In this highly technical monograph, Robert Crellin surveys the phenomenon of so-called “word dividers” in Northwest Semitic epigraphic texts in an effort to discern what principles, if any, might guide their distribution. It represents a product of the CREWS (Contexts of and Relations between Early Writing Systems) research project based at the University of Cambridge from 2016–2022. The texts surveyed are in Phoenician, Ugaritic, Hebrew, Moabite, and Epigraphic Greek, with Greek included because of the premise that its alphabetic writing system was derived from the same tradition as that of Northwest Semitic. The organization of the book reflects this attempt at a systematic investigation: the four main parts (Part I: Phoenician; Part II: Ugaritic alphabetic cuneiform; Part III: Hebrew and Moabite; Part IV: Epigraphic Greek), each of which consists of multiple chapters, are together framed by an introduction (chapter 1) and an epilogue (chapter 17).
The overarching argument of this book is that word division should be understood as a prosodic phenomenon in the texts in question: “graphematic words [in these texts] correspond either to prosodic words or prosodic phrases” (p. 294). In other words, the function of word dividers was primarily “to indicate the manner of oral delivery of a piece of written language” (p. 7), and thus they do not necessarily align with semantic or syntactic divisions. Following Sproat, Crellin identifies five possible levels at which word division could operate (with the “Orthographically Relevant Level” or ORL denoting the level that is targeted for a particular writing system)—those of semantics, prosody/phonology, morphosyntax, syntax, or graphematics (p. 10). These possibilities are explored throughout the book in light of the primary data from the various languages.
For Phoenician inscriptions (Part I; chapters 2–4), Crellin identifies two systems of word division at work: one targeting the (actual) prosodic word (chapter 3; based on KAI 1, 4, 7, and 24)—which the author sees as corresponding to the situation in Tiberian Hebrew with units joined by maqqef—and another targeting the prosodic phrase (chapter 4; based on the Yehawmilk inscription [KAI 10]). There is indeed a notable difference between the distribution of word dividers in the first group (KAI 1, 4, 7, and 24) and the second (KAI 10), with the first showing a strong tendency toward the use of separators and the reverse situation in the second. The body of these chapters includes discussion of the numerous exceptions and special cases arising in the texts.
Similarly, for Ugaritic (Part II; chapters 5–9), Crellin distinguishes between a ‘majority’ orthography and a ‘minority’ one, with the ‘majority’ orthography targeting (or at least consistent with the identification of) actual prosodic words, and the ‘minority’ orthography (which is found primarily in non-literary documents) making divisions on the basis of morphosyntax—a situation that runs counter to the pattern of focus on prosody elsewhere in Northwest Semitic writing systems. The ‘minority’ orthography is especially characterized by the tendency to separate prefix clitics with a word divider. Comparison with Tiberian Hebrew again plays an important role particularly in the analysis of the ‘majority’ orthography.
For Hebrew and Moabite (Part III; chapters 10–12), Crellin sees the minimal prosodic word as the operative unit of focus. In other words, the orthographic groupings correspond to the smallest units that could “in principle stand as independent prosodic words, although they need not do so in a particular context” (p. 228). This is the situation reflected in the consonantal text of the Hebrew Bible and in Hebrew and Moabite inscriptions, which Crellin illustrates with attention to the Siloam Tunnel Inscription and the Mesha stelae (KAI 181 and 306).
Finally, for Epigraphic Greek (Part IV; chapters 13–16), Crellin sees a more complex situation that requires distinguishing between the roles of rhythm and pitch accent in the delineation of prosody. Furthermore, even with this distinction, he concludes that “in order to account for the data it seems one must accept a certain level of fluidity between levels of phonological representation” (p. 293).
Crellin’s general conclusion regarding the overall importance of prosody in Northwest Semitic word division is not in and of itself surprising, though the major challenge that the book confronts is the unevenness of the primary data. At times it seems that the exceptions are so numerous, or prominent, as to render the details of the analysis in some doubt. This is all the more noticeable since the book does not claim to be exhaustive in its analysis; principles are illustrated (if not derived) from certain subsets of texts, the a priori grouping of which is not always clear. In at least one case, namely the Ugaritic text KTU 2.88, the author posits the combination of two different orthographies at work (the aforementioned ‘majority’ and ‘minority’ orthographies respectively), which raises questions about viability of positing any definitive descriptions of the orthographic practices. In other words, to what extent is it reasonable to expect consistency in the orthographic conventions displayed in these texts?
This dovetails with another critique of the book, which is that it would benefit from a greater attention to the place of scribal practice and materiality in the analysis. While Crellin does offer some comments along these lines, particularly in the final chapter (chapter 17), it appears that these extra-linguistic factors are appealed to only when strictly linguistic explanations are judged inadequate. And yet I would argue that, given the nature of the source material involved in this investigation, a fundamental interrogation of the purposes and practices of the scribe need to play a bigger role in the analysis if not serve as the starting point for it.
All in all, this is a well-researched and careful analysis that can serve as a useful entry into a challenging topic. But I retain my doubts as to whether we can go beyond the top-level general conclusion regarding prosody that Crellin offers, given the unevenness and scope of the primary data in question.