I may have just had a blinding flash of light concerning Hiaki echo vowels, which I have to write down before I forget the reasoning. For the uninitiated, an echo vowel in Hiaki is actually a syllable of the form -ʔV, where the vocalic content of the V is copied from the final vowel of the stem to its left. They tend to show up in various morphological contexts, for example, in the stem forms of certain verbs. A stem form of a Hiaki verb is a special form used when a derivational suffix is added to the verb. The verb kecha, ‘to stand (something) up’, has the stem form kecha’a-, with an echo vowel at the end (the ‘ is the Hiaki orthographic symbol for the glottal stop [ʔ]). If I wanted to add the prospective suffix -vae to kecha and make a verb meaning ‘going to stand (something) up’, I have to attach -vae to the stem form, yielding kecha’avae. There’s also echo vowels that are triggered on all stems, apparently as a lexical part of particular suffixes. Alvarez (2012) analyzes a sequence -‘Vpo as a locative nominalizer/relativizer; it’s used in lots of place names. For example, mahtawa is a verb meaning ‘to be taught’, and if we add -‘Vpo to it, to get mahtawa’apo (with the final [a] of the stem filling in the vocalic content for the first syllable), it means ‘the place where one is taught’–that is, a school.

Previous thinkers about this phenomenon (e.g. Hagberg 1992) have assumed that the extra syllable usually appears as a prosodic repair, e.g. to add a mora to a foot that would otherwise fail a bimoraic requirement on foot structure, which makes some sense. But I just ran across a pair of forms in a corpus that makes me feel like there’s a whole nother way to think about it.

The locative relative forms that end in ‘Vpo are clearly related to the postposition -po meaning ‘at, on, in’, as in, e.g. mesapo ‘on the table’. This has led me to suspect that ‘Vpo is bimorphemic, consisting of a relativizing suffix made up of the glottal+echo vowel combo, and then followed by the normal postposition -po which turns the whole thing into a PP and allows the locative adjunct to be introduced into the sentence (these forms always behave as adjuncts, not arguments). But I didn’t really have any proof.

BUT! In a narrative about traveling in the mountains of the Hiaki homeland, I came across these sentences:

Akita Va’apo Weeka’u yevihne.
“You would arrive at “Where the Organ Pipe Stands In The Water”
Hunuen teak.
“That’s how it’s named.”
Akita Va’apo Weeka’awi.
“Where the Organ Pipe Stands In The Water”

In particular, notice the difference in form between Weeka’u ‘stand.REL’ in the first occurrence of the verb weeka, and Weeka’awi in the second occurrence of the verb. The second occurrence has an echo vowel. The first occurrence does not. What? If the echo vowel is part of the relativizing morpheme, it should ALWAYS be there! What’s going on, I thought?

So you can understand what follows better, here’s the morpheme-by-morpheme breakdown of the two crucial utterances, as I now think I understand them.

The first key thing is to notice that the -u suffix on Weeka’u in the first example means ‘to’. This is fully expected; the verb ‘arrive’ in Hiaki likes to occur with PPs headed by ‘to’, and ‘to’ is realized by the postposition -u most of the time. So, for example, if I wanted to say I was going to the river, I’d say I was going vatweu –vatwe being ‘river’ and -u being ‘to’.

The second thing to notice is that in the second example, the suffix meaning ‘to’ is -wi . And again, this is not unexpected. The postposition -u and the postposition -wi are phonological allomorphs of each other. There’s a recognized coda repair process whereby a Hiaki [w] turns into [u] when it’s not followed by a vowel; that’s what’s going on here in the version without the final -i.

So far, so good. Example (2) seems to be promising fodder for my theory that -‘Vpo forms are really made up of locative relativizer –‘V plus the postposition -po. If that theory is right, then we expect to see other postpositions occurring with that relativizer, and here is -‘a-wi, relativizer plus postposition ‘to’. So that’s great.

But what about the –‘u form in example (1)? Where’s the echo vowel? My morphemic theory isn’t faring so well here.

BUT, BUT — what if it’s the case that the relativizer is *just the glottal* by itself, the way I’ve indicated in the glossing of the (1) example? And what if glottals are forbidden to be codas in Hiaki? (They are.) So then, when the postposition following the glottal turns up as a pure vowel, as in (1), nothing else needs to happen, and we get weeka’u.

But what if the postposition following the relativizer is consonant-intial, as it is when the postposition is -po or -wi? Then the glottal will be a coda, and it can’t be a coda. So THEN an echo vowel is inserted, so that the glottal can surface as an onset, as it wants to do.

I feel like this almost HAS to be right. But that then means that in all the stem forms with echo vowels, the thing that’s inserted to create the stem form may actually just be the glottal itself. But then when it’s followed by a consonant-initial suffix, as it invariably is, the glottal will be a sad little coda, and will have to be resyllabified by the addition of an echo vowel to save it.

This makes sense in light of Jason Haugen’s thesis work on reduplication and other prosodic processes in Uto-Aztecan, where he discovered that many Uto-Aztecan languages created heavy syllables where they needed them via the addition of a coda glottal stop. Maybe Hiaki was like that, but then developed a ban against coda glottals — which then triggered the development of the echo vowel repair strategy. I can’t wait to tell him about it next week!