Is rough vs smooth breathing predictable?

Question

Recently, I came across an excerpt from a scholium on Dionysius Thrax:

Διὰ τί τὸ "η" πρὸ τοῦ "τ" ψιλοῦται, ἐν δὲ τῷ ἧτα τῷ ὀνόματι τοῦ στοιχείου δασύνεται; Ἐπειδὴ παρὰ τοῖς ἀρχαίοις ὁ τύπος τοῦ "Η" ἐν τύπῳ δασείας ἔκειτο, ὥσπερ καὶ νῦν τοῖς Ῥωμαίοις.

Why is Ē before T pronounced smooth, when the name of the letter "hēta" is pronounced rough? It's because the ancients applied the shape of the hēta to the rough-breathing mark—like the Romans still do nowadays. (Trans. mine)

This is a nice little bit of trivia (that grammarians in the time of Dionysius Thrax still associated heta with /h/). But that first bit makes me wonder.

Is it a rule that heta before tau is always smooth? If so, are there other rules like this that allow us to predict rough versus smooth breathing? If not, what did the author mean by this? (I unfortunately have no context for the quote, which might provide more explanation.)

Answer 1

I'm not sure whether you meant for it to go without saying, but here are some basic facts about the distribution of the rough and smooth breathing marks in polytonic Greek orthography. The rough breathing is thought to have represented aspiration (possibly a consonant phoneme /h/, but there are more complicated suggestions for its phonemic representation) and the smooth breathing is thought to have represented the absence of aspiration. This contrast is phonemic, so the two types of breathing are not in complementary distribution in all contexts (I won't bother to try to find a minimal pair). For some words, the distribution is predictable, but for others, it isn't.

• The smooth and rough breathings don't appear on consonants, except for on ρ in some typographical traditions. When breathings are used on ρ, the distribution is almost completely predictable according to the following rules: ρ is written with rough breathing when at the start of the word, or after another ρ, and ρ is written with a smooth breathing when followed by ρ. The possible exceptions to these rules for ρ are very marginal: for more information, see When transliterating from Latin to Greek, what kind of rho is used? and Why does "ῤάρος" have a smooth breathing?

• The smooth and rough breathings don't appear on word-internal vowels, except:

  a) a rough or smooth breathing is used on the second vowel of a word-initial "diphthong"

  b) a "coronis", which apparently has the form of a smooth breathing in modern typography, is used word-internally to mark vowel contraction (Wikipedia)

  c) a rough breathing occurs exceptionally in the middle of the word ταὧς, and sporadically (but usually not) in the middle of certain prefixed or compound words (brianpck's answer to Greek pronunciation, invisible aspirations)

• Word-initial υ always takes a rough breathing, with possible exceptions for interjections or the name of the letter ("Gaps in the System", Nick Nicholas). The smooth breathing can occur on the letter υ as the second element of a diphthong. I asked on Linguistics SE about this gap, and it seems that we don't really know why it exists.

This would leave as unpredictable the use of rough vs. smooth breathing on word-initial vowels other than υ. You could try to derive rules to predict the occurrence from certain historical sound changes, but I don't think those would actually work to describe the synchronic distribution of the breathings, so I wouldn't recommend trying. (For example, you might think that "Grassmann's law" means that we shouldn't see a rough breathing when the second syllable of a word contains an aspirated consonant, but Ἥφαιστος is a counterexample to that candidate rule.)