English: "human" (phonology)
- Thread starter berndf
- Start date
What do you mean by "phonemically perceive"? The phoneme is undoubtedly /h/, isn't it, regardless of its phonetic realization?
How can it be [j] when the diphthong right after is [ju:]?
I totally agree. It's very inconsistent in English. In AE I mostly hear [j], in BE more often [ʝ].
No, that's just the spelling. The first sound is not /h/, maybe aspirated ʰʝ.
Right, the sound (or phone) is not [h], but the phoneme is /h/ (note that square brackets represent phones and slashes represent phonemes). It's not just the spelling (graphemes are represented by angle brackets, as in <h>). /ç/ and /ʝ/ are not phonemes in English.
That's why I'm not sure what Bernd meant by phonemic perception. The phonemic value of the sound is fixed regardless of its phonetic realization or perception.
I generally perceive it as [ç].
I usually perceive a notably difference to German [ç]. It is mostly much closer to [ʝ].
Not for me. The phoneme /h/ is used in help, hire. This is never used in human. I would suggest phoneme /j/ with different realisations as [j] or [ʰʝ] or [ʝ] or [ç]. But calling this the phoneme /h/ is completely off the chart for me.
There are at least two phonemically distinguished pronunciations of human: one with a mute h starting with the same sound as utility and one with a audible h. And relisations with [ç] and [ʝ] both relate to the latter. I wanted to give you an example in English where [ç] and [ʝ] belong to one side and [j] to the other side of a phonemic divide.
I once learned that a phoneme is the smallest linguistic unit bringing about a change of meaning in minimal pairs. I can't think of any one minimal pair in English where [ç], [ʝ] and [j] belong to different phonemes; they are all allophones or variations of the phoneme /j/, aren't they? In English, [ç] and [ʝ] only appear in words starting graphemically with <hu>.
Like this I could agree:
...a phoneme is the smallest linguistic unit with the potential of bringing about a change of meaning in minimal pairs.
What this means is that differences below the threshold of phonemic relevance cannot be used to distinguish words, i.e. pronunciations that only differ in non-phonemic characteristics will not be distinguished. Langusges have to distinguish between relevant and irrelevant differences. Otherwise poeple wouldn't understand each other. And phonemes provide a grid of equivalence classes of realization that carry the same information.
No, [ç], [ʝ] represent the phoneme sequence /hj-/. Another example is [ʍ], which occurs in dialects that do not merge which and witch. Usually, native speakers perceive [ʍ] as the phoneme sequence /hw/.
[ç] is clearly a pure phoneme for me, not a sequence. [ç] is not aspirated. One can realise [f] or [θ] or [ç] or [ʝ] without prior aspiration.
As the first phoneme in "human" it might be correct to assume the /hj/ sequence. Then [j] would be one possible realisation of /hj/ besides [ʰʝ] or [ʝ] or [ç] or [ʰç].
Then, however, I would argue that the sequence /hj/ is just a placeholder for a true phoneme not explicitly defined in English because it is so rarely needed.
Let's assume that every native English speaker would use the common AE pronunciation [juːmən]. Then I would straight-forwardly argue that the phonem is /j/, as in words like use, yes. Do you agree?
Since [ʰʝ] or [ʝ] or [ç] or [ʰç] are no possible realisations of /j/ in other words like use, yes, the phoneme /j/ can not be the pure phonem in "human". That's why I agree that "human" starts with either an undefined new phoneme or in lack of it with /hj/ as unit, but not as sequence.
Sure, for you. Phoneme structures are always language dependent. That is the whole point.
The phoneme is undoubtedly /h/ - the [j] is introduced because of the /u/ (as in 'use') and then leads to palatalisation, thus [hjumən] -> [çumən]. The same process occurs in many other words, for example 'sure' ([sjuɹ] -> [ʃuɹ] and onwards), 'duty' ([djuti] -> [dʒuti]) and 'stupid' ([stjupɪd] -> [stʃupɪd]) amongst many others.
Many British dialects drop /h/ entirely, thus [jumən] or even [ʔjumən], though I don't how prevalent this is as a general trend in America. I do know that palatalisation before /u/ is less frequent in America than in Britain, and they say things like [duti].
Many British dialects drop /h/ entirely, thus [jumən] or even [ʔjumən], though I don't how prevalent this is as a general trend in America. I do know that palatalisation before /u/ is less frequent in America than in Britain, and they say things like [duti].
Phonemic analysis comes down to the most parsimonious description of the meaningful distinctions made by a language.
<hue/hew/Hugh> and <you> are distinct in most if not all varieties of English. Influenced perhaps by the spelling, I'm sure most native speakers would perceive this as /hju/ v /ju/.
It seems no more useful to posit a new extra phoneme here than it would be to suggest that the two Ls in little should be counted as different phonemes: no one doubts that in most varieties of English, the realisations of /l/ are quite different, but as they pattern neatly within the overall phonology as one phoneme, this is the most obvious and, I think, universally adopted analysis.
The case is slightly different with /h/ v /ŋ/: while they are distributed distinctly, and one could use the same sign for each without causing confusion, most people would feel that the realisations are too far apart for them to be assigned as one phoneme, and there are no structural reasons in the language that suggest any relationship.
In the end there can be no perfect phonological analysis, it is simply the structure that seems to make most sense.
<hue/hew/Hugh> and <you> are distinct in most if not all varieties of English. Influenced perhaps by the spelling, I'm sure most native speakers would perceive this as /hju/ v /ju/.
It seems no more useful to posit a new extra phoneme here than it would be to suggest that the two Ls in little should be counted as different phonemes: no one doubts that in most varieties of English, the realisations of /l/ are quite different, but as they pattern neatly within the overall phonology as one phoneme, this is the most obvious and, I think, universally adopted analysis.
The case is slightly different with /h/ v /ŋ/: while they are distributed distinctly, and one could use the same sign for each without causing confusion, most people would feel that the realisations are too far apart for them to be assigned as one phoneme, and there are no structural reasons in the language that suggest any relationship.
In the end there can be no perfect phonological analysis, it is simply the structure that seems to make most sense.
Both can occur intervocally.
I hear [ʰʝ] or [j] depending on the person (it might be regional too). A slight /h/ followed by /j/ is most common, though more and more people are dropping the /h/. This is the same for huge too.
It is certainly not [ç] as in German though.
I may be wrong, but I don't think /h/ can occur as onset of an unstressed syllable. A word like Haha, Hee-hee, is always stressed on the first syllable, for me at least.
/ŋ/ by contrast can only appear intervocalically before an unstressed vowel, as in Ringer.
I agree. Intervocalic /h/ can only occur after an unstressed prefix syllable as in behold.
The Germanic /h/ originated as a word-initial allophone of /x/ and somehow it still shows.
Last edited:
Well, we do have 'rehash' and 'rehab'.
What do you mean by these examples? That intervocalic /h/ also occurs after stressed prefix syllables or that different rules apply to Romance loans? [BTW: rehash fits my description.]
I can't help but interprete all the recordings on Forvo as [çju̟ːm(ə)n] or [çu̟ːm(ə)n].
However, it should be noted that Russians percieve voicing differently (since all Russian voiced plosives and fricatives are intensively voiced and opposed by that single feature to their unvoiced counterparts).
However, it should be noted that Russians percieve voicing differently (since all Russian voiced plosives and fricatives are intensively voiced and opposed by that single feature to their unvoiced counterparts).
They were just examples of words where /h/ occurs as the onset of an unstressed syllable, which jimquk speculated was probably not possible in English. Another example is 'cohort' (another romance loan, of course, but from French so presumably the /h/ was added at some point under influence from spelling). Anyway, it's only a minor point. I don't think you find /h/ very often as an onset of an unstressed syllable, but it does happen.
To respond to the initial question, I (British English speaker) perceive a /h/ in the recordings you linked to. Careful listening, and introspection as to my own pronunciation, allows me to hear the [ç] in each. I don't think I would perceive a [j] as an /h/, however; I would consider this an alternative pronunciation of the word (one that I would primarily associate with varieties of American English) rather than allophonic variation.
I wouldn't describe these syllables as "unstressed" in any of these example. In cohort and rehab, I would argue that the second syllable carries secondary and the first syllable primary stress in these words. I would argue that unstressed pronunciation of the second syllable would produce cohort = [coʊɚt](General American)/[cəʊət](RP) and rehab = [ɹiəb](General American)/[ɹiːəb](RP).
Rehash is a bit special because it exists as a noun with primary stress on the first syllable and as a verb with primary stress on the second syllable.
Webster agrees with me. In all three examples (in the case of rehash only the noun of course), they mark the second syllable as carrying secondary stress.
Right.
Well, Webster generally puts a secondary stress mark on syllables with an unstressed strong vowel, e.g. the penultimate syllables in 'irritating' or 'hamburger', and the final syllable in 'educate'. It doesn't make any sense to me, but that's simply the American tradition. British English dictionaries don't do this, so secondary stress is not marked on those three words, or in rehash/rehab/cohort.
Exception that proves the rule?
The Cambridge, Collins and Longman dictionaries don't show secondary stress in those words. Oxford (in both the free Lexico dictionary and also the paywall OED) show it only for the US pronunciation. I imagine this is intended to reflect the different transcription traditions rather than an actual difference in pronunciation.
Collins does not indicate stress at all (koʊhɔːʳt). The phonetic transcription system of the OED has no symbol for secondary stress.
Apart from that, RP indeed has more unstressed syllables then General American. It is not only a question of transcription tradition.
Apart from that, RP indeed has more unstressed syllables then General American. It is not only a question of transcription tradition.
Look again – stressed syllables are underlined. It's not much of a system because it doesn't distinguish primary and secondary stress (e.g. see the transcription of 'rehabilitate').
I'm not sure why you say this – perhaps we're looking at different sources? On the Lexico site, the secondary stress symbol is present in 'cohort' if you change 'UK dictionary' to 'US dictionary' (it's a drop-down menu next to the search box). They use exactly the same symbol as Merriam Webster. On the full (paywall) OED, the UK and US transcriptions are next to each other. In the entry for 'cohort', the transcription for US English shows the secondary stress, again using the same symbol as Merriam Webster.
Probably. I still go to my bookshelf to consult the OED and not to a computer.
But as you said, it is a minor point. We should probably leave it at that.
I don't know how widespread it is in other dialects, but in NYC English /'jumən/ prevails.