Some Toughts on Creating a LanguageSince I have been a bit bored lately, I got some thoughts on an old and familiar subject, namely creating a language. I always wanted to do that, silly as it may seem... Of course
I am not an expert, but I do have a somewhat linguistic background, so that as soon as I started thinking about it (and I did this before years ago), the ideas grew complex rapidly. However, I do think that writing them down first is not only useful for me, but for other people trying to construct languages as well. Anyhow, I will keep the notions discussed below in mind when I start the project of creating a ‘Gallente’ descriptive grammar.
I will briefly look at linguistic points that require the attention of someone who is creating a language, but naturally the list is by no means exhaustive. This is not my only point, though, because I believe that constructing a language, especially in this fictional world, is particularly meaningful if it gets a fictional history. After all languages do not appear out of nothing. This etymology gives a language its character, but is also of importance if people ever wish to elaborate on fictional history and describe fictional archaeological findings, because language is everywhere.
Let us begin with stating the obvious: language changes. It changes all the time, and this cannot be stopped, although mass media, education, reading, writing and institutional and governmental prescriptions can slow the process. In only some millennia Proto-Germanic has become virtually unrecognizable to speakers of Germanic languages, and this gap grows even stronger the further one goes back in time. Scrolling through the list of reconstructed roots of Proto-Indo-European (PIE), it is unlikely one finds a lot of recognizable features. And then we are still only talking about lexis, not even about grammar. PIE is said to be a real language six thousand years ago. Now, if one reasons the other way around, it is extremely unlikely that anyone can find something of modern earth languages after sixteen thousand years of change.
The second observation is more comforting, and that is the fact that language does not change completely random and totally unpredictable. That is, there are certain rules and discerned processes that can be abstracted from the on-going language change. Some changes are more likely than others.
Lexical changeWords come and go, so reconstructed forms should contain words that are lost in later phases. This can be a language internal process. It can, on the other hand, also be the consequence of contacts with other languages. Bear in mind, though, that loanwords occur especially when the ‘borrowing’ language does not have a word for the object the word denotes, or when the source language has more prestige. One has to account for all this.
Some words are more resistant to change. Frequently a vast group of normal and basic nouns and verbs remain fairly intact. Auxiliary verbs change more often in form and meaning, sometimes becoming affixes. The more abstract a word in its meaning is, the more likely it will semantically change. Another group of words that remain pretty constant is that of the proper names. Names of people and places can therefore be very archaic. Lots of names for towns and natural phenomena are hard to explain and understand. Some linguists even conjectured a pre-Indo-European source for lots of names for natural phenomena in Europe.
A third point is the effect of substrates. Language communities are not secluded from each other. There is contact and eventually migration. This can result not only in the borrowing of many words pertaining to certain sociolinguistic fields, but also in change of grammar. One language may adopt structures that are typical for another. Compare for instance the effects of English on Afrikaans.
Finally a remark on the shift of meaning. In general there seems to be a tendency of neutral words to get negative connotations in the course of time. This is for example the case with Middle Dutch
wijf (‘woman’, cognate English
wife). It eventually became negative, meaning something like ‘hag’. A new neutral word came instead,
vrouw. It thus becomes clear that lexis is also influenced by semantics.
In all, some elements from e.g. German and especially English ought to be incorporated in the data that is used to construct Gallente.
Phonological changeNB: our alphabet does not have enough signs to represent all the sounds. For that purpose the International Phonetic Alphabet (IPA) has been developed. I will use it here. See
http://en.wikipedia.org/wiki/IPA.
All the sounds a human can make (and this is naturally limited) have specific ways of articulation. Some are made with the lips, hence they are called bilabial. Others are made with the point of the tongue, hence they are called coronal. And so forth. Now, if we have a coronal, say an R [r], and we wish to look at change, then we need to look at the articulation that changes, and therefore R is more likely to become L [l]. Thus from R to P [p] is rather absurd. On the other hand it is context that plays its part. Specific contexts with specific sounds may trigger specific changes in vowels and consonants. Think of the process of assimilation, where one sound gets one or more features of a nearby sound. One simple example is N becoming NG [ŋ] when followed by K: now NG and K are articulated pretty much at the same spot, while NG is still a nasal sound like N. This kind of change is partly due to the strive for efficiency in speaking. The other claim made above is the regularity of phonological change. One example can be found in Dutch, where Old Dutch [ u ] became Middle Dutch [y:] (a sound English does not have), and this in turn changed into Modern Dutch [ʌi]. This is regular. However, there is an exception, namely when [y] was followed by [r], and once again this exception is regular. Hence we get
huis (‘house’) versus
muur (‘wall’).
So if we take the French word
préférer, it could change into:
prefer (with the first syllable stressed) > prefɪr > prɪfɪr > pɪrfɪr > pɪrfi etc. We see that [e] becomes [ɛ] (as in
bed), and [ɛ] becomes [ɪ] (as in
hit). Moreover there is progressive assimilation of the first vowel to the second. After that, the desire for regularity in syllable structure comes in and causes metathesis, where the r shifts to the front. In the end, however, this regularity is obscured by another type of change, deletion. After all, it is the first syllable that is stressed, thus it is quite easy to lose material at the end of the word (a process attested in Germanic). If a language were to be created, these steps should be used for other words as well (e > i; loss of final r).
MorphologyChanges can be regular. In Germanic, the stress shifted to the first syllable of words (affixes not included), which made it possible that information at the end was lost. At the same time, the ‘full’ vocals in the latter syllables became nothing but schwa’s [ə]: an example of lenition. This explains why Old Dutch
vogala (‘bird’) became Middle Dutch
vogele [voxələ], and finally
vogel. Nonetheless processes like these cannot continue forever, making words smaller and smaller, otherwise we would be left with nothing after some millennia. Affixes (e.g.
be- and -
ing) are one possible solution (but not the only one).
Take the aforementioned example of
préférer. This became the fictional
pɪrfi. For reasons of harmonization this might change once more in a syllabically rhythmic
pirif. By now the word probably has altered conjugation as well, so this form is now a stem. What can happen next is the addition of the prefix
a (‘by’):
apiri. Semantically the meaning of
prefer + by is not strange at all. It happened before. In the course of time this prefix can become transparent and thus becomes a fixed part of the word, and is incorporated into the stem. This enables the next forms:
apir [ɑpir] > hɑbir > hɔbər > xɔbrɛ > go:brɔ > ko:bjɔ > ko:vjɔ
This final product stands many thousands of years away from current French and has become unrecognizable. The trick with affixes can also be done with auxiliaries and even articles, who become clitics, prefixes and suffixes. The addition of these parts may trigger changes in stress patterns, thus leading to schwa forming etc.
Two stray notes. The first is that there seems to be a tendency of synthetic languages (e.g. case systems; Latin) to become more and more analytic (loose words; Italian, although Chinese is a far better example). At the same time verb conjugations become simplified and irregular. Think of English, Swedish and Afrikaans. Mind, though, that again words can only become so small. ‘After a while’ other elements may get attached so that the whole cycle starts again: morphological cyclization.
The second is that a language can be accusative or ergative. Most Indo-European languages are accusative, i.e., they mark the Object of a transitive sentence:
I see him, where
him (‘he’) has the accusative case. This is by far the most common way of marking. Rather marginal is the ergative option: marking the Subject in a transitive sentence:
I(+ case)
see he. Subsequently one should ask himself to what extend the language has a case system at all.
Furthermore there are questions like: does the language have past and future tens? Because not all languages have all tenses, like Afrikaans. It then must be abstracted from the context, by using additional adverbs like then and yesterday. In some languages tenses are explicated by morphemes on the verb, but other languages use auxiliary verbs, like Dutch:
Ik zal dat doen (‘I will do that’), versus Italian: ‘
Lo (
io)
farò …’
SyntaxNaturally a sentence is more than just a string of words. It has a structure. Language is a hugely complex system of rules, and these rules change all the time as well. For this reason it does not suffice to merely build up a vocabulary.
There are several types of sentences, such as declarative, interrogative and imperative ones. They may have different word orders. One way to label word orders is to look at the positions of Subject, Object and Verb (SOV) respectively. English is a harmonious SVO type (a very common type indeed):
I look at him; I looked at him?; Look at him! Of course then there are the passive constructions as well…
Synthetic languages usually allow words to be moved around in sentences more than analytical languages, simply because the words keep their case, and the case makes it clear that it concerns a Subject, Object or Indirect Object. It follows that in a ‘caseless’ language the word order becomes fixed.
If one wants to construct and make clear syntactical rules (word order, it is best done by glossing. There are conventions for how to do it.
http://www.eva.mpg.de/lingua/resources/glossing-rules.php.
Additional points that require attention is how complex sentences are made, and if these subsentences get another structure or not.
In generalIn general it is worth looking at the language universals.
http://en.wikipedia.org/wiki/Language_universals. Think of the banal universality of verbs and nouns. For example, usually an SOV language tends to have postpositions, and not prepositions. This class of words consists of e.g. in, from, towards and for.
Now ExampleTo make the matter concrete I have made an example, but this is not final. It is just a sketch to serve as an illustration. Also, I do not know much about French (this semester I am going to do a beginners course), so please excuse me for possible errors.
Modern Standard French
1. ‘Le cadre du projet est défini par des principes fondateurs.’
Old Gallente (10.000 AD)
2. i chor du proi e-depi-n par i prichip podatur
the scope of.the project is-define-PAST by the first founder.PL
IPA: [i čɔr du prɔi edəpin phɑr i pričip phɔdatur]
‘The scope of the project is defined by the first founders.’
Modern Standard-Gallente: same sentence
3. se šavos as-aivelai asθuza an fes as-(e)vuz-e
is define DET-project scope by first DET-founder-PL
Observed changes are:
Phonological changes:
p > b > bh > v, or f when initial
r > z > s > š > j > l, but deletion when followed by other coronal.
č > t > θ
r > z, when in between vocals
ɔ > ɔi > ʌi > ʌ > u
Morphological changes:
There are remains of clitization and obliqueness of affixes:
as- signals a determined object, built up by the now lost particle
i, ‘the’, and an unknown -
s-, which can be found elsewhere too. It may also be derived from Proto-Gallente *
a, ‘is’.
The stem
vuz- has an allomorph,
evuz- after a prefix, possibly to avoid clustering of consonants.
Syntactical changes:
Originally the language must have been irregular in its base order of Subject, Object and Verb. Nowadays the order is perfectly SOV, accept for passive sentences, where it is exactly the opposite. The Verb is placed in front, followed by the Object, which of course is the focus in a passive sentence: VOS.
Lexis:
Also note vocabulary changes:
depi- is replaced by the (indigenous!) word
šavos.
Asaivelai is a good example of how changes work. French project became
proi. This in turn changed as follows: ploi > hflɑi > həfelai > hafelai > avelai > aivelai. We now know to cut the word into these morphemes:
as- and
aivelai. Initial vocals are rather rare in Proto-Gallente, and it is conjectured that it arose from aspirating words (initial
h), but this can only be so for short vowels, and not for the diphthong
ai. Therefore this later
ai is likely to be a form of progressive assimilation with final –
ai. In the process, f became v because it stands between two vowels. Furthermore we see a tendency to avoid clustering of consonants, because of the inserted
e:
hfl is awkward. And indeed the Proto-Gallente phonotaxis does not have
vl-clusters. It is likely that future developments result in a harmonious CV-structure (consonant-vowel) in Gallente.
NB: reconstructions are always marked with an asterisk *.
PS: for reasons of legibility I attached this as a Word-document.
[attachment deleted by admin]
Author
Topic: Some Thoughts on Creating a Language (Gallente) (Read 3644 times)