An Unhealthy Obsession with W (and its pronunciation in Latin)

Oh! Hello there! It has be a very long time since I’ve been around, but with good reason – I’ve gotten a new apartment with my wonderful boyfriend, and I’ve been starting up my life as a student and grad assistant at Villanova. I thought as a ‘returning to my blog’ post, I would write about my favorite linguistic topic: [w].

You may remember this post from when I first started this blog. You may have thought I was going to talk about when the Romans stopped pronouncing V as [w] and started pronouncing it as [v] instead. Well, I was being misleading – I don’t actually know a whole lot about that topic. What I really wanted to talk about was whether or not [w] was a phoneme in Classical Latin, or if it was just an allophone of [u].

What is a phoneme?

Quick clarification on the difference between phoneme and allophone:1 A phoneme is defined as a meaningful sound – i.e. a sound that can be used by a speaker to signal a semantic difference between two words. If we look at pat vs. bat, we can see that there is a phonemic contrast between /p/ and /b/ because these words differ only in that sound. In contrast, we are capable of making sounds that convey no meaning in our language. For example, if you pronounce pat and apt while holding your hand in front of your mouth, you will notice that you produce a puff of air when you pronounce the /p/ in pat, but not when you pronounce the /p/ in apt. These are two different sounds: [p] and [ph], but English doesn’t distinguish between the two of them (however, many Indian languages do). In English then, [ph] is an allophone of /p/.2

So to reiterate, what I will be discussing is whether [w] was a phoneme in its own right (i.e. did it meaningfully contrast with another sound) or was it just an allophone, another way of pronouncing the Latin phoneme /u/?

How do we know the Romans pronounced [w]?
Let’s start by asking how on earth we know that the Romans pronounced V as [w] in the first place:

I definitely just copied that from a poster I’d made a while ago, so please excuse the odd brown background. In the inscription you can see that the V surrounded by a red box is probably going to be a consonant (because it comes right before a vowel), and that the V surrounded by a blue box is probably going to be a vowel (because it is surrounded by consonants).3 If you look at the line of poetry, you can see again that the same letter has to represent both a consonant and a vowel (in red and blue respectively). Poetry is a great tool because the placement of consonants and vowels gives us the rhythm. By comparing this line of poetry to a line from the same poem with no Vs (us) in it, we can be fairly certain what the rhythm of this line was supposed to be.

Now we need to ask ourselves what sort of sound can behave like this – sometimes showing up as a vowel and sometimes showing up as a consonant. The answer is going to be a ‘semivowel’:

Well that’s all well and good, but how do we know which semivowel it’s going to be? I primed you a bit by using [u] and [w] in my example, but there is also the possibility that it could be [j] (I discussed this semivowel in this post). Fortunately, we know what this sound turned into:

Since the semivowel [v] is most similar to is [w], we can be pretty certain that the Latin semivowel was [w].

How can we tell if [w] was a phoneme or an allophone?

So now we’ve determined that the Romans definitely pronounced [w], we can try to answer our question: was [w] an allophone of /u/ or a phoneme in its own right?

First of all, we can check to see if [w] contrasts with other phonemes:

pestis contrasts with uestis

bonam contrasts with bonum

Score: Phoneme: 1, Allophone 0. u definitely seems to contrast with other phonemes.

On the other hand, it seems fairly clear that the the consonantal and vocalic realizations of u are conditioned by its environment. When u appears as a syllable onset or is between vowels, it acts like a consonant, but when it appears between consonants, it acts like a vowel – just as we saw above. Since this pattern is completely predictable, we say that [u] and [w] are in complementary distribution – they will never occur in a position where they can contrast with each other.

Score Phoneme: 1, Allophone 1.

Oh…but wait a second. What about uoluit (woluit) “he wanted” versus uoluit (wolwit) “it rolls” It looks like these us are in the same position and provide a meaningful contrast between two otherwise identical words. Sure looks like a phoneme to me.

Score: Phoneme: 2, Allophone 1.

…Or at least it would if that were a valid comparison. Unfortunately for team Phoneme, comparing uoluit ‘he wanted’ to uoluit ‘it rolls’ is comparing apples to oranges. In the first word, the u is a tense marker, and in the second word, u is part of the stem. We’re going to have to take that point away from team Phoneme for an illegal move. (I apologize for the horrifically cliche sports metaphor, but gosh darnit it’s convenient, and I’m going to stick with it.)

Score: Phoneme:, 1, Allophone: 1.

It also turns out that whenever u follows a consonant and precedes a vowel, we can establish a rule. If it is part of the stem, it will be analyzed as [w], and if it is a grammatical ending, it gets analyzed as [u]. If you’re good at reading generative grammar style rules, see the list below. There are some funky things going on here that I’ve expanded on in a paper that will hopefully be published pretty soon (in an online journal – nothing too exciting), but the pattern is pretty consistent:

u →
a. [w]
i. / V _ V (auis)
ii. / /l , r/ . _ + (silua)
iii. / . + /s/ _ V (suauis)
iv. (?) / /n/ . _+ (tenuis)
b. [u]
i. / . + C0 _ C0 . + (unus, cornu)
ii. / /l , r/ + _ . (merui)
iii. / . + /s/ _ + (sua)

The point is that even in an environment where u occurs between a consonant and a vowel, [u] and [w] are in complementary distribution.

Score: Phoneme: 1, Allophone: 2

Did the Romans think the sounds were different?

No, we can’t know exactly what the Romans thought about anything – they lived 2000 years ago, and no one’s invented a time machine yet. However, they did leave us some clues.

One of those clues is in poetry. We can find a lot variation in how tenuis, for example, scanned. Sometimes it scanned as though it were ‘tenuis’ and sometimes like ‘tenwis’. There’s also a very interesting example in Vergil where he scans silua as ‘silua’ instead of ‘silwa’ as it almost always scans. If this kind of variation was going on, maybe the rules I gave above were not so hard and fast. Poets seems to be able to perceive a difference between the sounds, and they would use whichever version fit their meter better.

Furthermore, the Emperor Claudius could definitely tell the difference between the sounds. He tried to make a letter to account for it. Here is an inscription from his time as emperor:

Particularly useful in this inscription is that the inscriber made some errors. Of course, it might be argued that these errors indicated some difference in pronunciation of the u, but it seems unlikely. The errors seem to be grouped and primed rather than occurring in any specific phonetic environment. I image the poor man just wasn’t used to this letter and kept screwing up. Every time he sets down his chisel for a small break, he comes back and starts writing Vs instead of inversed digammas (i.e. upside down F’s) all over again. I also imagine he had some choice words for the Emperor Claudius.

How to explain the disparity of perception with the rest of our evidence? I’ll keep it short because I’m running out of steam (and need some breakfast). In and around the time of the late republic and early empire, people were starting to pronounce their [w]s differently – they were lowering the back of their tongues from the velum:

Once the back of the tongue had lowered, they ended up producing a sound a lot like [b]. This, by the way, is what I was getting at in my weni, widi, wici post. There was a bit of a class divide about the whole thing (or so we can imagine). Pedantic, high-brow Romans were resistant to the change (in the same way you won’t hear Prince Charles pronounce ‘think’ ‘fink’ the way a Londoner might), and maintained the Proto-Latin allophony between [u] and [w], considering it to be more ‘proper’. However, the change grew increasingly apparent and poets were able to exploit the difference. Claudius, feeling it was his duty to clarify the Latin Alphabet so there would be one letter for each sound, decided to add an upside down digamma to represent this [β](a b pronounced in the same manner as a v) that was showing up more and more frequently in every day speech.

I think that accounts pretty well for the evidence we have concerning [w] and [u], but the last paragraph is fairly speculative and there is quite a lot of debate about the exact dates of the sound change, where it was happening, and who it applied to.

Anyway, I’d love to hear your thoughts as always! I have to go make myself some delicious oatmeal so my tummy stops yelling at me.

1. There has been some discussion in recent years (I consider the 90s recent because I can’t wrap my mind around 2000 being twelve years ago) about whether or not the phenomenon of phonemes is natural or simply a result of our writing system. I won’t get into it here, but it is very interesting and I recommend JSTORing it if you have access. Go Back

2. Phonemes are written between virgules /p/, and allophones or sounds I’m not ready to state the phonemic status of appear in square brackets [p]. Graphemes (symbols used to represent sounds – known to most by the much less pretentious term ‘letters’) are written between angle brackets, but since this is html, it’s more effort than its worth to get angle brackets to show up. Go Back

3. In almost every language, words are a strings of sonorous elements (vowels) and less sonorous elements (consonants). As I’ll mention above in a few paragraphs, some sounds can be either consonants or vowels. If they are surrounded by consonants, they are considered vowels, and if they are surrounded by vowels, they are considered consonants. In very simple terms, this makes the word easier to articulate. Go Back


3 comments on “An Unhealthy Obsession with W (and its pronunciation in Latin)

  1. John Cowan says:

    The fact that the Romans wrote both the consonant and the vowel with U is not really evidence that /w/ was realized as [w] rather than [v]. After all, Early Modern English wrote both /v/ and /u/ with U too until about 1530, though the split between /v/ and /f/ had happened several centuries before with the borrowing of French words with initial /v/.

    The Appendix Probi, which was written before the year 300, tells us BACULUS NON VACLUS, BRAVIUM NON BRABIUM, ALVEUS NON ALBEUS, PLEBES NON PLEVIS, TABES NON TAVIS, TOLERABILIS NON TOLERAVILIS, VAPULO NON BAPLO, which pretty well demonstrates that consonantal U had shifted toward a fricative, as intervocalic B also had, by then.

    It may have happened much earlier, of course: absence of evidence in the form of misspellings is not evidence of absence. A fine example is that although North American English has entirely merged intervocalic /t/ and /d/ as a voiced alveolar flap, misspellings like heridable are still pretty rare, which would naively suggest that the merger is only sporadic. As long as most people who write are more or less taught to spell, the appearance of the older phonology (which in English means roughly the 14th century!) will be preserved.

    • You’re right, of course, and perhaps I should have said more about it. I think the English use of u for v&u&w can be attributed to ecclesiastical Latin, and therefore is rooted in the reason Latin u was used to write [v], i.e. that it developed to [v], so i don’t know how I feel about that as an argument that Latin u shouldn’t be used as evidence. And though it goes against my (biased) instincts that it is natural to write [u] and [w] with the same letter because they are so similar, and therefore that [w] is the more likely candidate given the choice between [w] and [v], you’re right that the evidence I’ve given doesn’t preclude [v] as a pronunciation. I think for that I would have to go into discussions of the development of the Latin alphabet and justifications for indo-european phonology and learn a bit more about the development from [w] to [v]. I’m going to whimp out though – I have to get ready for work and I can’t remember off the top of my head which box I put that info in. I do have the paper up though and found this little nugget: In McCullagh’s section of Clackson’s Companion to the Latin Language he states “In Indo-European, and possibly also in Early Latin, /i/ and /j/ were allophones, as were /u/ and /w/”. I remember dying to get his research or find a better discussion of it, but running out of time. I really wish I’d been a good student researched that paper more thoroughly. Thank you for a good morning mind exercise though!

      Oh, also, the information from the Appendix Probi is really interesting, and really great to have. Thank you! I wonder if there are any misspellings among the less-educated scribbles at Pompeii – I know that there are some showing the loss of final ‘n’ and the loss of initial ‘h’, but i can’t remember anything about [w]/[v]/[β].

      Love your comments as always!

  2. John Cowan says:

    For some reason, both of us left off the minimal pair cui vs. qui.


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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s