I know the general concept behind natural classes but what is the philosophy for some classes like syllabic? and are there any relationship between any two classes (like being sonorant and being syllabic)?

As I suspect the question might be marked as too broad, I must add that I'm not looking broad explanation of the charts and tables it is easily reachable by searching in the internet. But what is the relationship between these binary forms?

  • 1
    It is too broad, but you could fix that. Problem 1: there is no feature [syllabic] as of 1982. What's another example that's like it? There are many relations between certain pairs -- round and back, voice and spread glottis... would you like to pick a specific pair.
    – user6726
    Mar 22, 2016 at 23:47
  • I saw syllabic's here: linguisticsnetwork.com/an-introduction-to-natural-classes I understand the general concept but why do we have for example: [Labial] and [continuant] these two are different being labial refers to manner of articulation but being continuant refers to the general nature of the sound. How can we compare the two?
    – blackkeys
    Mar 23, 2016 at 7:17

2 Answers 2


Phonological rules and sound changes are changes to articulation. The changes are blind to phonemes or other specific sound segments -- they see only articulations and change only articulations. This characterization is rather idiosyncratic; I don't know that any other linguist would agree with it. But that's my idea about the matter.

In the study of phonological changes, the only reason to be concerned with individual sounds is methodological. We don't observe the changes directly, but we see the effects of a change on individual sounds. We then make a theory about the articulations involved: what articulations are changed? What articulations are required of neighboring parts of a pronunciation in order for the change to happen?

So the theory of phonological changes and the facts we can observe to find out about the changes are different. The theory concerns articulations; the facts are usually changes to individual sounds. We have to be concerned about the relationship between these two things. All of the individual sounds that share an articulation make up a natural class. In Generative Phonology, this is commonly annotated by describing an articulation as a conjunction of feature specifications, but that is just a notation.

As we observe phonological changes through their effects on individual sounds, we expect to see all natural classes, since the changes are articulatory, and the set of sounds sharing an articulation is a natural class of sounds. If it appears that one of the sets of sounds involved is not a natural class, then that's a problem. The observation must be wrong, or the theory is wrong. Phonologists are usually especially interested in such problems.

Some phonologists think that perceptual factors, as well as articulations, play a role in phonological changes.

One difference between the account I have just given and the standard in Generative Phonology concerns describing articulations using disjunctions of features, which is countenanced in Generative Phonology. However, disjunctions of features in phonological rules make no sense, to me.

If a feature is required to describe the class of sounds that undergo some phonological change, or the sounds in the environment that condition the change, that is evidence for including the feature in the feature system. The feature [syllabic] that you ask about was introduced in SPE with the justification that it is required to formulate a phonological change discussed by C.-J. Bailey. I'm afraid I've forgotten the details.


The philosophy behind all features is simple: rules apply to classes of segments, and features are the basis for defining those classes. We observe empirically that the behavioral sets of segments in languages are not in a strict hierarchy, for example besides the set {p,t,k} vs. {b,d,g,m,n,ŋ,f,s,i,u,o,e,a} we also find the sets {p,b,m,f} vs. all else, and {p,t,k,s,f} vs. all else. The theory of segments is that every segment is uniquely identifiable by the conjunction of some set of feature specifications, and sets of segments are identified in a rule by referring to some conjunction of features (including things that the members of the set have in common, excluding things that are not uniform within the set).

There are two main ontological approaches to features, one where features are phonetic descriptions (this is the most common view and almost certainly reflects what you’ve seen), and the substance-free theory where features don’t have invariant phonetic definitions, but rather they are the product of analyzing how segments group together in a particular language. The answer to the question about Labial vs. Continuant in the substance-assuming theory is that these features define different phonetic facts about sounds of languages, and if you have enough of these kinds of facts, you can uniquely identify any sound as distinct from any other sound, in any language.

Numerous relationships between features have been observed, which leads some people to impose grouping structure on certain features, for example [voice], [spread glottis] and [constricted glottis] are often grouped together under the label “Laryngeal”, and there are claims about the possibility of operating on multiple features just in case they are together in such a group. Other such groupings that have been proposed are “vowel height”, “vowel place”, and these in turn may be grouped into a set “place”. These groupings allow for certain kinds of simultaneous change (i.e. change in multiple features by one rule), so that [back] and [round] could change at the same time by a rule, but not [back] and [nasal].

There are other kinds of relations that have been observed, for example we know that voiced uvular stops are very rare, i.e. [+voice] does not frequently combine with [+back,-hi]. Or, there is a frequently observed relationship in vowels between the features for height and the feature nasal, but no significant relationship between frontness and nasality. There have been various attempts to make such correlations be part of the theory of phonological features, but I have to say that they have not been particularly successful, especially since the problems are not about impossible or unattested patterns, just less-common ones.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.