At the outset of this article, it was observed that the adequacy of an etic typology of written symbols could be judged by its ability to describe all the emic distinctions in all the writing systems of the world. In conclusion, we should like to return to this point and briefly examine the extent to which currently available etic concepts can be used to describe the distinctions made by Western Apaches in relation to the writing system of Silas John. Every symbol in the Silas John script may be classified as a phonetic-semantic sign. Symbols of this type denote linguistic expressions that consist of one or more words and contrast as a class with phonetic-nonsemantic signs, which denote phonemes (or phoneme clusters), syllables (or syllable clusters), and various prosodic phenomena (2, pp. 2, 248). Phonetic semantic signs are commonly partitioned into two subclasses: alogographs (which denote single words) and phraseographs (which denote on or more words). Although every symbol in the Silas John script can be assigned to one or the other of these categories, such an exercise is without justification (21). We have no evidence to suggest that Western Apaches classify symbols according to the length or complexity of their linguistic referents, and therefore the imposition of distinctions based on these criteria would be inappropriate and misleading. A far more useful contrast, and one we have already employed, is presented in most etic typologies as an opposition between compound (composite) and noncompound (noncomposite) symbols. Used to break down the category of phonetic-semantic signs, these two concepts enable us to describe more or less exactly the distinction Apaches draw between "symbol elements put together" (ke?escin ledidilgoh) and "symbol elements standing alone" (ke?- escin doledidildaahi). The former may now be defined as consisting of compound phonetic-semantic signs, while the latter is composed of noncompound phonetic-semantic signs. Up to this point, etic concepts have served us well. However, a deficiency appears when we search for a terminology that allows us to describe the distinction between "symbols that tell what to say" and "symbols that tell what to do." As far as we have been able to determine, standard typologies make no provision for this kind of contrast, apparently because their creators have tacitly assumed that systems composed of phonetic-semantic signs serve exclusively to communicate linguistic information. Consequently, the possibility that these systems might also convey nonlinguistic information seems to have been ignored. This oversight may be a product of Western ethnocentrism; after all, it is. we who use alphabets who most frequently associate writing with language (22). On the other hand, it may simply stem from the fact that systems incorporating symbols with kinesic referents are exceedingly rare and have not yet been reported. In any case, it is important to recognize that the etic inventory is not complete. Retaining the term "phonetic sign" as a label for written symbols. that denote linguistic phenomena, we propose that the term "kinetic sign" be introduced to label symbols that denote sequences of nonverbal behavior. Symbols of the latter type that simultaneously denote some unit of language may be classified as "phonetic-kinetic" signs. With these concepts, the contrast between " symbols that tell what to say" and "symbols that tell what to do" can be rephrased as one that distinguishes phonetic signs (by definition nonkinetic) from phonetic-kinetic signs. Purely kinetic signs-symbols that refer solely to physical gestures-are absent from the Silas John script. The utility of the kinetic sign and the phonetic-kinetic sign as comparative concepts must ultimately be judged on the basis of their capacity to clarify and describe emic distinctions in other systems of writing. However, as we have previously pointed out, ethnographic studies of American Indian systems that address themselves to the identification of these distinctions-and thus provide the information necessary to evaluate the relevance and applicability of etic concepts-are in very short supply. As a result, meaningful comparisons cannot be made. At this point, we simply alack the data with which to determine whether the kinetic component so prominen in the Silas John script is unique or whether it had counterparts else-where in North America. The view is still prevalent among anthropologists and linguists that the great majority of American Indian writing systems conform to one or two global "primitive" types. Our study of the Silas John script casts doubt upon this position, for it demonstrates that fundamental emic distinctions remain to be discovered and that existing etic frameworks are less than adequatelyequipped to describe them. The implications of these findings are clear. On the one hand, we must acknowledge the possibility that several structurally distinct forms of writing were developed by North America's Indian cultures. Concomitantly, we must be prepared to aabandon traditional ideas of typological similarity and simplicity among thes systems in favor of those that take into account variation and complexity.