Latin and ancient (Attic) Greek look similar in vocabulary and in grammar. What make them into different subgroups in the Indo-European family of languages?

  • 3
    Apart from the phonetics, phonology, phonotactics, prosody (stress), and the way Proto-Indo-European phonemes are reflected in each of the two languages, do the grammars of Latin and ancient (Attic) Greek look really similar, as for you? What about the number of cases, genders, and numbers, with dual in Greek? What about the way tenses and voices are formed? How many personal pronouns are similar? Just one, ego/ἐγώ? What do you really find similar enough for the two languages to be classified into the same subgroup of IE, could you explain?
    – Yellow Sky
    Commented May 11 at 9:54
  • 5
    The burden of proof is really on the claim two languages are similar. Commented May 11 at 19:04

1 Answer 1


Latin and Greek are similar enough to clearly be related languages, but they really aren't that similar. Just a few incompatible divergences that must have happened very early in their development:

  • Greek has its so-called triple reflex of the Proto-Indo-European laryngeals: in positions where they're realised as vowels, *h₁, *h₂, and *h₃ became ε, α, and ο respectively. In the Italic languages they all became a, which in Latin subsequently weakened to i in non-initial syllables; at the start of a word, they were lost entirely (as in most languages). γενέτωρ vs. genitor < *genator 'parent', PIE *g̑énh₁tōr; πατήρ vs. pater 'father', PIE *ph₂tḗr; δοτός vs. datus 'given', PIE *dh₃tós.

  • Greek also has a double reflex of *i̯- under unclear conditions (usually assumed to be *i̯- versus *Hi̯-, where *H is some laryngeal): compare ζυγόν to iugum 'yoke', but παρ to iecur 'liver'.

  • The PIE syllabic nasals *m̥ and *n̥ both became α in Greek, but em and en in Latin: ὄνομα vs. nōmen, δέκα vs. decem.

  • Many differing developments of and restrictions on stop clusters; e.g. the "thorn cluster" *dʰg̑ʰ- underwent metathesis in Greek but simplified to *g̑ʰ- in pre-Latin (χθών next to humus 'ground').

That's just phonology. In morphology, Latin of course massively remodelled its verbal system, collapsing the perfect and the aorist and innovating a future that's different from the future innovated by Greek, as well as thematising almost all of its athematic verbs and generally making a muddle of things. Greek, for its part, innovated a passive in the aorist (Latin broadly continues the PIE medio-passive in its passive). The nouns diverged as well, with the fourth and the fifth declension being Latin innovations; Latin also merged the PIE instrumental and locative with the ablative, whereas Greek merged both with the dative and merged the ablative with the genitive instead.

The lexicons famously really aren't that similar, with Greek often (but controversially) being grouped with Armenian and/or Indo-Iranian largely based on lexical isoglosses. Greek's lexicon is much bigger than Latin's—it certainly has reflexes of a greater number of PIE roots—so it's often the case that you can find a Greek cognate for any given inherited Latin word; the reverse isn't a given.

For my money it's the development of the laryngeals that most strongly demonstrates the early divergence of Greek and Latin: Greek's development there was demonstrably already complete in Mycenaean times (the earliest Mycenaean Greek we have is from the 14th century BCE, also a good 700 years before the earliest attestation of any kind of Latin), and likely much earlier than that, so Latin and Greek must have diverged earlier still.
Obviously on the face of it Latin and Greek look more similar than, say, Armenian and English; that's really just because they were two or three millennia closer in time to their common ancestor than Indo-European languages today are, not because they were particularly closely related.

For more information on the defining features of the Indo-European language families see e.g. Thomas Olander (et al.)'s open-access The Indo-European Language Family for an excellent synoptic introduction.

(It is true that Dionysius of Halicarnassus considered Latin to be an Aeolic dialect of Greek, but that was a minority view even at the time.)

  • I wonder if you'd take a tangent and comment on whether Sanskrit is closer to Greek than Latin, by contrast. A friend (unpublished) has studied lexical overlaps between Greek and other IE languages and has informally concluded the highest overlaps of Greek are with Eastern IE languages, Sanskrit, Ancient Persian, and so on. I'd appreciate your broad take on such? Commented May 11 at 20:52
  • 1
    @CosmasZachos It's not an uncommon view, but lexical evidence has to be handled a lot more carefully than sound laws—if language x and y both show reflexes of a root the default reason for that is that they both retained it and it's therefore a shared archaism, which cannot be used as evidence for a subgrouping, and that doesn't change if you find a hundred or a thousand such correspondences. You can find shared innovations in the lexicon ("root x meant y originally but came to mean z", or certain derivational innovations), but I'm not aware of any large-scale studies that bother.
    – Cairnarvon
    Commented May 11 at 22:58
  • 1
    @CosmasZachos That said, it's not really my area of expertise and it's entirely possible I've missed important work. My instinct is that people are prone to seeing a close relationship between Greek and Sanskrit specifically because they're both very early literary languages for which we have very large early corpora and they both have notably large inherited lexicons (in part because they're early and well-attested literary languages) compared to other IE languages, and if Greek is close to Sanskrit is must be close to Indo-Iranian in general, because that is a well-established subgroup.
    – Cairnarvon
    Commented May 11 at 23:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.