I will discuss the following three problems: the apparent lack of phonetization and standardization of the proto-Elamite signary and its consequences, the similarity of some proto-Elamite and linear-Elamite signs along with the questions of the transmission and use of of non-cuneiform writing in Iran during the 3rd millennium BCE, and finally, the asymmetric distribution of signs and "decipherable" strings of signs in the linear-Elamite corpus noting the likelihood that not all of these texts belong to the same corpus.

Evidence of many types has accumulated in the last half decade that the so-called Indus script was not a speech-encoding or writing system in the strict linguistic sense, as has been assumed since the first Indus artifact carrying symbols showed up in the 1870s (Farmer, Henderson, and Witzel 2002; Farmer, Sproat, and Witzel 2004; Sproat and Farmer 2005). This find is radically changing views of the oldest Indian civilization and has awakened new interest in issues involving non-linguistic symbols, pseudo-decipherments, and related topics addressed in this Workshop. In this talk I illustrate a variety of nonlinguistic symbol systems, propose a typology of scripts and non- scripts based on semantic and phonological grounds, and discuss the light recent studies of Indus symbols throw on ways of distinguishing the two classes of symbols. The talk ends with a discussion of ways in which our archaeological understanding of Indus society is deepening as a result of abandoning the traditional Indus-script thesis (Weber, Fuller, and Farmer 2007).

Indus signs remain "unread" in part because the linguistic nature of the signs is in question and since little is currently known of the language or languages spoken in the Indus civilization. However, two contemporary sources can shed light on this: some 30 loan words in Mesopotamian inscriptions from the Indus (or the Dilmun) areas, and some 200-300 words in early Vedic texts composed in the northern section of the Indus civilization. Both point to a language that was prefixing in nature but radically different from agglutinative, suffixing languages such as Dravidian. Even if Indus signs do not encode full phrases or sentences of a spoken language, as recent studies suggest, determining the languages spoken in the region may be useful in interpreting Indus symbols, which may (like heraldic signs, Mongolian tamghas, and similar nonlinguistic symbol systems) contain occasional puns even without systematically encoding language.

In December 2004, Steve Farmer, Richard Sproat and Michael Witzel published (in Electronic Journal of Vedic Studies 11: 19-57) an article entitled "The collapse of the Indus-script thesis: The myth of a literate Harappan civilization." The authors deny that the Indus script constitutes a real writing system in the sense of being tightly bound to language. In this presentation I review their arguments, repeating what I have already stated in a paper published in 2005 (Transactions of the International Conference of Eastern Studies 50: 33-44) and adding some further considerations.

We study a number of natural language decipherment problems using unsupervised computer analysis methods. These problems include letter substitution ciphers, character code conversion, archaeological phonetic decipherment, and word-based ciphers with relevance to automated machine translation. We describe a handful of techniques that improve results across all these problems.

The connection between archaeological decipherment and alignment tasks is not at all new - a few of the more famous decipherments in history, in fact, were based on the alignment of parallel texts, anchored by occurrences of personal and/or place names. "Alignment" as a formal, computational problem, however, has been construed in several different ways in graph theory and engineering, many of which simply answer a different question than the one that the classical bitext alignment problem is asking. In this talk, we'll take a look at two of the most common alignment problems that have been required in the course of a decipherment's maturation, each in two different ways: what they really should be aligning, and how the quality of an alignment should be calculated.

The Phaistos Disk, discovered in Crete in 1908 and thought to date to between 1600 and 1850 BC, is one of the great outstanding archaeological enigmas. Since its discovery there have been no fewer than 20 published "decipherments", ranging in quality from intriguing to truly bizarre. In this talk I will present a variant of a proposal by Kevin Knight for a computational model of script decipherment, and use this to argue that in the absence of another sample of this writing system, it is simply pointless to try to decipher this artifact or any artifact with such a short text. I will also examine and critique a couple of the more recent proposed decipherments for the Phaistos disk in some detail.