Date: Sat, 6 Dec 1997 17:21:34 +0100
Subject: Re: EvolLang: on humanness of language

> Sent by: R.G.
> > Sent by: Waruno Mahdi
> >
> > Yes, for some reason, people say "language" when they mean "human
> > language", but then think nothing of saying "machine language"
> > (which is quite "non-human").
> What is non-human about machine language? It is invented by humans
> for devices used/designed by humans. Humans input to the machine, the
> machine communicates in its own way (machine language) to other
> components, the machine outputs to varying degrees of user-friendly
> formats. All these are very human stories seems to me.
The stories are human alright (even very much so :-)), and the machines are human-made and human-operated, so far so good. It is the "language" of these machines that is non-human, for many reasons. The trivial ones are: 
(1)  words in machine language, like in non-human animal signaling, have strictly defined meanings, in human language they are characterized by polysemy (as a consequence, machine language and non-human animal signaling cannot tolerated homonyms/homographs). In non-human animal and computer signaling systems there is a finite, concrete number of well-defined meanings for each of which there is a defined expression. In human language the meanings are not strictly defined and numerable like that, and there is no such strict tit-for-tat relationship between symbol and symbolized (we have synonyms, metanyms, homonyms, figurative speech, etc.). 
(2)  machine language, like non-human animal signaling systems, cannot change without loss of functionality. When a new version of a machine language is introduced (new dialect in animal signaling), it cannot be used with older machines, whereas the new machines cannot handle old programs unless the old language persists as subclass of the new version. Human language, on the other hand, not only changes without stop, but each speaker is constantly code-switching between several social dialects, including the age-group dialect of the own generation, that of the previous generation, and optionally a professional slang, a regional dialect, etc. (S)He can also cope with situations in which several of these dialects are mixed. Machines would go into a stupor, from which they can only be extricated with a reboot. Analogically, we may even understand a foreigner speaking broken English, or a drunkard. Machines likewise play dead in such situations. In some European languages you can say "he didn't go, they went him" (meaning "they made him go"), although it is just as ungrammatical as it is in English. Machines don't put up with such disdain for syntax. 
(3)  machine language does not distinguish styles (archaic, poetic, bookish, colloquial), and I don't think non-human animal signaling systems does either, although it is imaginable for the animals, that signal quality may express emotionality or some other condition. 
So you see, machine language is much more animal (non-human-wise), than it is human. After all, you can also talk with your pet. In fact, I've seen pet dogs and cats communicate with their owners much more "human"-ly than my computer does with me (or anybody's computer with anybody :-)). When compared with human languages alone, machine language comes closest to artificial languages (e.g. Esperanto), but this will only last so long one doesn't use such an artificial language as natural language. The moment one does that, it will transform into any normal flexible, variable, changing, idiosyncratically irregular, dialect-diversifying natural language (look what happened to Hebrew since the founding of Israel), because only then does the "human touch" come in. 
Those were the trivial points. The point that I see as being the principle (not just principal) one is: 
(4)  Every utterance in a human language first of all establishes a social relationship between speaker and listener/reader, i.e. it is an act of social communication, and only secondly, optionally, does it convey some informative content that can be infered by a formal analysis of the code. Even when you are reading a lecture, you are establishing yourself as lecturer and your listeners as students, and, depending upon whether you strike a more mentoring or a more jovial tone, you also indicate how you would like to see this lecturer-student relationship. It doesn't matter, how many percent of your students will understand your lecture. They'll all understand the social part of the message. Being impersonal in one's speech is not easy (or, when it come's naturally, one should perhaps consult a psychiatrist). That is perhaps one reason why speaking announcements into the intercom requires appropriate training (try let some unschooled layman announce something over the intercom). In machine language it is the other way round. It is the formal content that counts. The flowers may be inserted after special "comment" signs for the benefit of the (human) programmer. Such a "comment" sign indicates to the machine that it should ignore everything that follows in that line.... 
And connected with this point is another: 
(5)  withdrawal from human language communication can lead to mental depression. One form of mobbing is that none of the colleagues speak with the victim anymore. But if nobody uses a computer, it doesn't suddenly break down. Non-human animals, in this regard, seem to be closer to humans than machines are. 
The main reason why people consider machine language to be closer to human language, seems to be that both have syntax and, as a consequence, the limited number of available words can be organised into more or less complicated meaningful sentences, and varieties of these can in turn be ordered in sequence to build lengthy monologues (programs). But, although we do not seem to know of any non-human animal signaling system with elaborate syntax, I don't think we should as principle exclude the possibility of syntax in animal signaling at sub-human levels. I don't know, for one, whether one can safely exclude syntax in whale and dolphin signaling (I'd be grateful for comment from biologists in the know). I also don't know what came first in human evolution, syntax or conscious social organization (other than by biological instinct). 
Apart from that, of course, machine language syntax differs from human language syntax as indicated in (2). When Sapir said that all grammars leak, he meant grammars in human language. Machines cannot cope with leaking grammars. Theirs don't .