You are here

Rethinking discourse segmentation in Herodotus and Thucydides

Anna Bonifazi

University of Cologne

This paper claims that if discourse segmentation would consider further criteria beside hypotactic divisions and modern punctuation, our understanding of the length of segments, of their combination, and of the meanings expressed would significantly change.

What commonly happens when we read a contemporary edition of either historian is that we process discourse boundaries by relying on the syntactic sequence of main and sub-clauses, and by taking into account—almost unconsciously—punctuation. The latter might even guide our reading: for example, we may stop when we find a full stop, or we might infer from a comma that the thought is not yet complete. However, we tend to forget that modern punctuation represents the editors’ interpretation of mixed information provided not just by the words, but also (if not preponderantly) by choices made in previous print editions, by modern punctuation rules in writing, and by assumptions about periodic configurations (cf. e.g. Blass 1897). The result is that scholars such as Müller (1980) end up talking about long periods in Thucydides, and about the still primitive and/or imperfect periodic style of Herodotus. Seldom contemporary editions openly discuss the possible contribution of medieval punctuation to meaning and structure (Rijksbaron 2007 on Plato’s Ion is an exception). Further modern interpretations concern chapter and section boundaries (“our” chapter division of Herodotus appeared only in 1608; that of Thucydides in 1696). But also in this case, determining where major discourse units start or end is not an objective operation.

The central part of the paper points to four different kinds of input we could profit from to enrich the spectrum of possible discourse boundaries. The underlying perspectives differ, but on the whole matches overcome mismatches. Each input is illustrated by means of examples.

The first input is given by pre-print punctuation. The history of textual transmission tells us that for centuries the layouts of those Histories did not contain any punctuation, or they displayed different punctuation markers than the modern ones, and often in different positions. If digitalized versions of medieval manuscripts are available, checking the type and the location of punctuation markers is not only worthy but also relatively handy. It may reveal where scribes/readers/teachers felt important to pause/to breath beside at clause boundaries.

The second input comes from prose colometry, which modern scholars retrieved from antiquity. Relevant works range from Fraenkel 1932, 1933 to Scheppers 2011 (The colon hypothesis), and include prose colometry in Latin literature. The main point to underscore is that kôla are not necessarily clauses.  

The third input is offered by Goldstein’s study of the relationship between enclitic words and intonational phrases (Goldstein 2016). The role of “second position” words can be rethought of in terms of their contribution to intonational boundaries. This allows us to “listen to” kinds of boundaries that a hypotactic segmentation could not detect.  

The fourth input is given by the pragmatic perspective that takes several particles, hyperbata, anacolutha and other linguistic features as cues to segmentation into acts and moves—that is, smaller and bigger strategic steps in communication (Bonifazi et al. 2016).

These four inputs invite us to go back to what ancient orators and grammarians took for granted, that is, prose works are successful to the extent that both the grammatical and the performative dimensions of language are harmoniously mastered.

The arguments presented in this paper can be extended to ancient Greek prose in general.

Session/Panel Title

Greek and Latin Linguistics

Session/Paper Number


© 2020, Society for Classical Studies Privacy Policy