I don't have much of a background in linguistics, and I can't tell if some of the terms I am seeing are overloaded or unique in meaning. Specifically, I've been told that language timing can be stress, syllable or mora. At the same time, I've been told a language can be put on a gradient of stress, pitch accent and tonal to categorize it by intonation. So...does this mean that 'stress' refers to both timing and intonation, or are they different? It also seems like 'mora' and 'pitch accent' overlap, but they are used with entirely different concepts.

Perhaps the source of this information was merely convoluted. If not, can someone explain how the above terms are different? If anyone has a better metasegmentation typology I would be delighted to know what it is.


There is a phonetic concept of "timing" that relates to more or less constant units of duration in milliseconds, so that if a language is "syllable-timed" then you find more consistency of duration according to the number of syllables, but in a "stress-timed" language, stress units (e.g. the stressed syllable and the unstressed syllable after it) will be come closer to having a constant duration, regardless of the number of syllables. This presupposes that we know how to identify a "stress" (such as we have in English). [Just to be clear, my reporting this concept doesn't constitute approval]

There is also a completely different and very messy phonological (not phonetic) "gradient" of stress, pitch accent and tone. A canonical example of stress is Spanish, and a canonical example of tone is Chinese (pick your favorite dialect). Most languages are different from Spanish or Chinese, which has resulted in a third category "pitch accent". There have been unsuccessful attempts to reduce stress vs. tone to a physical difference, such as the idea that stress is about "expiratory force" but tone is about "controlling pitch".

We have found it most useful to focus on the question "what are the properties of this thing?", rather than "what a priori 'language type' is this best categorized as?". There do seem to be distinct entities "stress" and "tone", where "stress" is a rhythmic prominence relationship between syllables and tone is, well, something else -- usually about pitch contour, but often with other phonetic concomitants. There are languages, such as Eton or Zulu, which use both tone and stress. Tone tends to be more contrastive and more densely specified (e.g. most often, every syllable has some tone) and stress tends to be less contrastive and less densely specified. "Pitch accent", a concept that is highly disputed, is used for things that seem to fall between these poles, such as Tokyo Japanese or Lithuanian.

"Mora" is orthogonal to the tone / stress difference: a mora can be relevant to, or irrelevant to, both tone and stress.


The idea that language timing can be stress, syllable or mora is highly simplified, and it is actually false if taken to refer to phonetic facts about syllable length in actual speech. In real life, even in so-called "syllable-timed" or "mora-timed" languages, syllables or morae are often realized with measurably different lengths; and in so-called "stress timed" languages, the delay between subsequent stressed syllables can also vary significantly. So if there is anything real behind this classification, it must be something more subtle or psychological. This is a rather controversial topic; there's some good reading available for free online linked in the references of the Wikipedia page on Isochrony, including a Language Log post by phonetician Mark Liberman where he argues that the fact that English syllables are more variable in length than Spanish ones can be explained purely by the fact that English syllables are more variable in complexity (more coda consonants allowed, vowels can be reduced or unreduced, etc.) without needing the concept of a "stress-timed" and "syllable-timed" languages.


Here are some definitions and discussion of relevant terms that hopefully might help clarify some of the linguistic concepts. I'm sorry for the excessive length, which I'll try to trim down as I organize these thoughts better. Pretty much all of the terms you've brought up refer to phonemic concepts, or theoretical elements of the organization of a language's sound system, rather than to phonetic, directly measurable (or audible) features of language.

Overall, the “syllable” is a vaguely defined concept. Many languages can be analysed such that some sounds are considered syllabic nuclei, such as vowels or syllabic consonants. For example, the English word “chasm” has two syllabic nuclei: the first is the vowel “a”, and the second is the syllabic consonant “m”. Once you’ve determined the syllabic nuclei, a word can then be defined to have exactly as many “syllables” as it has syllabic nuclei. Then, by looking at the patterns of how the other, non-syllabic sounds vary depending on where they are in a word, we can come up with theories of syllabification that assign some or all of the non-syllabic sounds in a word to a particular syllable, as either part of the syllable’s onset (before the syllable nucleus) or its coda (after the syllable nucleus).
The difficulty lies in formulating a theory that clearly assigns every sound to a particular syllable. Sometimes, there are conflicting principles of syllabification, and so different phonologists may have different views on the syllabification of a particular word. Other contentious questions about syllabification: can a single sound simultaneously belong to more than one syllable? Can a word have sounds that are not underlyingly part of any syllable? Do all words in all languages have clearly identifiable syllabic nuclei? If anyone tells you these questions have a simple answer, be skeptical.

A mora is basically a unit of syllabic "weight" used in the analysis of a language, that may or may not constitute a full syllable on its own. A single syllable can have one or more morae. Typically, the onset (first part) of a syllable is weightless, and thus does not contribute any morae to the syllable. A short vowel is considered to be worth one mora, and a long vowel two. Consonants at the end of a syllable may also be considered to contribute a mora. When analysing Japanese, a short vowel is considered to have one mora, a long vowel is considered to have two, and the syllable codas ん and っ (sometimes transcribed /N/ and /Q/), are considered to each have one mora. It's useful to distinguish morae from syllables because some languages, like Japanese, have phonological rules that refer to both.

The use of pitch (also volume and other suprasegmental features like this) above the word level: for example, the use of rising pitch in English questions: the pitch is not associated with any particular word, but with a particular type of statement instead. All languages have some kind of patterns of intonation, but not necessarily the same ones as other languages. "Intonation" is not directly related to "tone" or "stress": non-tonal languages have intonation, and intonation usually doesn't interact with the tone system of a language, but operates on a different level from it.

Phonemic tone:
Specific words can be differentiated only by the pitch or the pattern of pitch changes that occurs when they are spoken. In some languages, tonal patterns apply to words as a whole, and knowing the placement of one particular pitch feature in a word is enough to know what pitch pattern to use for the entire word. Tokyo Japanese has a tone system of this type, where knowing a pitch value and the mora it is located on is enough to predict the pitch pattern of an entire word. The Japanese system is often called a “pitch accent”. Thus, the concept of the "mora" is used to help explain the pattern of pitch or "pitch accent" we see in Japanese words, but the concepts don't have to go together, and not all languages have both. Other languages with phonemic tone are analysed as having each syllable specified with an independent tone. However, many of these languages analysed as having an independent tone “for each syllable” in theory actually exhibit a phenomenon where tones on adjacent syllables or in the same word can affect each other: this is called “tone sandhi”. Standard Mandarin Chinese is a language analysed with tone at the syllable level and tone sandhi. Not all languages have phonemic pitch, or "tone". English doesn't.

Word stress:
In some languages, including English, some syllables in a word seem more prominent than others. This is “stress”. “Stress” may be realized in different languages with different combinations of measurable phonetic features, such as pitch, length, or vowel quality. The term “stress” is typically used when there is exactly one syllable that can be identified as the most prominent, and every content word has such a syllable.

Phonemic stress:
In many languages, if you know the precise sequence of sounds in a word, you can predict with certainty where the word's stress is. In other languages, the stress cannot always be predicted; there could be two words with the same sequence of sounds but different placements of stress. English falls into the latter category; this kind of unpredictable stress is called “phonemic stress” or “lexical stress”. Spanish is also in this category; although the stress is predictable from the spelling, it can’t be predicted if you only know the sequence of consonant and vowel sounds.

Connections between Stress, Tone, and Vowel Length
Of phomenic stress and phonemic tone, languages can have just one, just the other, both, or neither. Both pitch and stress often interact with vowel length as well; commonly, long vowels can have more types of pitch than short vowels, such as "contour pitches" like falling or rising-falling, and long vowels often attract stress in some way or get shortened in non-stressed syllables.

In many languages that have both phonemic stress and tone, the two interact in some way. This combined system of stress and tone (and perhaps also vowel length) is often called “pitch accent”. This isn’t particularly related to the Japanese “pitch accent”, so I wouldn't attach much importance to the fact that the same term is used for both phenomena. In particular, there are languages that have a typical stress system, with all words having a single most prominent syllable, and that only have a contrast between different pitches for that one stressed syllable. Many Baltic languages have a system like this, which also interacts with vowel length. Lithuanian is an example of a language with this type of "pitch accent".

Timing and Isochrony
The concept of “timing” is usually presented as the idea that fluent speakers of languages will try to evenly space out some linguistically significant feature of their words. Supposedly, a speaker of a “stress-timed” language will adjust the speed at which they speak syllables so that timing between stressed syllables is approximately constant. A speaker of a “syllable-timed” language is supposed to adjust the speed at which they speak so that every syllable is approximately the same length. And a fluent speaker of a “mora-timed” language supposedly adjusts their speech unconsciously so as to have an approxomately even timing between any two successive morae. So, it’s very possible you might have heard someone make the claim that all languages fall into one of these types.

One complication with this view is that other factors besides stress, number of syllables, and syllabic weight can affect how long it takes phonetically to pronounce a syllable on average. In English, both “large” and “bee” are single stressed syllables, but the first has a more complicated syllable structure than the sceond, which tends to make it take longer to pronounce. There are also good theoretical reasons to think of these three classifications as hypothetical extreme points on a spectrum rather than a tripartite division: even though English is supposedly syllable-timed, most analyses of British English also identify a phonemic distinction in vowel length, which would seem to conflict with a rule of isochrony based purely on the position of stress and nothing else. Despite these issues, there are linguists who believe there is a real phenomenon underlying the idea of isochrony, but as you can see, it is not at all a clear-cut classification.


So...does this mean that 'stress' refers to both timing and intonation, or are they different?

Both, and to other things, but it would be better to change "refers to" to the passive "is referred to by". You recognize stress by its footprints, sort of like Bigfoot.

Stressed monophthongal vowels in English don't reduce to schwa, so when I hear an unreduced monphthongal vowel, I know that vowel is stressed. Stressed vowels like to be accompanied by syllable onset and syllable offset consonants, while unstressed vowels reject them, so when I find an intervocalic consonant leniting in a way I associate with syllable offset consonants, that tells me the following vowel is unstressed.

For instance "bacon" can have [x] substituted for the [k], so I know from that that it's syllabified "ˈbac.on" with unstressed second vowel. But the [k] in "Baconian" doesn't lenite, so that [k] must belong to the syllable of the following stressed vowel: "Ba.ˈconian".

The pair mentioned in SPE, "Plato"/"motto" is like this. The t in the first is not flapped, but rather aspirated, so the second vowel must be stressed: "ˈPla.ˌto", while the t in the second is flapped, so it must precede an unstressed vowel: "ˈmott.o".

  • "Stressed monophthongal vowels in English don't reduce to schwa, so when I hear an unreduced monphthongal vowel, I know that vowel is stressed". That inference requires the further premise that all unstressed monophthongal vowels in English reduce to schwa, which is true only in SPE.
    – user6726
    Apr 17 '15 at 22:16
  • @user6726, I don't know just what issues you have with the SPE vowel reduction treatment, but one problem I know about is that some of us have reduced unstressed vowels other than schwa.
    – Greg Lee
    Apr 18 '15 at 2:41
  • So is it your opinion that we can move from the fact that stressed vowels in English do not reduce to schwa to the broader conclusion that being "unreduced" proves that a vowel is stressed? How do you determine that a vowel is "reduced" when it is not schwa?
    – user6726
    Apr 18 '15 at 5:19
  • No, no, @user6726. There's no discovery procedure you can follow to pop up the right answer. Modern linguistics is theoretical. You make a series of assumptions, pursue the implications, and see if this leads to correct conclusions. There is no guarantee you can figure out what the stresses are. I was just giving some hints.
    – Greg Lee
    Apr 18 '15 at 5:44
  • About other reduced vowels, consider "phonetic", which I say with a flapped t, so that the following vowel must be unstressed, even though it's not much like a schwa -- too high and back.
    – Greg Lee
    Apr 18 '15 at 5:53

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy