The idea that all signs are arbitrary cultural constructs has become popular
among many of those involved in the critical discussion of art. Although a
number of theorists have contributed to this idea, it originates with Ferdinand
de Saussure's assertion of the "arbitrariness of the signifier" (with regard to
the signs for words). Saussure's opinion was based solely on the lack of any
proof to the contrary but it has become the cornerstone of linguistics and a
major force in the philosophical dialogue of our language-driven society.

The use of written words as images in contemporary art is now quite common
so it seems only natural that language-based theories be applied to visual art.
However, to accept written words as images, the true nature of
letter-sequences must be disregarded. The sequence of letter-forms which
convey a word do not qualify as a visual manifestation of that word because,
although a letter's identity can be carried by an image, it does not reside in the
image. The identity of our letters is defined solely by their role in
patterning of lexical sequences.
The same letter can be transmitted using the
different letter-forms of uppercase, lowercase, script, Braille, or even
expressed verbally without altering any features of what we describe as "the
letter" (in the Arial text you're reading, the lowercase "L" and the capitol "i" are
represented by the same letter-form, "l"). Words are inextricably linked to the
sequencing of letters, not the images of letter-forms.

It's undeniable that our letter-forms are historically, culturally, and functionally
an aspect of our written language; their characteristics and context will often
impact the message (just as tone-of-voice impacts the message of spoken
language) but they don't visually manifest the words they convey. McLuhan's
idea of "the medium being the message" doesn't extend to the codification of
language. Although the message of a word can be altered by the features of
the medium in which it is read, it can't be defined by the features of a
dimensionally dissimilar medium unless it is transformed into a manifestation
of that medium--although written language is transmitted in a 2-dimensional
context, its code is not spatially formulated, alphabetic code is 1-dimensional
(a linear sequence unfolding in only one direction). We may learn to visually
process the letters of a word as a unit (this is doubtful, see Notes on Bouma
Theory & Parallel Letter Recognition) but that doesn't transform the structure
of the system or the nature of its content.

If the signs for words are arbitrary, this entire discussion is essentially
irrelevant but if Saussure was incorrect, the argument is fundamental. The
history of the English Language and the evolution of the spelling of its words
tell us any logic in the content of its signs would have to be the result of an
instinctive or intuitive criteria which we adhered to unknowingly--we gravitated
toward certain choices simply because they "felt right" (during the evolution of
Middle English, "line" felt more comfortable than "lign", although some of the
word's meaning was borrowed from the Old French, "ligne").

In 1980 while working on an experimental text-sound composition, I
discovered--quite by accident--that a circle could be used to produce
2-dimensional configurations generated by the 1-dimensional sequence of
spelling. This 2-dimensional transformation is isomorphic
(information-preserving) because it relies on a rigorous process:
the letters
of the Roman Alphabet are represented by a fixed circle of 26
(the fixed locations of the letter-points are defined by organizing
the consonants around a symmetry of vowels--the unevenly spaced
arrangement is rotated so that "E" rather than "A" is at the top, which balances
the symmetry).
When lines are drawn which interconnect the
letter-points in the sequence of the spelling of a word, an abstract form
is generated.
In my view, these spelled-forms illustrate a thematic structure
which could not result from arbitrary associations.