Traditionally, phonology is thought of as the study of the
systematic organization of sounds in spoken languages. This is not
surprising as the phon in phonology comes from the Greek
word for 'voice', and given the absence of voice in sign language, sign
language phonology seems like a contradiction in terms. But with the
advent of scientific interest in sign languages, it has become clear
that sign languages have phonology, too. How we can bridge this seeming
contradiction?
The answer to the puzzle lies in a property that has been claimed to
be a defining characteristic of human languages (in contrast to other
systems of animal communication, even the most sophisticated ones).
This feature is called double articulation (Hockett 1960). It
states that human languages have two levels of linguistic structure: a
level of atomic meaningful elements called morphemes and a
level of atomic meaningless elements called phonemes.
Words (by which term we include signs in what follows) can consist of
more than one morpheme - for instance, if they are compounds,
like blackberry or SISTER (GIRL+SAME). If human language just
consisted of morphemes, it wouldn't differ significantly from other
animal communication systems (apart from the size of the vocabulary).
But human language does differ in two ways. First, all human languages
have rules for combining the individual morphemes, and these rules are
more sophisticated than just stringing together the morphemes into a
list. The study of these rules is the domain of morphosyntax.
Second, and more relevantly for present purposes, the morphemes, though
atomic from the point of view of meaning, are not atomic from another
point of view. From this second, formal, point of view, they are
composed of smaller units, which can combine with each other according
to rules of their own. The study of this second type of rules is the
domain of phonology.
When we examine phonemes more closely, we find that we can group them
into various classes depend on various shared articulatory features.
For consonants, the main features are given in (1). The
wikipedia
entry for distinctive feature has more details
(https://en.wikipedia.org/wiki/Distinctive_feature, accessed March 24,
2020).
Dual articulation
Phonemes functioning as morphemes
Though phonemes and morphemes are conceptually distinct, it is important
to understand that phonemes can function as morphemes. In other
words, a sound unit can be meaningless from the phonological point of
view, but meaningful from the morphosyntactic point of view. (We use
the term sound unit to avoid committing ourselves to whether a
sound is a phoneme or a morpheme.) Take the sound unit 'd' in English.
It has no intrinsic meaning in dip /dIp/ or doom /dum/.
(Examples from spoken languages are transcribed in a keyboard-friendly
encoding of the IPA
called SAMPA.) But in played /pleid/, the same
phoneme has additional morphemic status as a marker of past tense.
Similarly, the /o/ in French beau /bo/ is a meaningless part of
the word 'beautiful', but in eau /o/, it functions as the word
'water'.
Phonological features
Phonological features in spoken languages
We just said that the meaningless units of language are atomic, and the
analogy with physics holds up well. Just as we now know that atoms of
matter have internal structure, we now know that phonemes, too, have
more complex structure than was initially supposed. We'll start by
illustrating for spoken language. The way that we ordinarily identify
the phonemes in a spoken language is to gather lists of minimal
pairs - word pairs that differ in meaning but differ in only a
single sound unit. Those distinctive sound units are the phonemes. For
instance, at the level of the phoneme, hiss constrasts minimally
with his with respect to final /s/ vs. /z/.
| (1) |
|
The features each have values. For instance, voicing has the two values 'voiced' and 'voiceless',1 and it is the contrast between these feature values that distinguishes hiss /hIs/ from his /hIz/ even more minimally than the phonemic distinction between /s/ and /z/ just mentioned. Analogous contrasts can also be found with respect to the other two distinctive features. For instance, mitt /mIt/ contrasts with knit /nIt/ with respect to the place of articulation of the first phoneme (labial for /m/, alveolar for /n/). And tin /tIn/ contrasts with thin /TIn/ with regard to manner of articulation (stop for /t/, fricative for /T/).
The above distinctive features apply to consonants. Spoken languages make a fundamental distinction between consonants and vowels. Consonants are produced with some hindrance to the air flow up and out of the vocal tract (up to a completely stoppage), whereas the air flow in the production of vowels is basically unimpeded. The basic distinctive features for vowels concern the position of the tongue and the configuration of the lips, as in (2).
| (2) |
|
| (3) |
|
As we have seen in connection with incorporation, classifiers, and related phenomena in sign phenomena, these distinctive features cannot be expressed on their own. For instance, the movement roots in classifier constructions need to combine with appropriate classifier handshapes to yield gestural units that can be articulated and perceived. We might be tempt to think that handshapes can be articulated and perceived independently of other features. But even when they are used as part of the fingerspelling alphabet, they are always produced in some location and with some orientation.
Now let's compare sign languages and spoken languages. Spoken language phonemes are combinations of various distinctive features (/b/ is a labial, stop, voiced consonant; /u/ is a high, back, rounded vowel). In the same way, we can say that sign language phonemes are combinations of distinctive features, as in (4).
| (4) |
|
For expository convenience, we'll use the term sign phoneme for a combination of phonological features as in (4). If our focus is on on the meaning-bearing aspect of a sign, we'll call it a sign morpheme. And we'll use sign unit as a cover term for both - either because because we're not interested in whether a sign carries meaning or because we don't know. is Using these terms, we can say that the sign phoneme in (4) is not necessarily a sign morpheme in ASL. (It's similar to AUNTstr in ASLSignbank, but it's simpler and differs in some feature values.) Another sign phoneme is shown in (5).
| (5) |
|
The sign phoneme in (5) is also a sign morpheme - namely, the morpheme COLOR. In other words, it is the sign counterparts to the earlier examples from English and French ('d' and 'o').
We're glossing over details here, especially concerning movement and the distinction between internal-movement and path movement. Also, in more modern treatments of handshape, handshape is further specified in terms of selected fingers, as you can see in the ASLSignbank entry for COLOR. But the basic idea - that sign language phonemes are combinations of distinctive features just like spoken language phonemes - should be clear.
It is sometimes claimed that signs exhibit more simultaneity than
spoken words. But under the approach outlined here, that claim is the
result of confusing phonemes with distinctive features.
For instance, the English phoneme /s/ and the ASL sign phoneme in (5)
are equally simultaneous. For instance, the voiceless value of the
voicing feature in /s/ can't be produced without a place or manner of
articulation, just as the wiggle value for the movement feature in the
sign phoneme in (5) can't be produced without a handshape or location.
Consequences of modality
The discussion so far has focused on the similarities between spoken and
sign language phonology. There's an strong intuition, though, that
spoken and sign language phonology differ in some fundamental respect.
This intuition can be stated as in (6).
| (6) | In sign languages, sign phonemes are very likely to function as sign morphemes. By contrast, in spoken languages, phonemes functioning as morphemes are the exception rather than the rule. |
Let's pursue this intuition. We'll begin with spoken language features. Notice that there are relatively few of them, and they have relatively few values per feature. Using the numbers in (1) and (2), we get 24 (3 x 4 x 2) consonants and 18 (3 x 3 x 2) vowels for a totel of 42 phonemes. Note further that individual spoken languages don't necessarily use all features or all feature combinations. For instance, English doesn't have palatal or velar fricative consonants or front rounded vowels or back unrounded vowels. So that cuts down on the total number of phonemes per spoken language. All in all, let's say that spoken languages typically have 30-40 phonemes.2
By contrast, the articulatory space that sign languages have at their disposal is much larger than that for spoken languages, and so the features in sign languages have many more values. In addition, there are two hands, not just one vocal tract, and then there are the nonmanual articulators (the mouth and the eyebrows). As a result, sign languages have many more possible phonemes than do spoken languages. We won't try to quantify the number of possible sign phonemes, as we did for spoken languages, but the point should be clear.
Now let's assume we want to use the phonemes to convey meaning, and
let's assume further for the sake of argument that we have a vocabulary
of 800 morphemes. In sign languages, there's a good chance that each
phoneme can be used as a morpheme before the language runs out of
phonemes. That is simply not the case for spoken languages. In a
spoken language, once the vocabulary exceeds roughly alphabet size, the
language has to start using more than one phoneme to express a morpheme.
So the modality difference betwen spoken and sign languages essentially
forces spoken languages into exploiting sequentiality, and the
insight in (6) follows as a generalization.
In spoken language, the association of phonological features with
morphological features is much less salient and productive. Here is a
possible example. Like many languages, Italian distinguishes masculine
and feminine gender, and it uses suffixes for the gender/number
combinations. The table in (7) gives the morphemes along with the
phonological feature combinations for the phonemes that express the
morphemes.
Features functioning as morphemes?
There is a final loose end left to discuss. Earlier, we saw that
phonemes can function as morphemes. The analysis of phonemes in terms
of distinctive features raises the question arises of whether
the features can function as morphemes. The question arises
naturally in connection with sign languages, because early on we learn
that for certain sign clusters, the location features 'at forehead' and
'at cheek' express the meaning 'male' and 'female', respectively.
Another example: the value '7' for the feature movement is associated
with city names.
| (7) |
|
Given the paradigm in (7), one could argue that the feature tongue height (high vs. mid) is used to express grammatical gender in the plural. But this argument is weak since tongue height isn't used in the same way in the singular.
It seems sensible, therefore, to conclude that phonological features aren't productively correlated with morphological features in spoken languages in the way that they are in sign languages. A natural explanation is that phonological features have iconic origins in sign languages (for instance, in ASL, male touches cap at forehead, female ties bonnet string at cheek) in a way that is almost never the case for spoken language.
2. This estimated phoneme inventory size corresponds roughly to the numbers of letters in the alphabet. That's not surprising since the purpose of the letters of the alphabet is precisely to represent the phonemes. In an ideal spelling system for a language, the letters of the alphabet would stand in a one-to-one correspondence to the phonemes. For most spelling systems (notoriously for English), the correspondence isn't exactly one-to-one. Many spelling systems use two letters to represent a single phoneme (English 'sh' for /S/, as in shin) or more than one phoneme ('th' for /T/, as in thin, or /D/, as in these), and they use diacritics like umlauts, tildes, and so on, to modify the basic letters in order to represent additional phonemes.