, and fifth syllables. Some words do have stress on syllable #2, however, and words with prefixes usually stress first, but not third. Typically pitch changes at the end of a sentence. This wasn't planned, but came about with speaking as it seemed to sound right.

Grammar
modifiers before nouns <- I assume by 'modifiers' you mean adjectives?

There's no distinction between adverbs and adjectives. Maybe there's a term I should be using rather than "modifier" to describe the class of words that modify other words?

Two cases: Nominative vs Dative (forms genitive by turning the noun into a modifier) <- I'd change the Nominative case into Direct.

OK; Direct vs Dative, then?

Vocabulary & Sample Sentences

Turns nouns into modifiers with suffix -iɛ
Turns nouns into verbs with suffix -ɛv <- Like "Dogev" means "To turn into a dog", or "To make a dog"?

"He dogged around the corner. He's dogging his food." What exactly the verb "to dog" means is vague, yes, but I wouldn't typically say that. The most common usage I can think of is when I tell my son to be quiet by turning "silence" into a verb.

Pluralizes with prefix s-
Negates with prefix θo-
Pronouns are basically as English <- How do you mean? In those examples you provided you had 'Lai' to mean 'I' and 'Me' to mean 'It'.

Pronouns:

Lai - I, me
Lir - to me
Me - it, it
Mir - to it
Ne - she, her
Nir - to her
Moi - he, him
Mir - to him
Thyu - you, you
Thir - to you

Add the plural prefix to make any of these into "them." We're playing around with a distinct male pronoun set; before we just used "me" for he or it, but it felt difficult to use. Even right now you may note that the male and neuter pronouns aren't distinct in the dative.

Prepositions as English with a couple exceptions (distinguishes between two forms of "we") <- Same goes here. What do you mean by 'distinguishes between two forms of "we"'? Do prepositions and pronouns have some sort of relationship?

Yikes! Sorry, I meant it distinguishes between two forms of with.

"With my hands" -> Ma laie sav (using my hands)
"With my friends" -> Le laie slealitek (alongside my friends)

Interrogatives as English but differentiates between "Why-what cause?" and "Why-what justification?" <- I don't quite understand what you mean.

Two forms of why

"The grass is green on Fred's lawn."
"Why?"
Answer 1: "Because it has chlorophyll."
Answer 2: "Because I visited yesterday and I don't think it had enough time to turn brown."

This is honestly not that bad for a first language. You seem to know the basics of IPA, and you have a fair bit of knowledge on affixes and the like. Hopefully my unhelpful questions can be of some sort of use to you.

They definitely were, since they pointed out areas where my descriptions were confusing. I'll have to go back and edit my previous post.

I think that you should keep up with your idea of not having any voiced fricatives or plosives bar /v/; if you're thinking of adding more phonemes, I'd suggest that they be voiceless too (ʃ/ʂ and/or ç, perhaps) to retain this aesthetic. As for the accusative, if you strictly keep to your specified word order, it may not be necessary because it will nearly always be implicit. If you don't, I'd suggest an accusative ending for the sake of clarity. You might want to consider an direct object particle instead if the latter is the case.

Clarity is an issue, and those sound like good ideas. I suspect that an accusative ending might increase word lengths beyond what we'd like, but I'll have to ask my wife. I note that you're not the only one suggesting not to add /d/, but non-linguists almost always regard the lack of a "d" as a poor feature, and I still think the vocabulary we have creates tongue twisters - it may be necessary to replace some of the "th" sounds especially with other things.

As for your language, Celinceithir, I do want to look at it, but my son is already complaining that I'm hogging the computer he uses for games, so it'll have to wait for tonight.

Don't worry about it. Natural languages like Ayacucho Quechua don't have any of those sounds.

Linguists are always good at this kind of thing - I've never heard of Ayacucho Quechua before today.

Does <me> also serve for "he" and "she"?

No. <moi> = "he" <ne> = "she"

Are the pronouns plural in the same way as other nouns (<slai> "we", <sthyu>, <sme>)?

Yes.

Is there some facility for pluralizing words that already begin with /s/ (or some other tricky thing like /ps/ or /x/)?

Not currently, and it does create vagueness in a few cases.

The romanization of your sample doesn't show any distinction of /e/ and /ɛ/. Are all of those <e>s really /e/?

No, many are /ɛ/.

Your samples are not particularly agglutinative (I see a couple of <thyuie>s and one <tho->). Maybe try some longer sentences so we can get an idea of the way these things string together?

Difficult; I'm not very good with longer sentences, but I can try these:

<Lai tiuswie sakiri kaithi sol thyu.> = "I fourteen flowers get-will for you."
<Lai thyuie aesmith avir, vex thyuie lainu thoavir." = "I your super-paper (money) want, but your life not-want."
<Eofl laniyar lelifalorir riak, iesethaini anu hea.> = "When love young-person comes, frightening thing is."

In the second, why isn't it <Thyu me avir?>

Checking back, I note that the correct word order for questions is (O-S-V).

Are you using a rising tone (as in English) to signal a question?

In theory, but I notice I often use a falling tone. Question words are all "la-" and usually easy to spot.

Riak owe. (Come here.)
Ves thyuie piev. (Eat your food.)
Sil rith lir. (Give that to me.)
Thoisil rith! (Not-touch that!)
Psel lai ma rith. (Help me with this.)

In all of these, it looks like the verb is first. Is this a grammar rule for expressing the imperative (to distinguish from declarative <Thyu thyuie piev ves> "You eat your food.")?

Yes.

Is there a predictable production rule for <lai> "I, me" becoming <lir> "to me"?

Dative and locative take the "ir" ending; pronouns retain only the first phoneme plus "ir."

"We" is a pronoun, not a preposition. I believe you got pronouns and prepositions mixed up.

Yes, I'm sorry about that; I meant to be talking about "with," not "we."

In what manner do you intend to teach your children the language?

By speaking it to them, and occasionally telling them sentences. My older son (he's 5 years old) knows "come here," "give me that," "be quiet," "bedtime," "go," "hot," "cold," and a few other simple questions and commands from immersion. His English is much better, of course, and he's not confident at all when speaking - the /θ/ and bounced /ɾ/ are tough for the guy.

We can help you to reach your goals, but exactly what your goals are can only be chosen by you and your wife. So what are they?

1. Learn what makes one language better than another, objectively speaking.
2. Apply knowledge within the context of our subjective needs.

We were disappointed when we discovered that many conlangers seem to believe point 1 is nonsensical - we both strongly doubt that to be the case. It is clearly not true regarding any other art form; for instance, studies find that abstract art painted by professional artists is preferred over abstract art painted by children or animals.

2. What I personally think makes an objectively good conlang (independent of aesthetic preference for one structure over another) conflicts pretty heavily with your goal of it being a language you and your wife can learn and teach to your children.

I sympathize; but I'm also aware that languages spread most effectively through children. If we created a language that was too complicated, it would be more likely to die out; our subjective needs require the language to outlive us.

I think a conlang grammar should read out like a math proof: Quite complex, but all the pieces fitting together perfectly. If you don't need ten pages to explain how your case system works, then your language is too simple for my liking. However, this will make your conlang difficult to learn for the same reasons natlangs are difficult to learn. Are you willing to sacrifice interestingness for learnability, or vice-versa?

I think that's a key question, and it's the heart of what we're struggling with. One feature may make the language a bit easier to learn, while making it much less clear; another feature may make it much easier to learn, while making it only a bit less clear. We want to choose the latter sort of tradeoff - wherever we can figure out what it is.

I hope you're not taking this seriously, but rather as a game... Maybe your children won't give a damn about that lang.

Maybe not; but if you find it funny, shouldn't you hope we're rabidly serious about it, in order to maximize its humor value? In any event my children do speak it, but not well. Part of the goal is to design it in such a way that they end up speaking it well and do give a damn. For right now, though, oldest is getting very bored waiting to play Plants vs Zombies, so I'd better sign off!

[Micamo]

We were disappointed when we discovered that many conlangers seem to believe point 1 is nonsensical - we both strongly doubt that to be the case. It is clearly not true regarding any other art form; for instance, studies find that abstract art painted by professional artists is preferred over abstract art painted by children or animals.

This is because conlanging as an art form hasn't even begun its development yet: The number of conlangers around is small, and those who take it seriously even smaller. The idea that there's nothing more conlanging can do than satisfy a personal whim is simply the null hypothesis.

[Systemzwang]

I hope you're not taking this seriously, but rather as a game... Maybe your children won't give a damn about that lang.

Maybe not; but if you find it funny, shouldn't you hope we're rabidly serious about it, in order to maximize its humor value? In any event my children do speak it, but not well. Part of the goal is to design it in such a way that they end up speaking it well and do give a damn. For right now, though, oldest is getting very bored waiting to play Plants vs Zombies, so I'd better sign off!

You know, there is research - scholarly, high quality research - about how linguistic issues affect the development of children. This is why some - like me - won't prefer you to be rabidly serious about it, or anything like that. I hope you have thought about issues like that, and that you're somewhat knowledgeable on the topic.

[Trailsend]

1. Learn what makes one language better than another, objectively speaking.
2. Apply knowledge within the context of our subjective needs.

We were disappointed when we discovered that many conlangers seem to believe point 1 is nonsensical - we both strongly doubt that to be the case. It is clearly not true regarding any other art form; for instance, studies find that abstract art painted by professional artists is preferred over abstract art painted by children or animals.

The problem many conlangers have with your point 1 (or at the very least, the problem I have with it) is that it presupposes the notion of "better."

Your comparison to abstract art assumes (justifiably) that "better" can be measured by "preference"; that is, the professional art can be said to be better than the children's art because it is preferred over the children's art. There is no such common metric applicable to conlangs. The best that can be done is to determine how well a conlang achieves the goals of its creators, which is why we tend to (like Micamo) first ask about what the goals for your project are so that we know how it should be evaluated. It's not like comparing abstract art painted by professionals to abstract art painted by children—it's more like trying to compare a window designed to maximize the natural lighting inside a building to a cartoon designed to satirize a particular political policy. Preference is not going to be of any use to you here; which is objectively better, the window or the cartoon?

You know, there is research - scholarly, high quality research - about how linguistic issues affect the development of children. This is why some - like me - won't prefer you to be rabidly serious about it, or anything like that. I hope you have thought about issues like that, and that you're somewhat knowledgeable on the topic.

Yeah, the issue is somewhat complex. There was a thread recently in which we discussed this, which included links to various studies about the effect of multilingualism on children's development. Check it out here. Posts with links to studies are here and here.

[Micamo]

It's not like comparing abstract art painted by professionals to abstract art painted by children—it's more like trying to compare a window designed to maximize the natural lighting inside a building to a cartoon designed to satirize a particular political policy. Preference is not going to be of any use to you here; which is objectively better, the window or the cartoon?

This is a valid argument against comparing Lojban and Quenya, but he (she?)'s already stated he wants an aesthetically pleasing language that wouldn't be unreasonably difficult for himself and his family to learn fluently. That's more than enough to build an objective criterion out of.

[Trailsend]

It's not like comparing abstract art painted by professionals to abstract art painted by children—it's more like trying to compare a window designed to maximize the natural lighting inside a building to a cartoon designed to satirize a particular political policy. Preference is not going to be of any use to you here; which is objectively better, the window or the cartoon?

This is a valid argument against comparing Lojban and Quenya, but he (she?)'s already stated he wants an aesthetically pleasing language that wouldn't be unreasonably difficult for himself and his family to learn fluently. That's more than enough to build an objective criterion out of.

Oh, definitely. I didn't mean to suggest that languages can't be objectively evaluated, because once you know their design goals, they very often can be. I was only talking about why there's no metric that can be used to compare any two languages.

[Ναθια]

Phonology of Standard Perisian Celinese:
Vowels: [ä] <a, à>, [ɛ] <e, è>, [e] <é, ê>, [ø] <i>, <í>, <y>, [ɔ] <o, ò>, [o] <ô>, <u>, [ə] <ë, ı>

Consonants:
p b t d k (c) ɡ
m n
r
f θ (th) ð s χ (ch) h
j (i) w l

(Allophony: r = /ʐ/ in coda; g = /ç/ in coda; c = /χ/ in coda; f = /v/ intervocally, s = /ʃ/ before <é>, <í> or after <ı>; n, t and l = /ɲ/, /c/ and /ʎ/ after /ı/)

Other points:
- Three genders, formally determined by their endings: maiðír (mother) would be neuter, garys (boy) would be feminine, gathal (girl) would be masculine simply because of this.
- There are no articles, so the main way gender comes into play is in the plural endings: ím - masculine, ot - feminine, ain - neuter.
- Most tenses are created periphrastically with auxiliary verbs: fyðí for the future, weðí for the conditional, etc. The present and the simple preterite on the other hand are formed by inflections on the infinitive.
- There are initial word mutations, like in Gaelic. Old Celinese had several mutations, modern Celinese dialects usually have just one, the soft mutation, occuring usually after prepositions, personal pronouns, numbers higher than three and adjectives when placed before a noun. E.g. cafel (/ˈkavɛl/) = room; in my room = na mo g-cafel (/ˈgavɛl/)
- Word order is very flexible, with different orders preferred depending on context. The most common is OVS (ané mairí aıs - you love I)
- Past continual and past habitual are split, as are the present and the past-present continuous.
- No cases other than in the pronouns (this is not the case with some dialects though)
- 30+ productive affixes

OK Celinceithir, I think this is a pretty good overview. But can you explain why you made any specific decisions? Which would you regard as crucial to differentiating it from other languages? Which features did you select because they complement one another well?

You know, there is research - scholarly, high quality research - about how linguistic issues affect the development of children.

Yes; the effects, both positive and adverse, appear to be minimal.

The best that can be done is to determine how well a conlang achieves the goals of its creators...

Really? I don't see how that way is best. Do we have to know the creator's goals before evaluating a poem, or a car? Yes, some might evaluate a language in terms of their sense of the creators' goals. But clearly that is not the only way. Others might rate it according to their sense of beauty, the ability to express visceral concepts, efficiency of expression, clarity, and so on. Unless some compelling reason can be given not to let every opinion count, why not simply average all together and see whether some general criteria should appear? In fact, can anyone offer suggestions about which languages are generally well regarded among conlangers?

In response to a "top ten" conlang list, I notice that readers had this to say:

1/26/04 - EEFanSite wrote, "I also think Quenya should be on top. It's much more better than Esperanto, even though it's not finished."

12/31/04 - Katherine Hudson wrote, "I would think that because Sindarin is equally intricate and certainly quite unique from Quenya (especially with regards to phonological phenomena such as consonant mutations and vowel changes), captivating the interest of linguists and language enthusiasts of all parts of the world, it would be necessary to give it equal mention next to Quenya in a 'Top Ten List'."

1/9/02 - Craig wrote, "In general, I agree with your Top Ten Languages list. However, I must object to placing Nadsat above Lojban. Unlike the rest of the languages there that I know about, it is only a selection of slang terms. It has no grammar of its own, and nothing can be said in it without the massive use of standard English words. Lojban can be used for fluent conversation (one person once went a weekend speaking only Lojban in the annual lojbanist gathering, LogFest). Of the languages listed, the most thought about the nature of grammar probably went into Lojban; it is the only one in which the grammar is nothing like that of any natural languages but is nonetheless easy to learn."

11/22/00 - Kleine Kaoten wrote, "I think that Quenya should be on top because even if it is not complete it has its own origin. Not like Esperanto.... I must admit that there are some Latin influences in Quenya, but they are not really important. There will be more Tolkien pieces of Quenya when the LOTR film comes out. I think that there are some handwritten items which are not published yet. My opinion is, that all Tolkien fans out there should try to complete Quenya in his sense!"

Evidently incompleteness is generally regarded as a drawback in a conlang. This should be obvious; does anyone really believe that a completed conlang is not better than that same conlang in its early stage of development? The two conlangs are not identical; one is definitely better, in the most commonly understood sense of the word "better." Clearly we shouldn't need to know a conlang's purpose to know it would probably be improved by completion.

The trouble my wife and I are having is that most of the features that make for good conlangs - rather than drawbacks which can be easily avoided - are unknown to us. Even some mistakes we're still stumbling upon in a vague sense; for instance we figured out that having words with related meanings carry related sounds makes those words difficult to distinguish in conversation. They may be easier to learn, but when there are enough of them, they add up. Another mistake I think is in overusing a few sounds. Our language uses a lot of "l" and "th," and when speakers hear these perpetually, it's easy to mix them up, particularly when they're used in common words, like pronouns.

I'm slowly coming into a sense that a crucially important aspect of languages is its number of distinctions. Distinctions provide information, but also require processing, and there's a tradeoff there. We know that many animals like dogs vocalize when danger is near; I recall from a physical anthropology class years ago that a certain group of primates had two vocalizations, one indicating danger on the ground, and the other indicating danger in the trees. We might simply regard a distinction between "danger-up" and "danger-down" as better than "danger," but the animals using these two words must learn to vocalize them, to vocalize them appropriately, and to interpret the vocalizations of others. The costs of having more distinctions may outweigh the benefits.

So I speculate that in human languages, insufficient distinctions make a language easy to speak, but make communication difficult; the reverse problems occur with superfluous distinctions. Thus a language without adjectives or adverbs will probably, if well designed, have larger numbers of nouns or verbs. But this may be spurious, or overrated in importance - I'm not a linguist, and really don't know. Does it make sense?

[Micamo]

I'm slowly coming into a sense that a crucially important aspect of languages is its number of distinctions. Distinctions provide information, but also require processing, and there's a tradeoff there. We know that many animals like dogs vocalize when danger is near; I recall from a physical anthropology class years ago that a certain group of primates had two vocalizations, one indicating danger on the ground, and the other indicating danger in the trees. We might simply regard a distinction between "danger-up" and "danger-down" as better than "danger," but the animals using these two words must learn to vocalize them, to vocalize them appropriately, and to interpret the vocalizations of others. The costs of having more distinctions may outweigh the benefits.

This is true to an extent (and is the reason that, say, the usual number of cases in a language is six when natlangs exist with up to twenty), but the tricky part is humans can invent and explain new semantic concepts on the fly: That's sorta what language itself is for.

Thus a language without adjectives or adverbs will probably, if well designed, have larger numbers of nouns or verbs. But this may be spurious, or overrated in importance - I'm not a linguist, and really don't know. Does it make sense?

Not really. The only important difference between nouns, verbs, and adjectives is their syntactic behavior. You can easily have an adjective act as a noun, or a noun act as a verb, if you just use syntactic tricks. For example, in English:

"The male one is a student."

As a substitution for:

"The man studies (for a living)."

Mark Baker's theory is that lexical categories are merely the conflation of a bare semantic head and a functional head. Constructions like "male one" expose this underlying structure inherent in nouns like "man." You could, in theory, make a language with neither verbs nor nouns nor adjectives nor adverbs, and simply have bare "content" words that combine with purely functional words that indicate syntactic structure.

[Trailsend]

The best that can be done is to determine how well a conlang achieves the goals of its creators...

Really? I don't see how that way is best. Do we have to know the creator's goals before evaluating a poem, or a car?

You most certainly do have to know the creator's goals—otherwise you run the risk of saying things like "This poem is awful, it doesn't even have any airbags!" or "This car is terrible, its rhyme scheme is all wrong."

Yes, some might evaluate a language in terms of their sense of the creators' goals. But clearly that is not the only way.

Many methods are possible, but that doesn't mean they make sense.

Others might rate it according to their sense of beauty

But this would assume that all conlangs are supposed to be beautiful. Many conlangs have been created with the explicit purpose of not being beautiful; Tolkien's Black Speech and Okrand's Klingon, for example. There are any number of reasons why a conlanger would want to do this. Maybe the language is to be spoken by hellspawned demons and should therefore sound grating to human ears. In such cases, it must be possible for a language to do a very good job of sounding ugly (to certain people).

Evaluating all languages by our own senses of beauty makes about as much sense as evaluating all paintings by how much blue is in them. We can't say that one painting is better than another just because there is more blue in it, because the un-blue painting might be depicting something that is not blue. Similarly, we can't say that one language is better than another just because it sounds more beautiful, because the un-beautiful language might be designed for a purpose that requires it to not be beautiful.

the ability to express visceral concepts

This assumes that all conlangs are supposed to be able to express visceral concepts. It is not at all hard to imagine a conlanger creating a language for an imaginary race of robots which have only recently developed sapience. In this case, the language might do a very good job of depicting what the thought patterns of only partially sapient creatures might be like, even though it is completely incapable of expressing a great number of visceral human concepts. Indeed, it should not be able to, because doing so would require it to do a worse job of what it was designed for.

efficiency of expression

This assumes that all conlangs are supposed to be efficient. The language of Tolkien's Ents was specifically intended to not be efficient, to reflect the way that the Ents take a very long time to discuss things. A language like this could do a very good job of being very long-winded; it would make no sense to consider it "worse" than some other language just because it didn't do something which it was specifically designed not to do.

clarity, and so on.

Similarly, it is not at all hard to imagine a conlanger creating a conlang specifically designed to do a very good job of being vague and ambiguous.

Unless some compelling reason can be given not to let every opinion count, why not simply average all together and see whether some general criteria should appear?

My reason: we should not count opinions like

• "This is an awful car because its rhyme scheme is all wrong,"
• "The Mona Lisa is an ugly painting because it doesn't have enough blue in it," and
• "This robot language cannot efficiently express the visceral concept of 'wistfulness,' so it is not as good as this fairy language that can."

Such opinions are nonsensical and should not factor into the evaluation of a language (or anything else).

In fact, can anyone offer suggestions about which languages are generally well regarded among conlangers?

Tolkien's languages are often considered to be very high quality. The elvish languages in particular were intended to be naturalistic, with believable historic developments and levels of usability. Because many conlangers believe that they did indeed achieve these goals, they are admired.

On the contrary, some conlangers consider Paolini's languages from the Inheritance series to be low quality, because they do not exhibit the degree of naturalism they should given their setting.

Many conlangers consider John Quijada's language Ithkuil to be very well done, even though it is extremely complex and would be very difficult for humans to learn. The language was not designed for ease of communication but rather as "an idealized language whose aim is the highest possible degree of logic, efficiency, detail, and accuracy in cognitive expression via spoken human language, while minimizing the ambiguity, vagueness, illogic, redundancy, polysemy (multiple meanings) and overall arbitrariness that is seemingly ubiquitous in natural human language." (Link)

(I provide these examples because you asked for them, but again, finding examples of people who behave a certain way does not necessarily mean that that behavior makes sense.)

Evidently incompleteness is generally regarded as a drawback in a conlang. This should be obvious; does anyone really believe that a completed conlang is not better than that same conlang in its early stage of development? The two conlangs are not identical; one is definitely better, in the most commonly understood sense of the word "better." Clearly we shouldn't need to know a conlang's purpose to know it would probably be improved by completion.

You definitely need to know a conlang's purpose, because otherwise you have no way of telling how "complete" it is. One conlang might be designed exclusively to provide names for 12 cities on a map in a fantasy book, while another might be designed to be a fully-usable language meeting all the communicative needs of modern-day humans. The first will be quite complete once it has a dozen words or so, while the other will be quite incomplete even when it has hundreds.

The trouble my wife and I are having is that most of the features that make for good conlangs - rather than drawbacks which can be easily avoided - are unknown to us. Even some mistakes we're still stumbling upon in a vague sense; for instance we figured out that having words with related meanings carry related sounds makes those words difficult to distinguish in conversation. They may be easier to learn, but when there are enough of them, they add up. Another mistake I think is in overusing a few sounds. Our language uses a lot of "l" and "th," and when speakers hear these perpetually, it's easy to mix them up, particularly when they're used in common words, like pronouns.

Again, I don't contest that there's some way to evaluate whether your language specifically is doing well, or whether making a particular change to your language would make it better or worse. Your language is supposed to do very particular things—conform to your personal aesthetic sense, be easy to learn, facilitate efficient day-to-day communication, etc.—and so any given decision about it can be weighed against these goals to see if it brings you closer or farther away from achieving them. My point is that there are very many languages for which these same criteria don't make any sense, and so we cannot use them to evaluate those languages.

I'm slowly coming into a sense that a crucially important aspect of languages is its number of distinctions. Distinctions provide information, but also require processing, and there's a tradeoff there. We know that many animals like dogs vocalize when danger is near; I recall from a physical anthropology class years ago that a certain group of primates had two vocalizations, one indicating danger on the ground, and the other indicating danger in the trees. We might simply regard a distinction between "danger-up" and "danger-down" as better than "danger," but the animals using these two words must learn to vocalize them, to vocalize them appropriately, and to interpret the vocalizations of others. The costs of having more distinctions may outweigh the benefits.

So I speculate that in human languages, insufficient distinctions make a language easy to speak, but make communication difficult; the reverse problems occur with superfluous distinctions. Thus a language without adjectives or adverbs will probably, if well designed, have larger numbers of nouns or verbs. But this may be spurious, or overrated in importance - I'm not a linguist, and really don't know. Does it make sense?

I think you're onto something here, although the issue is a little more complex than this. For example, I speak Chinuk Wawa, a language which in many cases makes fewer distinctions than English. (E.g., it does not distinguish between "and," "or," and "than"; or between "for," "if," "because," and "when"; or between "again," "more," "rest," "left," and "also".) This didn't always make the language easier to learn, because I still had to learn all of the different circumstances in which you were supposed to say "wəxt."

There's an idea from information theory that can be applied here, though. Human languages convey information over a couple of different channels, including semantics (root meanings of words), morphology (how words can be modified to tweak their meaning), syntax (how words can be arranged), and pragmatics (how context contributes to meaning). If we suppose that the average amount of information that humans need to communicate to each other is more or less constant, then the language must maintain a certain level of overall entropy. This means that any time you simplify (that is, decrease the entropy of) one channel, the entropy of the other channels must increase. If you try to simplify the semantics, morphology, and syntax channels, then a huge informational load is going to fall on your pragmatics channel, making it very hard to communicate without being misunderstood.

The key, then, is to find a distribution over all of the channels which works for you. I cannot tell you what that distribution is, because there is no one correct answer; Chinuk Wawa has a semantics channel which is in some places much shallower than English, and yet both languages are just as usable. Finnish has a morphology channel which is much deeper than English's, and yet children learn both languages just fine.