Symbol Project

Note: this appears to be a dead project; URLs pointing to symbolwiki.org on this page are BrokenLinks.

Symbolproject - is now SymbolWiki - http://symbolwiki.org - a site for collaboratively crafting an ideographic character system (or two, or three), in a wiki environment. It was formerly part of the TaoRiver WikiHive, and now runs on a MediaWiki hosted at MyOo.de.

A more complete, and up to date version of this discussion is at http://www.symbolwiki.org/index.php?title=Symbolwiki:Critique. -- KuniShiro? See also HieraticLanguage and SymbolicUserInterface


This project isn't going to work in its current form; it is founded on linguistic principles that are uncontroversially known to be false. Chinese writing is not based on pure ideas, it is based on (Chinese) words, as is every written language - for instance, ancient Egyptian hieroglyphs represented spoken Egyptian words. Pure pictographic systems that attempt to represent ideas, rather than words, are not able to express the same breadth and depth as do spoken languages, for precisely the same reason that word-for-word literally translations of one spoken language to another fail. All written languages began as pictographic systems, but evolved into symbol systems with an approximately one-to-one connection with spoken language for functional reasons.

In every language, spoken and written words represent multiple concepts, often a large number of concepts, that must be disambiguated by context, but also by conventions that are complex, culturally specific, language specific, and which have never been fully analyzed even for one language. Similar problems apply to grammatical categories and relations, which are also overloaded and just as difficult to analyse.

-- DougMerritt

Assuming pictographic language systems evolved into symbol systems, pictographic systems could be seen as a "seed language". Modern languages are symbol-based, but if a global pictograph system was built(and there is plenty of prior art), and was recognizable by most earthlings, then wouldn't a symbolic language evolve out of that? We'd then have a global language(that wasn't esperanto). Such a project might fail, but I don't see any technical reason behind it. I think the biggest hurdle is that a universal language wouldn't be designed so much as distilled. -- LayneThomas

(Cutting in) Well, the "form" of the symbol project - its process, and even its scope are ever-changing and adaptable. That a language system should conform to an abstract term like "ideographic," is a bit silly, and I should be more careful to correct that the goal is the production of an "ideographic language" - but its not exactly incorrect, either. The goal is a platform for agreeing on a conceptual symbols - symbols which represent a convergence of terms across the languages of its contributors. That a symbolic glyph must represent a spoken word is somewhat obvious - by default, the most efficient symbols for language are spoken words - written words fall far behind, and among written forms alphabets are far more efficient than ideographs, pictographs, logograms, etc.

:But alphabets are efficient only for their localized language. An [[SP:auxlang|]], in order to transcend various low-level spoken language, must reasonably be at a higher level - the symbols to represent these higher-level concepts need not be pictographic, logographic or ideographic - they simply need to be agreed upon and efficient. The web allows us to be more efficient with written words than we ever can be with a radio, for example. I agree that there are a lot of unresolved issues - but the point is that by a collaborative, multilingual process can serve the function of pointing various distinct and apparently incompatible symbols toward the goal of a single unique symbol, accessible from any angle.

:The issue of structure must be explained of course, but the development of linguistic studies (as humble I come understand it) is the rejection of of localized notions like grammar, and the simplification of the language concepts down to particular symbol relationships, following a structure for the normal order in which they are used. So "green and colorless" are not, (in most cases) properties of "ideas." Regardless of what language you use, this is true. Even if the term "green ideas" is a regular idiom in some particular culture - the underlying meaning is about something other than color. The symbol project needs to account for idiomatic issues, certainly. But I submit that idioms are not as incompatible as people may believe. Although some cases are rather extreme - like Nikita Kruschev's "we will bury you" comment did not mean 'we will destroy you,' rather meant "we will outlive you." -- StevenCooney (www.symbolwiki.org )

On top of everything else, it's not like this hasn't been tried before. I used to be deeply interested in constructed/artificial languages, and studied many of them individually and also their general history, participated in email lists, worked on designing them, etc. Speculate all you like, but if you want success then you should study "the technical reasons behind it", beginning with modern linguistics and then continuing to technical case histories of important conlangs such as Esperanto and Loglan/Lojban.

Note that Esperanto is the most successful conlang ever, with an estimated one million worldwide speakers, and yet it failed in all of its goals (except of course that its existence as a language succeeded, which is far from true of conlangs in general).

Same thing with the SymbolProject. At best it will succeed in existing (although that is far from certain), but it is guaranteed to fail in all of its other current goals, again because they are based on 19th century naive assumptions. It's like trying to design a better set of roman numerals so that people will intuit arithmetic better; it misses the point entirely due to missing out on all of the important state of the art.

-- DougMerritt

What important state of the art exists now that didn't exist when Latin and Chinese were being developed? Linguistics. Language is emergent. There is virtually no way to design an icon language from scratch - but over the years we have learned to read icons, even speak them. Not because it was designed, or analyzed by linguistics experts, but because it is used by many. Perhaps you are referring to Chomsky's(mostly discredited) deep structure?

Chomsky sucks, but it wasn't Chomsky's deep structure that was discredited, it was the extremely wide misunderstanding of it that was discredited. I won't even try to address anything Chomsky has said since roughly 1970, and some things before then, however, because he is incoherent. His notions about language innateness, for instance, are simultaneously overly vague yet also self-contradictory. If you want to know about modern linguistics, avoid Chomsky and his followers like the plague. Aside from his sycophants, other linguists consider Chomsky to be an actively negative influence on the progress of the field (in recent decades, that is; his original contributions to linguistics are undeniable - but are also 99% concerning his hierarchy of grammars).

And no, we have not "learned to read icons, even speak them." That's what I mean by needing to know modern linguistics. Is it so hard to believe that the bazaar method of visual language creation would eventually allow a distillation of universal iconic concepts? I'm a believer in the cathedral and the bazaar; obviously it can yield a visual language creation with iconic concepts. It is the "distillation", "universal", and natural-language-equivalent-in-power parts that are naive. Also see SymbolicUserInterface and the MediaGlyphs? project( http://www.mediaglyphs.org/ )

The top critique is really a criticism of the conlang and I agree with it, where it applies to monolithically constructed conlangs. But where WikiWiki collaboration is the central design principle for a "conlang," it should quite naturally transcend all the criticisms of conlangs being stale. All that the symbol project espouses is the use of wiki technology (i.e fast and worldwide collaboration) to facilitate the "distilling" (as LayneThomas picturesquely calls it) This use of wiki seems like a no-brainer to me, considering where this discussion is. And "the process" of developing and re-developing this symbol library as I imagine it would be perpetual. -StevenCooney

The MediaGlyph? project appears potentially more attainable, but I see a large number of glyphs that violate their own design principles in ways that are typical of linguistics-unaware amateurs. I mean, I'm an amateur too, but at least I've studied linguistics thoroughly. Being completely innocent of a subject, on the other hand, pretty much guarantees naive mistakes. Look at the "grammatical" glyphs for instance. For pity's sake, they labeled them with English suffixes and Chinese characters, since the icon itself was completely opaque. They forgot what the goal was. The project would be extremely difficult even if they never forgot. Pathetic.

Rooting it partially in sign language is not a bad idea, except that people misunderstand the linguistics of sign language just as much as they do that of pictographs. Sign language, also, is not a direct representation of concepts, it is a full language with grammar and symbols, and one of the key things there is that grammar and symbols are arbitrary, not connected one-to-one with ideas, despite the concrete iconic basis of a certain number of nouns, verbs, and grammatical parts in most sign language; the latter are the exception to the rule, yet linguistics amateurs are always full of hope that they are the rule rather than the exception.

And not only "are arbitrary, in general" but also "must be arbitrary, in general". Deixis is insufficiently rich to represent a full language.

So the theory is that direct representation of concepts is counter to an arbitrary symbolic language? i.e. One is a symbol of a concept, one is a symbol that allows construction of a concept. The assumption then, is that the construction symbols are more powerful? This seems awfully close to low-level language(assembly/construction) vs high-level (c/java/concrete concepts). Obviously both have a place, and high-level languages have definitely survived. It seems that an iconic(direct concept) language would naturally arise from a constructive language. So if an iconic language was based on powerful spoken/written/common language(s) - the icon language would survive as it represents high-level concepts. Indeed, many icons are near universally recognizable. Obviously, a high-level language cannot have the same ability to fine-tune a statement, but that's not the point. A high-level iconic language just needs to abstract away some low-level details. Perhaps it is a matter of how succinct an icon language can be, and what the problem domain is(user interface, international encoding, a language for kids, etc)

''I email with Giuseppe (MediaGlyphs?) and we agree on a number of things - But symbol project focuses on an efficient "writable" interface, where MG has a much broader focus. It was a logical step to borrow from Chinese for SP, but some additions/amendments to the basic building blocks (strokes) are also viable, if there is good reason. I am quite open. This above is fairly good analysis. -StevenCooney

In a nutshell: the link between symbol and referent of symbol must be completely arbitrary, not based on real or imagined resemblance, or else the system will be unable to represent the things that human languages represent. This means that it is irrelevant at best, and usually highly misleading as well, to even start with a symbol system which is pictographic, onomatopoeic, sign language deixis, metaphorical, analogic, or in any way has anything but a random connection with what it references. To actually hope that a symbol system can not only begin that way, but continue that way up to the full power of a natural language, is naive. In this context, "symbol" specifically includes parts of speech (noun, verb, adverb, adjective, etc), and grammar rules and morphology as well. I.e. it includes everything that human language includes, not just, say, nouns.

I am mildly reluctant to say "impossible even in principle even for non-human brains", but I am very comfortable to say "impossible even in principle for designers who are not familiar with the history and theory of the subject". Which is not limited to linguistics; one should also be aware of the vast history and literature that relates from philosophy (semiotics before the post-modern crap), mathematics of semiotics, and much much more. Stephen Pinker's "The Language Instinct" is a good survey, although still not adequate by itself.

You mentioned high-level notations. Here's a much less difficult challenge (i.e. is probably possible): instead of low-level chess notation (either the older positional or the modern algebraic), invent a high-level language to describe chess games, which is descriptive enough that it can meaningfully be "compiled" into the usual low-level notation. I've been thinking about doing so for a quarter century and still can't figure out a way to do it that isn't uselessly vague. -- DougMerritt

It seems like the above is trying to argue a connection between alphabet efficiency and high-level language. I agree that an alphabet is most efficient for the task, and this leaves more braincycles for dealing with conceptual, rather than the translation of symbol to concept. But alphabets are perfectly useless for the ideal auxlang. So for example, English is working fine for this conversation right now, but its doing little for someone who cant read English. The chess example is a nice one, but rather misleading - the current format can be compacted significantly, but no more than the data loss principle can allow. The symbol project thesis (Thank you all for the valid criticism, by the way) isn't that the rules for sentence completeness are condensable, rather that humanity's "uniqueness" (commonality) demands that a common symbol system, to use in lieu of localized symbols, be commonly devised - to be used when necessary or convenient. Also, I'm not unaware of the perceptual differences between generations with regard to the substance of their conventional conventions, and the difficulty in accepting that these naturally (WikiNature in this case) will change. -- StevenCooney

Continued at http://www.symbolwiki.org/index.php?title=Symbolwiki:Critique. -- KuniShiro?


CategoryWikiFarm CategoryWikiForum


EditText of this page (last edited November 4, 2014) or FindPage with title or text search