Joseph Dexter, Harvard University, Pramit Chaudhuri, and Elizabeth D. Adams, University of Texas at Austin
This paper describes a large-scale computational study of anagrams in classical Latin literature. Notwithstanding a small number of high-profile examples, such as the close collocation of Latium, maluit, quoniam latuisset, and paulatim at Aen. 8.322-326 (Ahl, O’Hara, Nelis, Chaudhuri and Dexter 2022), there remains a general skepticism about both the frequency and literary significance of anagrams in Latin poetry (Cameron, O’Hara). Large-scale studies that might corroborate or reinforce this skepticism have been limited, however, by a lack of benchmark datasets and quantitative analysis that could establish baseline rates for anagrammatic rearrangements in Latin corpora. Our paper takes an integrative approach to filling this gap and reports three primary contributions: development of a flexible, open-source computational tool for anagrammatic search, a corpus-wide dataset of 172 anagrams identified using the tool, and literary critical analysis of salient examples from the dataset.
We have created a software package that can identify anagrams in any text corpus subject to user-specified definitions. In the broadest sense, anagrams are words that can be rearranged into other words through a limited number of additions, deletions, substitutions, and transpositions. The minimum number of character-level operations required to change one word into another is known as the Damerau-Levenshtein edit distance; edit distance calculations are used in spelling correction and have been applied to research on literary intertextuality (Zhao and Sahni, Chaudhuri et al., Chaudhuri and Dexter 2017, Vierthaler and Gelein). As anagrams inherently involve transposition of letters, the tool places no limit on the number of transpositions, but the user can set a maximum number of additions, substitutions, and deletions, as well as various filtering options. The tool can be used to search either for anagrams of a specific word or for all anagrams of words appearing in close proximity to each other (in a manner similar to a Tesserae intertext search; Coffee et al.).
Since discussion of anagrams tends to focus on individual cases, it is hard to generalize to an expectation about the prevalence of anagrammatic play in general. An initial test bears out previous skepticism about anagram frequency (Cameron): for words of seven characters or more, our tool found only 172 exact rearrangements within a 10-word window in the entirety of the Tesserae corpus of classical Latin (Coffee et al.). We then proceed to the qualitative examples, noting, for instance, Lucr. 1.106-7 …turbare timore! / et merito, where the rearrangement, signaled by the verb, embodies the order that Lucretius’ rationalism creates out of disorder. More complex cases, such as Medea’s name located in caede madens in Valerius Flaccus (Houghton 2017), can involve multiple words. Vergil’s and Ovid’s famously intertextual Polyphemus episodes contain two particularly developed examples (Morison 1995), both of which convey not only the image of the Cyclops’ eating but also the reversal of nature: eructans et frusta cruento (Aen. 3.632, signaled by commixta at 633); elisi trepident sub dentibus (Met. 14.196, signaled by elisi).