Inventories and phonotactics (1 hour)

↵ Back to module homepage

Phonology focuses on several main things, all of which are interconnected:

In the previous activities, we mainly focused on alternations: we examined the sound changes that happened in languages. But to figure out the alternations, we were also thinking about the other issues too.

Phonotactics is the rules about what sounds can or cannot go together in a language. For example, brick is a real word in English, and blick is not, but both sound like English. On the other hand, bnick doesn't sound like English, and it's hard to even pronounce. Crucially, both [b] and [n] exist in English (think of words like bat and neck); they just cannot go together in a consonant cluster at the beginning of a word. That's phonotactics. You can think of phonotactic rules in Chinese, too. For example, Mandarin has [s] (in words like 訴) and [t] (in words like 土), but they can't go together: there is no Mandarin word pronounced stu. That's phonotactics.

Think back to how we discovered the alternations in Maasai. We did it by looking at phonotactics. We noticed that [g] always occurs after [ŋ], and [k] never occurs after [ŋ]. That's a phonotactic restriction.

Inventory is the phonemes that exist in a language. Different languages may organize the same sounds into different categories (different phonemes). For example, [s] and [ʃ] are sounds that belong to two different phonemes in English (think about the words seep and sheep; these are a minimal pair, so they prove that [s] and [ʃ] are not just versions of the same sound, they are actually pronunciations of two entirely different phonemes). In Japanese, though, these are just two versions (allophones) of the same phoneme: Japanese has a sound which is pronounced as [s] in most contexts, but is pronounced as [ʃ] if it's before an [i] (think about the Japanese word for chopsticks: hashi).

As you can see from the above discussion, discovering the phoneme inventory of a language is not just a matter of listing all the sounds you hear in a language. Sometimes you might hear several different sounds (phones) which are all just different versions of the same phoneme. So, to figure out what phonemes a language has, you have to consider alternations and phonotactics, look for minimal pairs and complementary distributions, in order to figure out what the phonemes are.

That's the reason why I say the things we look at in phonology are interconnected. Alternations, inventories, and phonotactics are kind of different concepts, but it's actually impossible to analyze one without analyzing the others at the same time.

Now let's apply these phonology concepts to examine a question that's closer to home: how many tones does Cantonese have?

Many people say that Cantonese has nine tones. Here are examples:

fan1 (55)
fan2 (35)
fan3 (33)
fan4 (21)
fan5 (13)
fan6 (22)
fat7 (5)
faat8 (3)
fat9 (2)

In the above list, I have written the pronunciation of the tone in Jyutping (like the Cantonese version of Pinyin), where the small number indicates the tone category (Tone 1, Tone 2, etc.). After each one, in parentheses, I have written the phonetic tone description using Chao letters (if you don't remember what these mean, check the Phonetics module for a reminder).

Tones 1-6 are considered "full" tones, and tones 7-9 are considered "short" or "checked" tones.

Think about this group of words using all the concepts you've learned so far from phonology. Look for minimal pairs, look for complementary distributions, think about things that might be different versions of the same tone vs. things that might really be different tones. After thinking about this, do you agree that there are 9 tones? Why or why not?

The standard phonological analysis of Cantonese is that there are actually six tones, not nine. Notice that tones 1-6 are in complementary distribution with tones 7-9: tones 1-6 only occur in "open" syllables (syllables that end with a vowel or nasal), and tones 7-9 only occur in "checked" syllables (syllables that end with a stop like [p], [t], or [k]):

Open syllables Checked syllables

Tone 1 (55)
Tone 2 (35)
Tone 3 (33)
Tone 4 (21)
Tone 5 (13)
Tone 6 (22)

Tone 7 (5)
Tone 8 (3)
Tone 9 (2)

In other words, Tone 7 is probably just another version of the higher tones: historically, when those tones were in a checked syllable, they got shortened into this tone. Likewise, Tone 9 is just another version of the lower tones, etc.

This is another example of a situation where examining phonotactics (which tones appear in which kinds of syllables) shows us something about the phoneme inventory.

Now let's think about standard Mandarin. People usually stay Mandarin has four tones (or sometimes they say it has 5 tones, if they are counting 轻声 [neutral tone]). But in the Phonetics module we already saw examples of more: for example, Tone 3 is sometimes a full tone 3 (with Chao numbers 213: it starts out in the middle, then gets low, then rises again), and sometimes a half tone 3 (with Chao numbers 21: it goes down and doesn't come up). There are other similar variations: for example, a "full" tone 4, when pronounced by itself, has Chao numbers 51 (it starts very high and drops very low), but a tone 4 in a compound word with another syllable after it (like 罷工) is pronounced more like 53 (it doesn't drop quite as much). Why are these not treated as different tones? Using phonology concepts, explain why people treat Mandarin as having four tones—i.e., why are "213" and "21" both considered to be Tone 3, and why are "51" and "53" both considered to be Tone 4?

When you have finished these activities, continue to the next section of the module: "Optimality Theory and L2 speech models".


by Stephen Politzer-Ahles. Last modified on 2021-04-19. CC-BY-4.0.