= 27d567bc2d48d9f40e9ccc20ea370b15)
ChristianKl comments on Linguistic mechanisms for less wrong cognition - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (130)
Please forgive me a bit for mixing different ideas over multiple post in this thread with a bit of overlap. I consider the ability of a language to specify relationships very valuable and underdeveloped in English. Latin has a word for mother of father. English has only grandfather or grandmother. It has ugly constructions like great-grandfather.
In my draft I have the following root words:
ba = 0
ce = 1
di = 2
ma* = female
ne* = male
caiq = parent
Out of those roots I can create: caiqma = mother
caiqne = father
caiqce = grandparent
caiqcemaba = grandparent (parent of the mother)
caiqceneba = grandparent (parent of the father)
caiqcemace = grandmother
caiqcenece = grandfather
caiqcemana = grandfather (father of the mother)
caiqdi = great-grandparent
This way of specifying relationships is quite efficient. In case you want to distinguish your parents not by gender but by which parent is older and which is younger, you can simply use the syllable for "younger" instead of the on for "female". That way the language can translate easily from languages that have different words for older and younger brothers, while not forcing lanugage users that don't want to make distictions based on gender or age.
Why four letters for
caiq? Because it's based oncaiwith simply points to the parent node in any graph. Combingcaiwith the sylable for knowledge from authoritiesfwe, caifwe becomes teacher. It's easily extensible so that caifwece is the teacher of my teacher. English has no word for teacher of my teacher and my language can still do it in 8 letters. It can even do teacher of the teacher of my teacher in 8 letters a case where English feels like Pirahã.Do other words for family relationships are:
fuiq = sibling
caiqfuiq = aunt/uncle (parent's sibling)
Out of that a person with the same teacher as me (classmate) becomes from the structure we already have
fuifwe. We get a new word ofcaifuifwewith means a person with whom your teacher learned together under his teacher. We get that word without the language learner having to learn it explicetly.There will be many cases where more complex relationships can be easily expressed with that system. Via Sapir-Whorf I would expect that this well structured system of relationships makes it easier to think about more complex relationships.
*ma/ne : Those are very provisional. Likely it's no good idea to have two nasal consonants at this place but instead use two consonants that differ more from each other to reduce the cognitive effort that's required to hear whether someone says one or the other.
This compounding system is mostly good, but there's a problem in the phonology:
My linguistics-trained but English-speaking brain refuses to accept "qc" as a valid mid-word consonant cluster, and insists on a phonology rule to put a vowel in between. (I realize there are several ways of mapping q and c into IPA, but none of them worked for me in this case.)
I drafted the words with the phonology rules of http://selpahi.de/ToaqAlphaPrimer.html
caiqis the first syllable of the word andcethe second.But I grant you that at the moment I don't understand enough about phonology to publish a working draft of a language. My intent with this post was more to present the compounding system that I consider to be useful.
Ohhhh, <q> is pronounced /ŋ/. Knowing that, I can pronounce it now. (English usually spells /ŋ/ as <ng>.)
It's not possible to represent every possible set of relationships this way (you can't even say "adopted child of the gay spouse of my stepfather's brother", let alone variations on teacher and classmate). So what you're actually doing is creating a system that can easily represent some sort of relations, at the cost of making it more difficult to represent others.
I think you are confusing "easy to create a system for" and "most useful". It is easy to create a system which specifies "father of father of mother of father of..." It is hard to create a system which specifies things you would actually need to specify often. Your system is efficient in the first sense but not in the second sense.
I'm not making it more difficult to represent others. I don't lose anything that English can do.
You can't add extra features to the language without increasing the cognitive load in deciding when to use the extra features. You're still making everything else more difficult, it's just a distributed difficulty where everything is made more difficult by a miniscule amount, rather than one particular thing made difficult by a large amount.
I don't think that sentence would get added complexity. "adopted child" will likely be a 7-8 letter word using the same root as stepfather. Stepfather is a word that you can't derive from knowing "adoption" which make things harder for the language speaker. You can't derive spouse from knowing the word marriage.
Should "father" be "caiqne" rather than "caiqma" as your comment currently says?
I could count on one maimed hand the number of times I've needed to say "teacher of my teacher". That a language wastes short possible-words on such things is not obviously a recommendation.
[EDITED to add a missing space.]
Teacher of my teacher might not be a good example to show usefulness. Boss of my boss is likely more useful. Even boss of the boss of my boss is a concept that's worthy of being expressed in big modern corporations.
But even a phrase like teacher of the teacher of my teacher can be useful when talking about martial arts lineages.
There's no waste. There only a limited number of possible one-sylable words. If I would give
teachera one sylable word I would spend one slot for it that I couldn't use otherwise. As it stands teacher is made up of two syllablesceiandfwewhich also get used elsewhere.ceican for example be combined with the syllable for love to have a word forperson I love. That automatically gives me also a word for person who loves me via the root that also makes up son/daughter. There also a relations root for bidirectional relations (all the basic categories of graph theory have a one syllable word). If you have a polyrelationship you get a word to describe a person who loves the same person as you do in 6-7 letters. In 8-9 letters you get "the person, that the person I love, loves".There is a saying, don't know by whom: 'To love one's beloved is to love one's beloved's friends, and one's beloved's dog, and one's beloved's children, and one's beloved's wife, and one's beloved's beloved one.'
Yeah, but if I understand correctly ChristianKI's language has special provision for things like "my boss's boss" and "my beloved's beloved" but not for "my boss's husband" and "my beloved's friends". You pick a particular relationship and then you have efficient ways of describing complicated paths through the graph it defines, but there isn't special machinery for combining multiple relationships.
I haven't presented here a way to combine multiple relationships but the language certainly should have mechanisms to handle them. I'm not sure whether it makes sense to have all in one long word or not, but when it comes to language design, it's worth thinking about how those cases get handled.
When it comes to kinship relationships it's worth noting that not every language has a word for "brother". Pitjantjatjara for example has a no word for brother but one "younger sibling".
A language that allows both of those concepts to be expressed is more culturally neutral and doesn't force the speaker into categorising his relationships in the way our culture does.
Yup. But again there are tradeoffs: it could be that complete neutrality ends up making a less useful language than any of several different non-neutral options. (E.g., because you definitely want some words for siblings, but you don't want too many because there are other things to do with the possible-word-space they would occupy, and then every way of having not-too-many ends up not being "culturally neutral" because it inevitably favours some categorizations over others.)
Possible word space is vast. None of the words I used even compete with words in the English language or are easily confused for English words.
Possible word space within a given language is not so vast, and shouldn't be filled too tightly.
Do you think it's just incompetence that has led to existing languages not using every possible short combination of sounds to make words?
Incompetence would assume that the existing languages are designed to be the way they are.
English has 12 vowels (not counting diphthongs) and 24 consonants. Does that mean that English needs 296 different words with two sounds? No, but maybe 100?
Then everything is alright isn't it? The Oxford dictionary contains 100 two letters words. No, it isn't. It contains words such as
aawhich isBasaltic lava forming very rough, jagged masses with a light frothy texture. Often contrasted with pahoehoe.and a lot of other junk likekiwhich isa plant of the lily family.You are correct. I removed the error.