What Word Embeddings Show About Plato

Two previous notes applied fastText models to Herodotus and Euripides, tracing how the semantic neighborhood of βάρβαρος shifts between a military cluster (Herodotus) and a political-social cluster (Euripides). The Plato model extends the comparison to a third register — philosophical prose — and produces results that complicate any simple story of progressive naturalization.

The first thing the Plato model shows is that βάρβαρος and Ἕλλην have become each other’s nearest neighbors, at a similarity of 0.620. In Herodotus the two terms are related but not dominant over each other; in Euripides they are close; in Plato they form a consolidated dyad. The pairing has become a formula, a standing antithesis that Plato can invoke without elaborating. Among βάρβαρος’s other neighbors: μειξοβάρβαρος (0.527), Plato’s own compound for “half-barbarian” used in the Laws to categorize peoples of mixed origin; εἱλωτεία (0.460), the Spartan institution of hereditary enslaved labor; and Λακεδαιμόνιος (0.494). The Spartan connection is distinctive and not present in the Herodotus or Euripides models at comparable proximity. In Plato’s political writing, the Spartan constitution is a constant reference point, and the Helots sit at the intersection of ethnicity and slavery in a way that makes them the structural counterpart to the barbarian within Greek territory.

The three-author comparison for βάρβαρος now spans Herodotus, Euripides, and Plato:

The political-social profile from Euripides broadly holds in Plato: τύραννος (0.292), δοῦλος (0.204), ἐλεύθερος (0.193), γένος (0.181) all remain significant. The association of βάρβαρος with tyranny and slavery, established in tragedy, persists in Plato’s dialogues where he writes about constitutional theory and historical examples.

Three entries break the pattern and carry the most analytical weight.

φύσις continues to barely move: 0.029 in Herodotus, 0.043 in Euripides, 0.083 in Plato. Across all three authors, βάρβαρος never gains a strong association with nature. This is the result that most directly addresses the naturalization hypothesis — the idea that Greek discourse progressively grounded group difference in inherent biological constitution. The distributional evidence does not support it, at least not at the level of the βάρβαρος term itself.

ψυχή goes negative (−0.097) in Plato, and λόγος (−0.028) and ἀρετή (0.006) collapse to near zero. The philosophical vocabulary — soul, reason, virtue — is kept almost entirely separate from the ethnic label. Where Plato discusses ψυχή he is writing in a different register from where he uses βάρβαρος. The two terms run in parallel without strong overlap. This bifurcation is what the model reveals most clearly about Plato: the philosophical elaboration of human nature and virtue proceeds in a conceptual vocabulary that does not substantially organize itself around the Greek/Barbarian antithesis, even though Plato frequently invokes that antithesis in his political and historical writing.

ἀρετή collapses from 0.291 in Herodotus to 0.006 in Plato. In Herodotus, virtue is a comparative ethnographic category — something various peoples are described as possessing or lacking — and βάρβαρος therefore appears in contexts where excellence is being assessed across groups. In Plato, virtue has become a philosophical category with its own technical vocabulary (σωφροσύνη, φρόνησις, ἀνδρεία, δικαιοσύνη, εὐδαιμονία all cluster tightly around ἀρετή), and that vocabulary is entirely separate from the ethnic antithesis.

Where the naturalization story does appear is in γένος, and here the three-author trajectory is the clearest diachronic signal the models produce:

In Herodotus, γένος clusters with political dynasty: ἀπόγονος (descendant), τύραννος, γαμβρός (son-in-law), ἄποικος (colonist). In Euripides, it shifts toward biological reproduction: σπέρμα, φύω, σπορά, φυτεύω. In Plato, it becomes an ontological category: εἶδος (0.511), φύσις (0.491), γένεσις (0.474), ζῷον (0.445), ἀγέλη (0.442 — herd, a zoological term), ἰδέα (0.383). The concept of group membership itself has migrated from political succession to biological reproduction to logical taxonomy — the vocabulary of natural kinds, species, and forms. This is the naturalization trajectory: not in βάρβαρος directly, but in the concept of what makes a group a group, which in Plato is a matter of φύσις and εἶδος rather than lineage or political belonging.

The distinction matters for how we understand ancient racializing discourse. The label βάρβαρος is not naturalized across any of these three authors — it never gains strong φύσις associations. But the conceptual framework for thinking about group identity does undergo naturalization, at the level of γένος, across the same period. What shifts is not the ethnic label but the ontological scaffolding for grouping. That scaffolding is available to be applied to ethnic categories, but in these texts it operates somewhat independently of them.