Hiatus Resolution: ‘Your search returned no results’

So, my posts have been distinctly sparse as of late. There are two reasons for this; the primary one is the arrival of my boyfriend in Cambridge, but running a very close second is my thesis.

As is the way with theses, they do tend to take up a lot of one’s time, especially when left to the last minute *coughmighthavedonethatcough*. What I found particularly challenging about my topic though was the odd lack of literature. What I’m researching is hiatus resolution. Classicists may have encountered this term when dealing with Homer and epic correption, ‘vocalis ante vocalem corripitur‘ (a vowel is shortened before [another] vowel). This Latin phrase refers to the tendency in epic poetry for a word final long vowel or diphthong to shorten before a word initial vowel:

τὴν δ’ ἐγὼ οὐ λύσω: πρίν μιν καὶ γῆρας ἔπεισιν
(the final omega in ἐγὼ scans short, even though omega typically scans long, because it is followed by a vowel – in this case omicron)

This is one method Greek (and in a sense Latin – more later) employs to deal with hiatus. Hiatus can be defined as the heterosyllabic adjacency of two syllable nuclei. In non-technical language, that means ‘two vowels or vowel-like segments occurring next to eachother, but in different syllables’. For example, the word hiatus contains an example of hiatus; it is syllabified hi-a-tus. The i and the a are next to each other, but occur in different syllables. Often, languages try to avoid this, presumably because it can be a bit awkward to pronounce. Since hiatus consists of two aspects, heterosyllabicity and vowel adjacency, a hiatus resolution method will try to eliminate one of these aspects so there is no longer a hiatus:

resolving heterosyllabicity (where C = consonant and V = vowel)

resolving vowel adjacency (where C = consonant and V = vowel)

There are quite a lot of interesting things that can happen here, but I’ll discuss that in a later post once I’ve finished my thesis.

You’ll note that I said, ‘there are quite a lot of interesting things that can happen here’. Despite this, there doesn’t seem to be much literature on the topic at all. As far as I’m aware, only Roderic Casali has published anything explicitly to do with the topic. The rest is hidden in grammars and theoretical texts on syllabification (but not really). Sidney Allen and Michel Lejeune both include sections on ‘vowel-junction’ in their grammars, and this sort of thing is, I believe, what is typical of discussions of hiatus resolution[5]

One might imagine that books that discuss syllabification would be keen to talk about hiatus, but I have found no recent literature that even touches on the matter (as discovered by the very scientific method of looking through the indices of every book on syllabification, syllable weight, phonology, and phonetics found in the linguistics section of the Modern and Medieval Languages Library at Cambridge).

I would like to say that this post has more of a point, but mostly I just wanted to share this oddity and see what people thought about it. Why on earth is hiatus so rarely mentioned in theoretical texts? Am I looking in the wrong places? Or is it, perhaps, that it isn’t really that interesting after all?

1.Casali, R. F. (1998). Resolving hiatus. New York ; London, Garland.; Casali, R. F. (2011). ‘Hiatus resolution’. The Blackwell companion to phonology. D. W. Oostendorp. Malden, Mass., Wiley-Blackwell. 3: 1434-1460.
2.Allen, W. S. (1987). Vox Graeca : a guide to the pronunciation of classical Greek. Cambridge, Cambridge University Press.; Allen, W. S. (1978). Vox Latina : a guide to the pronunciation of classical Latin. Cambridge, Cambridge University Press.
3.Lejeune, M. (1972). Phonétique historique du mycénien et du grec ancien. Paris,, Klincksieck.
4.Aguilar, L. (1999). “Hiatus and dipthong: acoustic cues and speech situation differences.” Speech Communication 28: 57-74.
5. Though Lourdes Aguilar has provided what is, to my knowledge, the only phonetic analysis of hiatus. This is useful, because it confirms that real hiatus (i.e. hiatus that isn’t interrupted by a glottal stop or other ‘silent’ consonantal element – like what you would get if you pronounced ‘a apple’) can be pronounced. Perhaps this wasn’t up for debate, but it certainly was never particularly clear. In this study, Aguilar shows that hiatus is essentially a ‘diphthong’ that is super long – long enough to be perceived as to separate syllables.


12 comments on “Hiatus Resolution: ‘Your search returned no results’

  1. John Cowan says:

    Since there are tens of millions of Americans (mostly black AAVE-speakers, but not exclusively) who don’t have a/an alternation, using only a, I certainly hope it’s possible, or they will all be condemned to mutism rather than being able to say /əæpəl/.

    • Lol, very true! I admit that that statement comes from a bit of a misconception that I had when I first started my research: I’d assumed that what was meant by ‘languages that permit hiatus’ was ‘languages that don’t have a phonemic hiatus resolution strategy’. I thought that words were syllabified by sonority peaks occurring between ‘consonant troughs’. I didn’t think it was actually possible to create two syllables without some dip in the sonority (so hiatus would *have* to be pronounced [hijatus] (couldn’t figure out how to do a superscript). Turns out I was wrong!

  2. rsmease says:

    On the face of it, I’m interested in reading Vox Graeca and Vox Latina over the summer. Between the covers, though, they look like they have a very demanding linguistics jargon. Should I be scared? What can I / did you gain from reading these?

    • Hmmm. I’d say they’re more the type of book that it’s worth buying and saving for reference. It would sort of be like reading a grammar in one go. The book does have a small introduction to linguistics in it that would certainly be worth reading, but though Allen’s jargon doesn’t get too thick, a lot of the phenomena he mentions are more easily understood with some linguistic training. Even with my (admittedly not awesome) linguistic training, I find myself rereading sections a few times before I get what he’s talking about. If you’re interested in understanding the reason reasons for certain assumption we make about Greek and Latin Phonology (like how we know φ,χ,θ are aspirated stops and not fricatives, or how we know that Latin could be pronounced [w]) then you’ll find a lot worth checking out. I certainly wouldn’t be able to sit down and read them cover to cover though.

      Does that help?

  3. Tom Recht says:

    “the tendency in epic poetry” – and in inscriptions. At least, Cretan inscriptions sometimes show word-final long diphthongs spelled short before a vowel-initial word; I don’t know if this happens in other dialects too. There must have been some phonetic reality behind this, which didn’t get phonologized so doesn’t show up in standard written Greek.

    But why call this ‘hiatus resolution’? The hiatus remains, it’s only the vowel quantity that changes.

    • Ooh! do they? That’s exciting. I would be grateful if you could point out a reference of two. What’s the date?

      The phonetic reality behind it is one of the points I’m discussing in my thesis – why does hiatus cause the shortening of long vowels and diphthongs in hiatus? In Greek, i’m suggesting glide formation of V2 in diphthongs resulted in the loss of on of the moras (i.e. through the creation of an onset), but it seems like a bit of a cop out, even to me; more troubling is that I’m not sure how to explain the similar word internal process in Latin. I don’t know that there’s really any precedent with diphthongs to justify it.

      As to whether or not this should be considered hiatus resolution, well, I was initially bothered by calling it that myself. Actually, I had a very long rant about it to my supervisor. However, since it seems to be triggered by hiatus, it seems that it must be some sort of resolution method. This is the part where I feel like a crazy person – I’m imagining loss of a mora through onset creation with the high V2 of a final diphthong; this process applies in all instances where there is hiatus at a word boundary, and where there is no high V2 to use as an onset, the rule deletes a mora instead. Thus, the surface form has forms with both phonetically resolved and unresolved hiatus. The problem i’m seeing is with long dipthongs. In order for my speculation to work, Homeric long diphthongs would have had to have lost their yod element already…..i don’t really feel comfortable saying that.

      Any thoughts?

      • Tom Recht says:

        I don’t have a precise reference, but I remember that Monique Bile’s grammar of Cretan lists examples of inscriptions where long diphthongs are spelled short, and that looking at them it struck me that most of them were before a vowel-initial word. The only specific example I remember, which was actually what led me to look this up in the first place, was in the Poinikastas inscription (which is pretty early, c. 500), but it’s in a very badly understood line so it’s far from certain whether the there stands for a long diphthong or not. But there are other such cases given in the grammar which are certain.

        I don’t know if anyone has ever explicitly made the connection between this kind of apparent orthographic variation in inscriptions and epic shortening in hiatus, but it does seem plausible that they’re reflecting the same phonetic process. I bet if you looked at the phonology sections of some dialect grammars you might find more such examples.

        It does seem like a fairly understandable process from a phonetic viewpoint, though – given such a long sequence of vocalic-sounding stuff smooshed together it would be easy to miscount the units. Presumably in Latin this got phonologized into actual vowel shortening, while in Greek either there was some variation for a while where some dialects phonologized it and others didn’t, or maybe even all dialects phonologized it but the original quantity later got restored by analogy. (I wonder if there are any word-internal cases in Greek, like in Latin; I can’t remember offhand.)

        • Tom Recht says:

          Sorry, “the there” = “the OI there”, with angle brackets. I forgot those had a different modern use.

        • Thanks for the tip on the dialects. It would certainly be interesting to see if there is dialect variation and if there’s any variation between diphthongs and long vowels.

          As to the mushing together of vowels sounds (:P), do you think it might be a reaction against superheavy syllables? Though, I think this would require an attempt to syllabify all of the vocoids into one syllable first – eek! Not sure if I’m comfortable with that. Surely there are limits on what vowels can be syllabified into one syllable, as described by the diphthongs permitted in the phonemic inventory. Hmm. So maybe not that.

          Miscounting the weight units as you suggest would work a bit like syncope I think, but whether or not this is possible might depend on whether one sees the weight tier (moraic tier) as projected from the segments or the syllable. I imagine that syncope of a mora would be more difficult if it were projected from the segment. Maybe that’s irrelevant. Dunno.

          I’d still like to find a specific formal motivation. What about this environment allows deletion that other environments don’t? Going back to what you said about the long sequence of vocoids and what I said about a reaction against superheavy syllables, perhaps it’s not a reaction against superheavy syllables but rather the number of weight units which can occur adjacently without regard for the syllable. If there was a rule that, say, no more than 3 moras could occur adjacently, then perhaps deletion of a mora is the only way to compensate. I’m going to go take a look at Hyman (1985) and see if he says anything useful.

          Apologies for the train-of thought writing and the loads of speculation…it might be the case that i’m in the middle of editing my thesis and I’m finding this discussion very thoguht-provoking :P

  4. trecht says:

    I think some restriction on the number of adjacent morae sounds on the right track, in terms of formalizing the rule phonologically. I wonder if you need to include the word boundary in the conditions at all. What I’ve now remembered (and you may already know) is that there are word-internal cases of vowels shortening before another vowel in Greek as well as Latin — for example in the etymology of (one of the forms of) the word for ‘cheek’, παρεά, where ε is historically from η. But I don’t know if this is completely regular. I’m pretty sure there are Greek words that contain a long vowel or long diphthong followed by another vowel, though I can’t think of one offhand; and of course if such words do exist you’d want to know whether the long vowel/diphthong scans long or short in epic poetry.

    • Just wanted to let you know that a large chunk of my thesis ended up being about the possible mora deletion rule that our conversation inspired. Turns out it works rather neatly and brought my thesis from ‘shitty ramblings about hiatus’ to ‘interesting and insightful’. Thanks for the comments. I didn’t get a chance to use dialects (because I was too pressed for time), but I’m going to apply for funding to research the possibility of the rule more next summer. Anyway, our conversation really got me out of hot water! Thanks! I’ll be writing a blog about it sometime in the near future :)


Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s