Sound: Phonology – Fonologie

In the previous chapter we saw that sounds can be different from each other in many ways. At the end of the chapter we discovered that even sounds that we think of as ‘the same sound’ can come in different variations, such as English [p ^h] and [p]. The same is true for the Dutch tapped [r] and the velar [ʀ], which are two different ways of pronouncing /r/.

We say that English [p ^h] and [p] are two different realisations of the same >phoneme /p/. As you noticed, phonemes are put between forward slashes (/). >A phoneme is the smallest unit of sound that makes a difference in the meaning of a word. It doesn’t matter if you say [p ^h ɪ n] or [p ɪ n] in English, in both cases you hear the word ‘pin’ (the latter will sound like a South African or Dutch accent). However, it does make a difference if you say [b ɪ n] (‘bin’) or [p ɪ n] (‘pin’), since these two words mean something different. Whereas [p ^h] and [p] in [p ^h ɪ n] and [p ɪ n] are different >phonetic realisations of the phoneme /p/, [b] and [p] in [b ɪ n] and [p ɪ n] are two different phonemes (/b/ and /p/). Words that differ only by one phoneme are called >minimal pairs.

A phoneme can be regarded as a basic sound of a language. The average number of phonemes per language is 35. Dutch has about 40, English is said to have 44, although these numbers depend on the method of analysis and can differ between dialects.

Allophones

You may wonder how it is possible to tell if something is just a different realisation of a phoneme (allophone) or actually a different phoneme. There is an easy way to tell the difference. We know that a phoneme is a unit of sound that makes a difference in word meaning. All we need to do now is make a minimal pair, two words that differ in only one phoneme, and mean something else. Let’s do this test for Dutch:

Although the r’s are pronounced differently, we have only one single word ‘raar’. Dutch tapped, alveolar [r] and velar [ʀ] are two instances of the phoneme /r/. As we noted above, the velar [ʀ] is pronounced at almost the same place as the Dutch [x] (often written as the letter ‘g’). So as far as pronunciation is concerned [ʀ] is much closer to [x] than to [r]. Yet, [ʀ] and [r] are different realisations of the same phoneme /r/, whereas [x] is a realisation of the typical Dutch ‘g’ sound /x/ (obviously a different phoneme than /r/).

In the south people often speak with a ‘zachte-g’ (‘a soft g’ ) which is [ɣ] in phonetic transcription. Both [ɣ] and [x] get interpreted as /x/:

So both [ɣ] and [x] are different instances of one and the same phoneme. [ʀ] and [r] are said to be different allophones of the phoneme /r/, and [x] and [ɣ] are different allophones of the phoneme /x/.

We can see that if we use [ʀ] or [r] the word meaning doesn’t change, both words get interpreted as /raar/, so [ʀa:r] and [ra:r] are not a minimal pair. Between [x] and [ɣ] the word meaning also doesn’t change, both words mean /gaar/. On the other hand, /xaar/ and /raar/ do form a minimal pair, and so /x/ and /r/ are different phonemes.

An important part of identifying phonemes in a language is to find out which ‘different sounds’ (allophones) belong to the same phoneme. Certain groups of sounds behave the same in the same phonetic environment. We have already seen an example of this in English where the group of voiceless plosives (/p, t. k/) get aspirated at the beginning of a word. This means that the change of sound from the phoneme /p/ to the allophone [p ^h] in the beginning of a word is predictable. By trying to understand these processes that change a sound, we can predict how a certain phoneme will sound in a certain phonetic environment (i.e. at the end of a word, or before a long vowel etc.).

Distinctive features

It turns out that sounds can be classified in a structural way by looking at the phonological features of a sound, and that all sounds in a certain natural class of sounds behave the same in the same environment. Phonological features are sometimes called distinctive features because these features make the distinction between two phonemes. Phonemes in turn can make a distinction in the meaning of a word. For example /d/ and /n/, both alveolar and voiced, differ only in the fact that /n/ is nasal and /d/ isn’t. This difference in only one feature of the sound already makes the difference between, for example, the word /naar/ (nasty) and /daar/ (there).

We already know 2 distinctive features. One of them is VOICE. If your vocal chords vibrate during the production of a sound, that sound is said to have the feature [+VOICE]. The other feature is NASAL. For example the /n/ is [+NASAL], but the /d/ is [-NASAL]. If we add two more features to that list we can start listing all the consonants and see if we can tell them apart by just looking at their features. The two features we need are CONTINUANT and LATERAL. If the air stream in the vocal cavity is obstructed in any way, the sound is [-CONT]. In Dutch only the [l] is lateral, [+LAT], which means that the air escapes through the side of your oral cavity, as your tongue blocks the middle.

Now let’s make a list of some of the consonants of Dutch and see if we can categorise them using these four features:

This list is far from complete. First of all there are obviously more consonants, such as [ɣ], [ʃ], [ʒ] and more. Second, there are far more features. Most phonology books list about 14, including VELAR and LABIAL, which should sound familiar.

However, using the knowledge you’ve gained in the chapter about phonetics and the simplified table above we can already explain a few phonological processes. First we can make some generalisations from this list. For example, only /p/, /t/ and /k/ have the features [-VOICE], [-NAS] and [-CONT]. Together they form the natural class of voiceless plosives (a little explosion of sounds with no use of the vocal cords). Phonological rules apply to natural classes, so we can say that in English the rule of aspiration applies to the natural class of voiceless plosives.

Natural classes aren’t mutually exclusive: for example, /p/ is also in the natural class of labials, together with /m/, /p/, /b/, /f/, /v/ and /υ/ (note that both bilabial and labio-dental counts as labial!). We will see below that this natural class plays a role in a phonological rule of Dutch.

The most important thing you should learn from this is that, with the help of these distinctive features, from the hundreds of sounds in a language we can deduce about 40 ‘underlying sounds’, phonemes, that play an important role in meaning differences. The other sounds are all predictable results of phonological rules. Apart from English aspiration, we have also seen that the /i:/ sound in /bead/ is longer than the one in /beat/. This is also true in Dutch /bi:r/ (‘bier’, beer) and /bi:t/ (‘biet’, beet root).

In fact, the Dutch rule is that vowels are always longer before [r] and in English sounds are always longer before a voiced consonant. This difference in sound quality (vowel length in this case) is predictable from the rules.

As in syntax we have again found structure in our language. All the variation is the result of a limited set of rules. Where syntacticians try to find the rules for grammar, phonologists try to find the rules for sound patterns.

Variation of a Dutch phoneme

Now that we’ve seen what we can do with phonology, let us take a closer look at a Dutch phoneme and see what happens when it is placed in different phonetic environments. The phoneme we will look at is the phoneme /n/. The /n/ is an alveolar, voiced nasal sound. We know /n/ is a phoneme of Dutch because we can form minimal pairs like:

Sometimes sounds change if they’re next to another sound. In some cases a sound gets deleted completely, this is called elision. In English this happens in words like ‘friendship’. In normal, non-deliberate speech this often gets pronounced as [f r ɛ n ʃɪp] (‘frienship’, without the ‘d’). The same goes for the Dutch ‘vriendschap’, which is often pronounced [f r ɪ :nsx ɑ p] (‘vrienschap’, without the ‘d’). In other cases the sound adapts to its phonetic surroundings, this is called assimilation. This is what happens with Dutch /n/ (and many other sounds).

To investigate whether this is indeed true, we need to find a word that attaches to other words so we know we can find /n/ in different surroundings. A good word for this is ‘on-‘ (compare to English ‘un-‘). ‘On-‘ isn’t a word by itself, it is called an affix because it needs to attach to other words. In this case it is a prefix, a type of affix, because it needs to attach to the beginning of a word (as opposed to suffixes, which attach to the end of a word).

It seems reasonable to presume that ‘on-‘ indeed ends with the morpheme /n/, because there are a lot of examples where the sound goes unchanged:

In these cases the phoneme /n/ gets realised as [n]. The alveolar /n/ also gets pronounced [n] if it precedes another alveolar sound:

However, there are other cases. When an /n/ precedes a velar sound, it changes toward the velar [ɳ] itself. Notice that the first two sounds are [ɔ ɳ], with the velar [ɳ] as opposed to [n]:

If /n/ precedes a bilabial sound, it changes towards bilabial [m] itself:

Notice that in English this change has even influenced spelling; in-partial and in-possible are actually spelled impartial and impossible.

We can catch these sound changes of the phoneme /r/ in a phonological rule. We could just make two separate rules:

1. [n] changes to [ɳ] in front of velar sounds
2. [n] changes to [m] in front of labials

However, it is more efficient to formulate one rule, as the process is very similar in both cases; [n] changes towards the following sound. First let’s formulate the rules in 1 and 2 a bit differently, using the distinctive features:

1. alveolar voiced nasal velar voiced nasal / ____velar
2. alveolar voiced nasal bilabial voiced nasal / ____bilabial

We can read this as follows:

[sound a] becomes () [sound b] when found in the following position:
( / ) before [sound c] ( ____ [sound c]).

Now we can summarise these two seperate rules as one:

• alveolar voiced nasal α voiced nasal / _____α
  (alveolar voiced nasal becomes alpha voiced nasal when found before alpha)

This sort of rule is called a context sensitive rewrite rule. In this rule α (alpha) is a variable which can represent velar or bilabial.

Note that this rule isn’t completely accurate, because α cannot represent every value (as it normally does) because /n/ never becomes a palatal sounds before a palatal ([j]). This is because we haven’t talked about all the features that can be used in the paragraph about distinctive features. However, it shouldn’t matter as the real rule is much the same but just involves a little more detail. For now, we conclude that [n], [ɳ] and [m] are different allophones of the phoneme /n/.

Diminutive – verkleinwoorden

The fact that a phoneme can present itself in different forms, allophones, leaves us with a lot of variation in pronunciation. The question is whether we can find structure in this variation. The answer is yes, of course, and we can show this by looking into the Dutch *diminutive.

The diminutive in Dutch is a suffix ‘-tje’, which means something like ‘little’. Examples of Dutch diminutives are:

This brings us to a peculiar feature of Dutch. The word ‘paard’ is written with the letter ‘d’, but is pronounced [pa:rt], with a [t]. This is because in Dutch there can be no voiced consonants at the end of a word. Pay close attention to the phonetic transcription of the following words, and notice the difference between Dutch and English:

Where we write a ‘d’ in Dutch at the end of a word, we pronounce a [t]. However, /d/ is a phoneme of Dutch, because it does make a difference in meaning in the beginning and middle of a word. Apparently there is a rule that states that voiced consonants get devoiced at the end of a word in Dutch. We know that phonological rules apply to natural classes, so if our rule is correct it should apply to other voiced consonants as well, and this is indeed the case (it even applies to loanwords from English!):

Notice that we had to borrow words from English to find words spelled with a voiced consonant. Dutch words are often already spelled with a devoiced consonant. For example the plural ‘rozen’ (roses) is spelled with ‘z’, but the single ‘roos’ (rose) is spelled with an ‘s’. Yet the pronunciation of [s] at the end of ‘roos’ is probably also a result of the devoicing at the end of a Dutch word.

Now that we know this aspect of Dutch phonology we can look at the diminutive in a little more detail. It turns out that the devoicing rule doesn’t only apply at the end of a word, but also at the end of a syllable if it attaches to a suffix. Before we attach the diminutive suffix ‘-tje’, we apply the devoicing rule. This explains the pronunciation of the following words:

So even though we spell a voiced consonant, we pronounce a devoiced consonant. This proves once more that spelling is unreliable.

The diminutive is also subject to an assimilation rule (like the one with /n/ we saw above). The sound of the suffix depends on the last sound of the stem to which we attach the suffix. If the last sound of the stem is alveolar, the first sound of the suffix is alveolar. If the last sound is labial, the first sound is labial and if it is velar a velar sound will follow:

Note that after ‘m’ the diminutive is –pje and after ‘-ing’ it is –kje.

Again, it seems that the assimilation is the result of a single phonological rule, so we can formulate a rewrite rule.

The rule we need for the diminutive based on what we have learned so far is something like: -tje becomes –pje or –kje when found after a labial or velar sound. More formally, and generalised, this is written as follows:

[tje] [αje] / α___ , where α can be labial or velar.

This rule seems to be accurate, but it doesn’t cover all the Dutch data. However, it is an example of how we can give a linguistic analysis of the variation we find in the Dutch diminutive.

For the more accurate rule we’d need more detail, but hopefully you understand how the formulation of such a rule would work. Being (passively) aware of such rules can help you in your pronunciation of Dutch. Let’s just look at a few cases where the above rule doesn’t work, without trying to explain them:

Notice that the final consonants of the stem are the same, but the diminutive suffix is different. Apparently when the vowel before the last consonant in the stem is short the suffix is different. An interesting case is the diminutive of bloem (‘flower’), which can be both bloemetje and bloempje. A possible explanation of this may be that ‘oe’ can be interpreted as a long vowel or as a short vowel. This only goes to show that there are a lot of things involved in the study of phonology that we haven’t covered yet. Another example is the ‘rhythm’ of language. There are differences in intonation and word stress between languages and these turn out to be the result of an underlying structure as well.

Conclusion

Just as sentences are words and inflections that are combined in a structural way, so words themselves are sounds combined in a structural way. Not only can you look into the way these sounds are produced and received, but it is also possible to look at the way these sounds get interpreted. It turns out that behind the hundreds of sounds used in a language there is an underlying system of a limited number of phonemes and phonological rules. The fact that there are more sounds than phonemes can be explained by the rules which predict certain sound changes.

The underlying structure is dominant in the way we interpret sounds. For example, Dutch /x/ and /ʀ/ are very close phonetically, but they’re still different phonemes because they make a difference in word meaning. On the other hand [l] and [r] seem to be rather unrelated phonetically, yet a speaker of Chinese will have trouble keeping the two apart because in Chinese they are not different phonemes.

Phonologists also study things like intonation and word stress, which are also partial to an organised set of rules. If you’re interested in this you should visit the internet, the library or consult the reference list of this online course for further reading.

The next chapter of this course considers the meaning of language. In the past chapter we have mentioned that sometimes a sound can make a difference in the meaning of a word, but what do we mean when we say that something means something? It won’t surprise you that again we can structure our thoughts about this and form a theory of meaning, and it’s actually really exciting and will definitely make you think about things from a different point of view!

next
(meaning)