The problem with long alpha

This week, a lot of us who teach Classical Philology and Linguistics at Cambridge have been teaching our first essay of the first year course on the sounds of Greek and Latin. I had a request from a student for a good example of the difference between long and short alpha, which really stumped me for a while. The kind of example I needed takes a bit of explaining, but stick with me (or skip to the end if you just want to know the great example I eventually got from a colleague).

The most important point that we are trying to get across in this essay is the difference between a phoneme and a phone. A phone is the smallest unit of sound that it’s possible to make with your mouth, vocal chords and other speech apparatus – we write these in square brackets, for example [p]. But there are hundreds and hundreds of possible sounds you could make, some of which are very similar to each other, and so not all of them are thought of as separate units of sound by speakers. So, when we’re describing a language, we also want to think about how speakers of the language conceptually group all those phones together into phonemes. A phoneme is the smallest contrastive unit of sound in a particular language – we write these between slashes, for example /p/. What do I mean by “contrastive unit”? This is easiest to explain with some examples.

In English, if I use the phones [bet] then someone listening to me would understand this as the word bet. If I use the phones [pet], then the listener would understand this as the word pet. So, if I replace [b] with [p], I’ve changed the meaning of the word. I can therefore say that the phones [b] and [p] belong to two different phonemes, because [b] and [p] contrast with each other when used in the same environment. We can refer to these phonemes as /b/ and /p/.

I could try the same experiment with other phones. For example, I could try [pet] and [pʰet]. This time, my listener hears the word pet both times. In fact, in British English, the normal pronunciation of pet is [pʰet], with an aspirated [pʰ] at the beginning. But speakers of English tend not to notice this aspiration, because in English the phones [pʰ] and [p] do not contrast with each other. These phones both appear in English, but in different environments from each other – so for example pot starts with [pʰ] but spot has [p], because the sound isn’t appearing at the beginning of the word. This same pattern is repeated in all other words – in English, [pʰ] is the phone we use at the beginning of words, and [p] is the phone we use in the middle of words. So we can say that [pʰ] and [p] are both allophones (or alternative phones) of the English phoneme /p/, and that they are used in different environments from one another.

It’s important to remember, though, that the way in which phones are grouped together into phonemes is different in each language. So, if I were to try the same experiment with ancient Greek, I would get a different result. For example, if I said [pʰō], that would mean “I speak” (in the subjunctive), whereas [pō] would mean “yet”. In Greek, unlike in English, the phones [pʰ] and [p] are part of two different phonemes, /pʰ/ and /p/. These two phonemes are also conveniently written with two different letters, <φ> and <π>. Learning the differences between the phonemes of your own language and another is one part of why pronunciation is difficult when you’re learning another language.

In these examples, the pairs of contrasting words I’m looking for, like bet and pet, are called minimal pairs. Finding a minimal pair is the easiest way to check whether two sounds belong to different phonemes or are allophones of the same phoneme. This isn’t always easy, though. For example, English has two phonemes /θ/ and /δ/. The first is the <th> in think and the second is the <th> in that. The trouble is, there are very very few minimal pairs for these phonemes. The only one I’ve ever come up with is mouth (noun) and mouth (verb). [Edited to add some further noun/verb pairs received from a reader by email: teeth/teethe, sheath/sheathe and wreath/wreathe.] I only just realised this week just how difficult it is to find minimal pairs for some of the Greek phonemes, especially the vowels.

In both Latin and Greek, the long and short versions of the same vowel are different phonemes. There are tons of good minimal pairs for Latin: so, for example, malus means “bad” and mālus means “apple tree”. There are also plenty of examples for long and short /e/ and /o/ in Greek (which are also written with different letters depending on whether they are long or short, which makes it easier to spot the minimal pairs). But what about Greek /a/, /i/ and /u/? They’re written with the same letter whether they are long or short (<α>, <ι> and <υ>, respectively). But surely there are some good minimal pairs to demonstrate that long /ā/ and short /a/ are different phonemes?

I posed this question to my colleagues at dinner last night, and we were all momentarily stumped. It is incredibly hard to think of long /ā/ ~ short /a/ minimal pairs. This is partly because most of the long /ā/ sounds in Attic Greek became long /e/ sounds instead – but there are still enough long /ā/ sounds around in Attic Greek that this problem is very annoying.

A learned colleague rescued us all though, by pointing out that ariston means “best” and āriston means “breakfast”. This is a fantastic example that I’d never heard before, and I’ll be using it forever now. He also had a good one for long and short /u/, which is tough just because long /u/ is not very common: puros “wheat” vs. pūros “of fire (genitive)”.

So basically, the point of this blog post is to share these fantastic minimal pairs with the world. If you’re trying to figure out some Greek linguistics, use these examples as your long/short vowel minimal pairs – your students and/or teacher will be very impressed.