You are here


Benjamin VICTOR

Over the last generation the linguistic differentiation of characters in Roman comedy has come in for investigation, and has been held more thorough and subtler than used to be thought. While some work along these lines has been excellent, some, too, is in need of serious revision. This paper will address the traces of sociolect that have been hypothesized for the speech of low-class characters, especially slaves— a subject studied thus far on the basis of inadequate data.

Karakasis 2005, pp. 83-89 has claimed that in Terence the language of slaves is marked by Greek turns of phrase. His starting point was a ‘consensus among scholars’, as he put it, regarding which Terentian features of syntax are of Greek origin. But scholarly consensus is worthless until serious research is done, and I daresay no proper study has ever been published on Greek-influenced grammatical constructions in Plautus or Terence: Brenous 1895 and Löfstedt 1933, the most significant work on Greek syntactic influence upon Latin, took only passing notice of comedy—a lacuna yet to be made good. Accordingly Karakasis found only five Terentian constructions agreed to be Hellenisms, and it so happens that three of them occur in slave-speech. A proper collection of syntactic Hellenisms in Terence, or for that matter in the palliata, would reveal them vastly more numerous. I myself have begun a catalogue with a view to writing a monograph on them someday. So far I count 191 instances of some sixty constructions and am still finding more. Most, it turns out, are said by respectable free characters.

Karakasis was following the lead of Maltby 1985, who had presented a detailed justification of the thesis that Terence’s slaves and other low-class characters use more Greek loan-words than the high-class characters, a difference which he took to reflect sociolectal division in the Latin of real life. This work, too, is flawed. Maltby began by defining a set of Greek lexical loans recent enough as of the second century BC to be likely recognised as foreign. He then calculated their frequency in the speech of slaves as one occurrence in 40 lines, in that of all low-class characters as one in 45.5. By contrast, the rate for high-class characters was found to be just one in 102 lines. The flaw was not to count the interjection ehem and the interjection hem expressing incomprehension or surprise. These, I shall argue, must have been borrowings from the Greek (the models being ἠήν and á¼¥ν respectively) and been recognized as such. Once they are included the figures for lexical Hellenisms become: one occurrence per 25 lines for slaves; one per 29 for all low-class characters; one per 35 for high-class characters. Such a slight difference between low-and high-class speech would be imperceptible even to the most attentive public. It is therefore excluded that the simple category ‘Greek words’ should be relevant to the definition of a sociolect, and the problem needs to be rethought in consequence.


  • Brenous 1895 = Brenous, Joseph. Étude sur les hellénismes dans la syntaxe latine. Paris.
  • Karakasis 2005 = Karakasis, Evangelos. Terence and the language of Roman comedy. Cambridge.
  • Löfstedt 1933 = Löfstedt, Einar. Syntactica. Studien und Beiträge zur historischen Syntax des Lateins. 2nd ed. Lund, vol. II., pp. 406-57.
  • Maltby 1985 = Maltby, Robert. " The distribution of Greek loan-words in Terence." The Classical Quarterly 35, pp. 110-123.

© 2020, Society for Classical Studies Privacy Policy