Coining new words

[bradrn]

When I need a new meaning in a conlang, I have several options:

1. Extend the senses of an existing word to cover the new meaning
2. Use derivational morphology on an existing word to derive the new meaning
3. Express it as a compound (or with serialisation or incorporation etc.)
4. Some other sort of periphrasis
5. Use a word from the parent language, but with a semantic shift
6. Coin a completely new word

EDIT: And 7. Borrowing

I’d be curious to know: which option do you prefer for your conlangs, and how do you choose between them? (I’d be especially interested in what you do for protolanguages or other isolates, since in that case option (5) is no longer available.)

[zyxw59]

Typically I've done (2) if I can, otherwise falling back on (6)

[Vilike]

(2), except if I am sure I can go away with (1) (do I really need a new word?). More often than not playing with (2) and (3)-(4) comes first and the definitions next: I am not always in a hurry to get a specific meaning. Like, recently I wanted to make a progressive out of a certain phrasal construction, and discovered that it made more sense as an abilitative. Needs and deeds may not coincide.

(6) as a last recourse, or for Lexember.

[xxx]

3&4, in a triple S philosophical language (1Sound=1Sense=Sign) this is the same thing and no other choice is possible...

[evmdbm]

2 if I can. That's not always possible, mind - although I could always come up with derivational morphology. In fact sometimes thart's how I decide I need to create a way to turn a adjective into a noun (say to identify the quality of being whatever...) or I just mess about with a root. Grelaq means to stick out, so someone sticking out would be grelek (except the -ek suffix is only used with transitive verbs); something that is stuck out is grelel (except grelaq is intransitive) so I just fiddled it and came up with grela for peninsula, which is of course is a pretty specific thing that sticks out. Otherwise I might try compounding in a rather unscientific way, but at least I try to follow Zompist's advice not to create a new root every time

[zompist]

Everybody is being reasonable here (i.e, like me), but I'd add one more important category: borrowing. Obviously this gets easier the more languages you have, but you can get going with just one other language.

Occasionally you can use interdialectal borrowings, especially for minor differences in meaning.

I usually have a tier of words marked "imitative". This can obviously be used for sound effects (splash), and can be metaphorically extended (flash); it's also good for motion verbs, musical instruments, and animal names.

[bradrn]

Otherwise I might try compounding in a rather unscientific way, but at least I try to follow Zompist's advice not to create a new root every time

This is what worries me about coining new words… I’d like to follow zompist’s advice and not make up new roots all the time, but given that I tend to be very slow at creating both words and derivational morphology (my current language has ~40 words and just 1 derivational affix), I feel that I am forced to make new roots anyway, simply because I don’t have any alternative. (And then of course I feel bad about doing ‘the wrong thing’… yes, I’m an inveterate perfectionist.)

Everybody is being reasonable here (i.e, like me), but I'd add one more important category: borrowing. Obviously this gets easier the more languages you have, but you can get going with just one other language.

Occasionally you can use interdialectal borrowings, especially for minor differences in meaning.

Ah, of course! Let me retroactively add that as ‘option 7’.

I usually have a tier of words marked "imitative". This can obviously be used for sound effects (splash), and can be metaphorically extended (flash); it's also good for motion verbs, musical instruments, and animal names.

These seem to be universally called ‘ideophones’ (except for Uralicists, who call them ‘expressives’). Other categories include colour, smell, spatial state etc. (I’ve been reading a book on them, as it happens.)

[zompist]

This is what worries me about coining new words… I’d like to follow zompist’s advice and not make up new roots all the time, but given that I tend to be very slow at creating both words and derivational morphology (my current language has ~40 words and just 1 derivational affix), I feel that I am forced to make new roots anyway, simply because I don’t have any alternative. (And then of course I feel bad about doing ‘the wrong thing’… yes, I’m an inveterate perfectionist.)

As I say in the Lexipedia, you're going to have to invent a bunch of roots. It's just good not to have that as your default option.

I usually have a tier of words marked "imitative". This can obviously be used for sound effects (splash), and can be metaphorically extended (flash); it's also good for motion verbs, musical instruments, and animal names.

These seem to be universally called ‘ideophones’ (except for Uralicists, who call them ‘expressives’). Other categories include colour, smell, spatial state etc. (I’ve been reading a book on them, as it happens.)

If I'm not mistaken, ideophones are a slightly different concept-- often a separate part of speech. They're pretty marginal in English, but very important in some languages.

E.g., in "He drummed on the table, taratata", "drum" is imitative but not an ideophone, and "taratata" is an ideophone.

[bradrn]

I usually have a tier of words marked "imitative". This can obviously be used for sound effects (splash), and can be metaphorically extended (flash); it's also good for motion verbs, musical instruments, and animal names.

These seem to be universally called ‘ideophones’ (except for Uralicists, who call them ‘expressives’). Other categories include colour, smell, spatial state etc. (I’ve been reading a book on them, as it happens.)

If I'm not mistaken, ideophones are a slightly different concept-- often a separate part of speech. They're pretty marginal in English, but very important in some languages.

In fact, according to said book, ideophones are usually not a distinct part of speech at all. For instance, Ewe has ideophonic adjectives (its non-ideophonic adjectives are a closed class): legbee ‘long’, wódzóé ‘light’, ɖiɖǐ ‘ripe’. Hausa has ideophonic nouns: ɓàlli-ɓàlli ‘palpitations’, wàndàr̃-wandar̃ ‘zigzag’. (So does English: mumbo-jumbo, ping-pong.) Mundang has ideophonic verbs: gàl ‘cackle’, vɪɗyàl ‘wave’, ŋgə̀lwàɓ ‘pound hard’. And Pastaza Quechua has ideophonic coverbs, e.g. dziriririririri-shi aysa- ‘drag’. In general, it seems that ideophones are usually defined in terms of phonology and semantics, and only rarely show distinct syntax; more often they cut across the other word categories.

[Ares Land]

Here's what I use for Simbri
A long time ago, I generated about 500 roots with a vocabulary generator (I don't remember which one) for proto-Tarandic.
I assigned more or less randomly 200 or 300 of these a meaning, using the Swadesh and as I recall, a Sumerian grammar.

If I need a word, I check a few etymologies for ideas (though sometimes I just come up with an etymology that makes sense).

Than I derive an ancestral form, generally by combining two roots (like Mandarin, the early Tarandim languages dealt with homophony with compounds), then the Simbri form, at least trying to come up with a semantic drift.
I may also use derivational morphology in either Simbri or the parent language (or both).

When I need a new root, I just assign the meaning I like to one of the proto-Tarandic rule that haven't been assigned a meaning yet. I'm actually running out of unassigned roots, but it's not really a problem. It seems that after 400 roots or so you don't really need new ones so much.

Sometimes I don't like the way the Simbri word looks. So I cheat with the sound changes and attribute it to a cross-dialect borrowings (there are lots of these in natural languages, so good). Sometimes I just make up a new root or change an existing one to get the results I'd like.

Simbri is supposed to have a lot of Katli borrowings. I haven't worked on the Katli language yet. So I just make something up and apply the most recent changes to it. Figuring out how Katli works is future me's problem.

[evmdbm]

Ah yes

7 too! Although there's not much of that going on yet. I do imagine Vedreki borrowing from Cheyadeneen and vice versa. Since the two languages (and their ancestors) have been spoken cheek by jowl for over 3000 years it would be surprising if there weren't borrowings, but since (say) Vedreki borrowings are going to be legal/political/governmental material and I haven't had cause to write an essay on the governance of the Cheyadeneen Empire in Vedreki those words aren't here yet - in either language

[vegfarandi]

To supercharge borrowing, have an older stage of the neighbor language and your main one and do borrowing in two stages. Old ones get run through the second half of sound changes and you can get yourself pairs like warden and guardian.

[Qwynegold]

The language I've been working with lately is an old proto-lang, so 5) taking something from its proto-lang is not possible. 7) borrowing is practically not possible either, because I have no well developed languages of the same time period. There is going to be one other proto-lang they have contact with near the end of the time period in question, so a few borrowings will have to come that way.
I use 6) coin a new word and 2) use derivation the most. 1) extend the meaning of existing words and 3) compounding get used a bit. I don't know what 4) other periphrasis could be.

[Vardelm]

Right now, everything is #6 since I haven't done anything with vocabulary aside from making up words on the spot as I need them for my posts here.

Eventually, I'd like to have a systematic use of all of those for all 4 of the languages I'm working on, but that's still a ways out.

[kodé]

Usually 2 (or 3 or 4, depending on the morphosyntax of the language I need the word for). Partly because I love morphology a lot more than lexical semantics (apologies to lexical semanticists), partly because it’s a nice way to tinker with the language. 6 if I have to, though I have a weird aversion to creating lots of roots for new concepts.

5 and 7 are good motivators for me to do more diachrony (outside of conlang relays, haha) and conworlding. Most of my conlangs are isolates with no neighbors!