Model Languages
The newsletter discussing newly imagined words for newly imagined worlds
Volume I, Issue 6 (1/2) -- October/November 1, 1995
Table of contents
I won't use words again
They don't mean what I meant
They don't say what I said
They're just the crust of the meaning
With realms underneath
Never touched
Never stirred
Never even moved through
If language were liquid
It would be rushing in
Suzanne Vega, "Language"
If language were liquid, we could enter a submersible
and use sound waves to reveal the subterranean terrain. For each word
floats like a buoy, anchored to some unseen spot far below. The meanings
of the word brother seem easy to pick out from the waters, but
in fact the possible meanings stretch deeper than you might expect, ranging
from "full brother" to "any kinsman" to "any
fellow human" to "anything related" (as in the brother
vices of greed and selfishness). Since we can't use sonar, how can
we sound out the meanings of words?
It used to be thought that any word could be described in terms of semantic
primitives. For instance, M. Bierwisch, writing in 1970, said that semantic
features do not differ from language to language, but are rather part
of the general human capacity for language, forming a universal inventory
used in particular ways in individual languages.
According to this theory, every word can be broken up into primitive
kernels of meaning, called semantemes (also called semantic
features or semantic components). Some sample definitions
using semantemes:
Word
|
Semantemes
|
father |
male + parent |
mother |
female + parent |
son |
male + offspring |
daughter |
female + offspring |
brother |
male + sibling |
sister |
female + sibling |
The process of breaking words down into semantemes is known as componential
analysis and has been most often used to analyze kinship terms across
languages. The components are often given in more detail. For instance,
kinship terms like those shown above might have three components: sex,
generation, lineage. Sex would be male or female; generation would be
a number, with 0 = reference point's generation, -1 = previous generation,
+1 = next generation; lineage would be either direct, colineal (as in
siblings) or ablineal (as in uncles and aunts).
This is obviously a highly technical way to define words we all know
and use without overdue consideration, but -- by using these components
-- you can concisely define a variety of English kinship terms.
Word
|
Generation
|
Lineage
|
Sex
|
|
mother |
-1 |
Direct |
Female |
|
father |
-1 |
Direct |
Male |
|
aunt |
-1 |
Ablineal |
Female |
|
uncle |
-1 |
Ablineal |
Male |
|
sister |
0 |
Colineal |
Female |
|
brother |
0 |
Colineal |
Male |
|
daughter |
1 |
Direct |
Female |
|
son |
1 |
Direct |
|
Male |
niece |
1 |
Ablineal |
Female |
|
nephew |
1 |
Ablineal |
Male |
|
This can be the starting point of a more detailed analysis of English
terms. One advantage of isolating and identifying each component is that
it then becomes possible to identify "holes" in a language's
vocabulary, areas for which it lacks a direct term. For instance, English
lacks a genderless word for an aunt or uncle; you can't fill in the blank
for the statement "parent is to mother and father,
as *** is to aunt and uncle". You can
still express this concept in English (we typically refer to aunts
and uncles but you can more formally refer to parents' siblings),
but you will likely express it less often than you would if there was
a word for it.
Another gap is the lack of words for either "male cousin"
or "female cousin". The paradigm parent/mother/father,
sibling/sister/brother is just not carried out for cousin.
This is unlike other Germanic languages, including Danish (Faetter
and Kusine for male and female cousins respectively), Dutch (neef
and nicht) and German (der Vetter and die Kusine).
Old English probably also made this distinction, but lost it under influence
of Norman French, which -- like most, if not all, Romance languages --
does not make this distinction.
The following table more fully fleshes out the distinctions English
does make in kinship terms.
Word
|
Generation
|
Lineage
|
Sex
|
parent |
-1 |
Direct |
x |
mother |
-1 |
Direct |
Female |
father |
-1 |
Direct |
Male |
|
|
|
|
[parent's sibling] |
-1 |
Ablineal |
x |
aunt |
-1 |
Ablineal |
Female |
uncle |
-1 |
Ablineal |
Male |
|
|
|
|
sibling |
0 |
Colineal |
x |
sister |
0 |
Colineal |
Female |
brother |
0 |
Colineal |
Male |
|
|
|
|
cousin |
0 |
Ablineal |
x |
[female cousin] |
0 |
Ablineal |
Female |
[male cousin] |
0 |
Ablineal |
Male |
|
|
|
|
child, offspring |
1 |
Direct |
x |
daughter |
1 |
Direct |
Female |
son |
1 |
Direct |
Male |
|
|
|
|
[niece or nephew] |
1 |
Ablineal |
x |
niece |
1 |
Ablineal |
Female |
nephew |
1 |
Ablineal |
Male |
There are many more "holes" or gaps in the vocabulary than
those labeled here. What about terms where the generation is not specified?
Where the lineage is not specified?
Please note that the combination of {Generation 0, Lineage Direct} is
meaningless (except for some backwater place -- choose your own to make
fun of -- where brothers marry their sisters and people can be their own
fathers). Since the sense of {Lineage Colineal} only applies to a generation
of 0, it could be thought of as the manifestation of Direct in that area.
(The term Ablineal can apply to any generation.)
A fuller componential analysis (yes, fuller) of kinship terms is presented
in a sidebar below.
While componential analysis is useful for some exercises, it is not
a representation of how language works; no linguist has ever been able
to develop a complete list of semantic primitives. Invariably, some of
the primitives identified are actually molecules that can be broken down
into new atoms. For instance, parent, offspring and
sibling are all interrelated terms; the word parent
can be defined as "a person who has offspring" and sibling
can be defined as "a person with a parent who has other offspring".
If semantic primitives were to exist, they would number in the thousands
and would resemble a mathematical logic system more than the mind's loom
of language.
While Suzanne Vega sings of language being liquid, the rigidity of componential
analysis makes language seem like frozen liquid: ice cubes. While semantemes
have their place, especially to compare and contrast languages' lexicons,
they do not indicate how we actually define terms in our minds.
One of the problems with semantemes is that they assume words have a
single basic meaning. Take brother, which was defined above first
as "male sibling" and then as {Generation 0, Sex Male, Lineage
Colineal}. The English word actually has a much broader meaning than either
of these definitions, with many degrees of brother-ness radiating
out from a core meaning of "male sibling", as shown below in
a no-means exhaustive list.
- brother-german (male sibling)
- half brother
- stepbrother
- kinsman
- comrade
- fraternity member
- co-religionist
- lay person
- racial brother
- fellow man
Defining these labels in more detail we have:
- A male having the same biological parents as another person: a brother-german
- A male having one biological parent in common with another person:
a half brother
- The son of one's stepparent by a previous spouse or lover: a stepbrother
- A male with the same ancestor as another person: a kinsman
- A male friend who is loved as if he were a biological brother: a comrade
- A male friend who belongs to the same fraternity: fellow fraternity
member
- A man who follows the same religious beliefs: a co-religionist
- A male lay member of a religious order: a monk or lay person
- A person of the same race or nation
- A fellow man
- Something that closely resembles another in kind
Even this radius of meaning is not exhaustive: a brother can
mean a "brother-in-law", a brother can be an adopted
son raised by your parents, a brother can be used in the strict
sense of "a fellow African American". The gender of a brother
does not even have to be male -- we must help our brothers in the
Fatherland uses brother to include both men and women (as
illustrated in meaning #9 above).
As this example shows, people think of words not as fixed definitions
composed of semantic primitives, but as examples or prototypes. The prototypical
brother has the same biological parents as another person and has an emotionally
close relationship with his siblings. On a scale of brother-ness
from 0 to 100 the prototype is 100. A 90 might be a brother who was twenty
years older than another person and as a result was never close to him,
or the brother who was abusive and was disliked -- even though these examples
are biologically brothers, they do not share in that emotional closeness
of true brother-ness. A best friend can be considered a brother
on the basis of emotional closeness, rather than kinship. The brother
virtues of love and charity are considered brothers only because they
resemble one another -- this is a metaphorical use of the primary sense
of brother, scoring perhaps a 10 out of 100 on our hypothetical
scale of brother-ness.
The word brother, then, is defined not in terms of semantic
primitives, but in terms of a network of associations with other words.
The human brain recursively defines words by words (just like a dictionary).
When you decide to translate brother into your model language,
you will have to decide which of its many meanings you wish to convey.
Too often we assume that an English word has exact counterparts in other
languages. We say that English brother = Spanish hermano,
when in fact hermano has different connotations. For one thing,
hermano is the expression of a root form herman- with
a masculine ending; give it a feminine ending and you have hermana,
"sister"; hermano has less of a distinctively masculine
connotation than brother does.
Rather than considering the breadth of the meaning of brother,
let's take a simpler example. It is tempting to say that casa
in Spanish equals house in English, like 1+1=2.
casa = house
In fact, casa also equals home, since Spanish does
not distinguish between house and home with separate
words. (Spanish does make a similar distinction, but it does so grammatically,
by saying el casa, "the house", contrasted with casa,
"home".)
casa = home
Of course, in English house and home are different.
house <> home
If this is so, then:
casa <> casa.
Clearly, semantics can never be reduced to an algebra of translation.
As further evidence that words with common meanings are not exactly
equivalent, review some double-translations. The story is told of an American
in the USSR who received a telegram, Your daughter was hung for juvenile
crimes. In fact, the Soviet censor had translated the telegram into
Russian, and then back into English. The original English telegram read
Your daughter was suspended for delinquency. The words suspended
and delinquency had different prototypes (different spheres of
meaning) than the Russian words they were paired with.
Words in other languages will make different distinctions. Some will
encompass a wider range of meanings than corresponding English words.
For instance, Rick Harrison's planned language Vorlin has some interesting
words: the basic sense of the word bat is "a ball-hitting
tool", with its radiated meanings including "bat, hockey stick,
and tennis racket", while the word sop means "soup"
and "stew". Other Vorlin words cover a smaller range of meanings,
so that for, "form, shape", does not include other
senses of English form like "a paper document to be filled
in", "a molding to be filled with concrete" or "manners
or conduct".
As you determine what the words in your model language mean, you have
to keep in mind that they will not exactly equal English words. But, as
a practical matter, you probably don't want to create words for each separate
meaning of brother and have each word assigned only to that meaning,
like the following English words: brother german ("full
brother"), half-brother, stepbrother, brother-in-law,
comrade, fellow, kinsman, fraternity member,
coreligionist. Doing so loses much of the flexibility of brother.
As an example, here's how I translated the different realms of brother
into the model language Negasi.
nemi |
[ Nagada nama.] |
A brother, ranging from the meanings of a full biological brother
to a distantly related kinsman, but excluding the broader senses
of fellow man, fraternity member, or coreligionist. |
sanami |
[ Nagada dunama du + nama, "near
brother".] |
A half-brother, a stepbrother or a brother-in-law. |
henami |
[ Nagada hanama ha + nama, "far
brother".] |
A comrade, fellow or kinsman. |
sanemi |
[ sa + nemi, "near brother".] |
A best friend. |
lunanemi |
[ luna + nemi, "divine brother".]
|
A coreligionist, though for this imagined culture it would refer
to a specific religion. |
Translating the English word comrade into Negasi henami
will result in totally different associations. The Negasi view comrades
as close to kin, and their word that would be translated comrade
of course has no taint of communism. So the word henami has a
stronger familial association than English comrade does. While
an English-to-Negasi dictionary might list henami = comrade,
this oversimplifies the relationship between the prototypes represented
by each word. The words intersect; they are not mutually inclusive.
As an aside, please note the difference between sanami and
sanemi. The word sanami was coined in the Nagada language,
so its literal meaning of "near brother" has been forgotten,
since it underwent sound change differently than sa nemi -- "near
brother" as two words -- did. This allowed the literal meaning "near
brother" to be used to coin a new word in Negasi, in this case referring
to "a best friend", a previously absent word meaning.
Another useful example to describe semantic prototypes or semantic stereotypes
is birds: the prototypical bird has feathers and wings and can fly. Yet
penguins, ostriches and Big Bird are considered birds, even though they
can't fly. A duck-billed platypus, on the other hand, isn't considered
a bird, despite the fact that it lays eggs and has a beak; it is not considered
a bird since it has no feathers, no wings and can't fly.
Prototypes are also a more useful way to describe meaning than semantic
primitives, because prototypes embrace the connotations of a word, rather
than just the denotations. The notion of prototypes can be used to show
how words overlap. For instance, the following table roughly summarizes
different types of body builds:
Above-average weight
Below-average weight
|
fat |
obese |
|
|
chubby |
|
plump |
|
|
|
|
|
|
|
thin |
|
skinny |
|
scrawny |
|
The words skinny and scrawny as meanings are subsets
of thin; plump, chubby and obese are
subsets of fat. Noticeably absent are any words for average weight.
English, like many languages, rarely has words to describe midpoints,
only extremes.
Words are often grouped together like this in semantic networks. However,
these word sets can be exceedingly complicated. For instance, the words
used to describe body builds each have complex connotations as part of
their prototypes: plump is used more often to describe food such
as meat and fruits than fat is; similarly, chubby is
more often used for little boys (or girls) than fat is. The word
scrawny also suggests bony; there are other words not
shown on this list that also have connotations and specific uses, like
lean, which suggests "muscular", and slim,
which suggests "tall", and slender, which suggests
"graceful". Connotations are not specified in dictionaries and
are rarely articulated.
Besides connotations, there are habits governing what other words to
use with a word. The synonyms pursue and chase are almost
interchangeable, except that pursue is preferred when the object
to be chased is highly desirable - pursue truth, pursue wisdom, pursue
happiness, but chase a thief, chase a bus, chase a fox.
You could write an essay on the difference between he pursued love
and he chased love.
Obviously, it is very difficult to translate these prototypes from one
language to another. Failure to properly account for the radius of meaning
of a word often has comical results, as evidenced by this sign in a Bucharest
hotel lobby: The lift is being fixed for the next day. During the
time we regret that you will be unbearable. Historically speaking,
the word unbearable had began with a strict literal meaning,
but over time its radius of meaning had expanded to include a figurative
sense as well. The combinations of connotations and detailed usage preferences
for any word are not articulated, but mastering them is one of the hallmarks
of literary writing.
Meaning is therefore a combination of prototypical examples. The important
thing to keep in mind, when creating your own languages, is that the words
you invent will not exactly equal the English words you define them as.
While, for practical purposes, you will define most words very straightforwardly
in English, you will want to highlight the unique culture of your language's
speakers by noting how the range of meanings and the range of possible
uses distinguishes your invented lexicon from English words.
No wonder Suzanne Vega sang, "I won't use words again. They don't
mean what I meant. They don't say what I said."
Contents copyright 1995 Jeffrey
Henning. All rights reserved.
Last updated: March 1996
Original HTMLizing by Piermaria
Maraziti