Re: Survey on criteria for judging something to be a writing system
- First, of all, thanks to all of you who completed this short survey. I got 38 responses in all, from people with a variety of backgrounds.
Participants were asked to rate, on a scale of 1 to 4 (very unimportant, to very important) the following factors in making something "look like" a writing system:
1. How important is the linear arrangement of symbols in assessing whether a symbol system is a writing system?
2. If the symbols are pictographic, how important is it that they be abstract rather than realistic?
3. How important is the length of the "texts" to a judgment of whether a symbol system is writing?
4. In order to judge something to "look like" writing, how important is it that symbols repeat?
5. How important is the number of distinct texts to a judgment that a symbol system is writing?
The results were for the most part unsurprising. The modes, number of respondents (out of 38) choosing the mode, and the means for each of these questions were as follows:
Linearity Very important 16 3.24
Abstractness Swht unimporant 17 2.21
Text length Swht important 16 2.92
Repetition rate Very important 18 3.29
# of distinct texts Swht important 18 3.00
Thus it was considered to be very important that symbols be arranged largely linearly, and that symbols should repeat in texts. The length and number of texts was considered to be a somewhat important factor. And abstractness of symbols in pictographic systems was considered to be somewhat unimportant: presumably enough people are familiar with Egyptian and other highly pictographic scripts to know that a system can look like a bunch of pretty pictures, yet still be writing.
I also asked respondents to give a bit of information about themselves, and in particular what writing systems they were familiar with, and which ones they considered themselves an expert on. On the basis of the answers, I attempted a crude classification of respondents into three bins:
0. largely unfamiliar with writing systems (5 respondents)
1. familiar with and/or expert on several writing systems (22 respondents)
2. familiar with and/or expert on several writing systems from a variety of writing system types (11 respondents)
There was no significant interaction between group affiliation and the five main questions in the survey. The largest difference between group 2 and groups 1 and 0, was on question 3, where members of group 2 considered length of texts a somewhat *less* important factor than members of the other groups.
Finally I asked respondents for other suggestions on other features that might be relevant to deciding if a symbol system looks like a writing system. I quote directly some of the suggestions below.
First, a few suggestions that relate roughly to the statistical distribution of symbols:
- How important is it that symbols are made on an object such that they could not have been made randomly (i.e., as an expression of a thought rather than meaningless scribble)?
- Statistical distribution of signs should accord with statistical distribution of graphemes, syllables, or words in attested languages
- The existence of random combinations of the symbols. If we are lucky to have a long enough texts, or many texts, the existence of random combinations that repeat themselves.
- The number of variations of symbols. How many are there? Does it correspond at all to the number of sounds in a given language?
Next, a couple of suggestions of the importance of provenance:
- Provenance, resemblance to known systems
- Archaeological setting. No provenance will tend to make me highly suspicious.
Finally, one suggestion having to do with the degree to which cursive styles have developed, suggesting a long tradition of use:
- Degree of cursivity; a fully evolved system should show signs of use, and therefore influence of the writing instruments, medium, and human dexterity.
Center for Spoken Language Understanding
Oregon Health & Science University
--- In ANEemail@example.com, "richardwsproat" <rws@...> wrote:
> Many people who are not experts on writing systems seem to have a sense of when
> a symbol system "looks like" writing, as opposed to some sort of non-linguistic
> symbol system. Even when the system's true nature is unknown, non-experts may
> feel that it is probably a form of writing, because for some set of reasons or
> other it "looks like" writing.
> I am interested in understanding what factors might make a system "look like"
> writing to the layman.
> To that end I put together a short survey:
> It should take less than 5 minutes to complete, and in most cases probably less
> than 1 minute.
> Richard Sproat
> Center for Spoken Language Understanding
> Oregon Health & Science University