A priori word creation

A priori word creation

adapted from a 03 Dec 96 post by Houldsworth

Methods of word creation can be categorised in several ways. This is one that may be useful:

A-priori
- Random [e.g Ferrengi]
- Semirandom
- Permutative [e.g Classical Yiklamu]
- Categorical [e.g Ro,SLAS]
A-posteriori
- Direct [e.g Dunia, Vorlin]
- Conformative [e.g Esperanto]
- Algorithmic [e.g Pralin (my own work-in-progress)]

A-priori words are created by the linguist. A-posteriori words are derived in some way from already existing words. Most a-posteriori conlangs have a-priori words.

Random words are just that - randomly chosen sequences of letters. Ferrengi is the only language that I know of that generates words like this:

AGDWIF
MKO
SQLAAD
TYTZF
EIFGHO

Semirandom words are sequence of letters that are chosen randomly from a set of possibilities determined by morphological rules and those letters already chosen.
For example, suppose we have a language where all words are [CVCCV]+[N/F/0] or [CCVCV]+[N/F/0]. The first letter can be any consonant, but vowels are forbidden.If the first letter chosen is {P}, then the next may be a consonant or a vowel - we have complete freedom within the phonology.

If the second letter pulled from the hat is [I], then the third must be a consonant. And so on for the next two letters.

Now, the five-letter-set may or may not have a consonant appended. If there is one, it cannot be a plosive. This give us words like:

TOKNI
SKAPON
DIFFO
NIPLI
ZGAGIM

Permutative lexicons are created en-masse according to morphological rules. Say we have the rule that all words are [PV]+[N/F]. This might give us:

Once the list of valid words has been generated, meanings can be assigned, perhaps mapped randomly from a list of glosses.

Categorical word-building is the process whereby the meaning of a word is indicated by it's letters.

Suppose that the initial letter indicates the general category of noun to which the noun belongs, e.g:
- P - AMIMAL
- B - VEGETABLE
- T - MINERAL
- D - ABSTRACT
...and that the second letter - a vowel or diphthong - indicates some more specific detail, e.g:
- A - RURAL
- AI - INDUSTURIAL
- AU - DOMESTIC
- E - PERSONAL
- EI - GLOBAL
...and so on, e.g
- L - TOOL
- K - FOOD
- D - FURNITURE etc.
This might eventually give us, say:
- BAKIME - Tomato
- DEPIK - Chair

Direct words are those taken, either the whole word or part of it, from words already existing in other languages, with minimal change, or even none at all e.g:

MON - World (in Vorlin)
MI - Me (in Interlingua)

Conformative words are those found in most IALs. Words are taken from already existing languages, and modified, if necessary, to fit the phonology and phonotactics of the conlang, e.g

ABELO from ABEL
TABLO from TABLE
VI from VU (all in Esperanto)

Algorithmic words are those derived from source words by some algorithmic process. This category is something of a rag-bag of techniques, and those given are merely a few example of many possible.

So, in our first algorithm, suppose our source words are:

TABLE [teibl] (english)
TABLE [tabl] (french)
MESA [mez@] (spanish)
TISCH [tiS] (german)

Now we take a poll of the initial consonants. [T] occurs three times, and [M] once, so the initial letter of the derived word is [T]. By the same process, [E] is chosen as the first vowel, and [B] as the second consonant. So our comosite word is: TEB

In a different example, suppose we have only the source words:

MESA [mez@] (spanish)
TISCH [tiS] (german)

Now, we conform these words to a phonology of our own, to produce:

We can string the two together, to make TISMES or MESTIS, or we can be more sophisticated. Both words end with [S], so the first can be dropped, to give METIS or TIMES.

This was one attempt by me to create a vocabulary that was easily learnable for both german and spanish speakers, by making words that contained elements of the corresponding words in both languages. You will not be surprised to hear that I abandoned this particular algorithm, although it did give me the name for my currant project

:
   PRALIN from SPRAK and LINGUO

Return to Conlang-related topics|Back to FAQ page