- English vowels in most dialects have various possible degrees of length.
In the following words, relatively long variants of the vowels are indicated
with a following [:].
Based on these examples, say what the phonetic context is for the long allophone of English vowels.
- hat [hæt]
- had [hæ:d]
- gas [gæs]
- jazz [jæ:z]
- mate [met]
- made [me:d]
- roast [rost]
- rose [ro:z]
- hoop [hup]
- tube [tu:b]
- buck [bʌk]
- bug [bʌ:g]
(This is not part of the answer, just how you might get to the answer.)
Looking at the whole list of words, we can see that both the long and the short vowels
occur between consonants in all of the words.
Since we're looking for complementary (non-overlapping) distributions, it can't have
anything to do with occurring before or after consonants or vowels.
So it must have something to do with what kind of consonant occurs before
or after (or both) for the two allophones.
Looking at the context before the vowels first, we notice in the first two wrods that
exactly the same consonant, [h], can occur before the short and the long allophones of
The same thing is true for /e/ since both allophones can occur after [m].
Obviously what comes before the vowels can't be complementary because the same
consonsants can occur with both allophones.
So the relevant context must be after the vowels.
Let's see which consonants occur after the short allophones: [t, s, p, k].
And which after the long allophones: [d, z, b, g].
No consonants are shared in the two sets; that is, there's no overlap.
So we're on the right track.
But we need to generalize about the sets, that is, say what it is that all of the
consonants in each set have in common.
All of those in the first set are voiceless, while all of those in the second
set are voiced.
The long allophone occurs before voiced consonants.
The Japanese phoneme /s/ has two allophones: [s] and [š]. Based on the following
words, say what the phonetic contexts for the two allophones are.
- [saya] 'pod'
- [kasa] 'umbrella'
- [senkyo] 'election'
- [mise] 'store'
- [sono] 'that'
- [heso] 'navel'
- [suši] 'sushi'
- [hanasu] 'speak'
- [šiku] 'spread'
- [šima] 'island'
- [šite] 'doing'
- [kuši] 'skewer'
- [sašimi] 'sashimi'
- [meši] 'rice'
- [sasemašita] 'caused'
[š] occurs before [i]; [s] occurs elsewhere (before
In Modern English, as you know, the fricatives [f, v, θ, ð, s, z] are all
But in Old English, although all of these phones occurred, they made
up only three phonemes, each with a voiceless and a voiced allophone:
[f, v], [s, z], [θ, ð].
The voiceless allophones are the more general (default) forms.
Given the following words, (i) say what the phonetic context for the
voiced allophones is, and (ii) say how the change from voiceless to
voiced in this context is an example of assimilation.
Hint: the context includes both what precedes and what follows the
([:] indicates vowel length, and phonetic details of vowels are not
indicated because they are irrelevant.)
- [fæst] 'firm'
- [full] 'very'
- [æfter] 'after'
- [klif] 'cliff'
- [heəvon] 'sky'
- [seva] 'mind'
- [hævde] 'had'
- [hweərvan] 'return'
- [æ:vre] 'always'
- [sunu] 'son'
- [la:st] 'track'
- [hu:s] 'house'
- [hors] 'horse'
- [ræ:zan] 'to attack'
- [i:zern] 'iron'
- [ræ:zde] 'attacked'
- [bizgu] 'occupation'
- [θæ:əw] 'custom'
- [wraθ] 'angry'
- [so:θ] 'true'
- [θiyeθ] 'receives'
- [kweðan] 'to say'
- [swi:ðre] 'right hand'
- [wraðu] 'support'
- [furðor] 'further'
(This isn't part of the answer.)
The fact that I know assimilation is involved can help me with the
Since the allophones in each pair differ only in voicing and
since assimilation means changing one allophone into another to make it more like
its neighbors, this must mean that the voicing of the neighbors of these consonants
has something to do with the complementary distribution that we're looking for.
So let's start with what comes before.
One thing that seems to be true is that the voiceless allophones can occur at the beginnings
and ends of words whereas the voiced allophones can't occur in either of these places,
but that still doesn't qualify as complementary because the voiceless allophones
don't have to occur at the beginnings or ends; they can also occur in the middle
So we have to show how the allophone pairs fail to overlap when they occur in the
middle of words.
Let's start with what comes before.
Clearly both the voiceless and voiced allophones can follow both vowels and consonants.
so that doesn't help.
And both the voiceless and voiced allophones can precede both vowels and consonants (though
not necessarily in the middle of words).
But remember this should have something to do whether things in the context are
voiced or voiceless.
Remember that vowels are voiced.
Let's look at the voiceless allophones first.
In the words we have, these are always preceded by voiced phones (unless they appear
at the beginning of a word).
And they can be followed by either voiced or voiceless phones.
What about the voiced allophones?
In all words these are both preceded by and followed by voiced phones.
Now let's check to see whether the voiceless allophones are never
both preceded by and followed by voiced phones.
So the easiest way to state this is to start with the more specific allophones,
that is, the voiced ones, and then treat the more general ones, the voiceless ones,
as the "elsewhere" case.
(i) The voiced allophones ([z], [v], [ð]) occur
between two voiced sounds. (ii) This is assimilation because the
fricative is agreeing with the sounds around it in voicing (it is
taking on the voicing of its neighbors).
The following words are from one dialect of Tzeltal.
['b] is a voiced glottalized bilabial stop, roughly a [b]
accompanied by a glottal stop.
Recall that [t'], [p'], and [k'] represent voiceless
ejective (glottalized) stops.
- [bi] 'what'
- [bu t'il] 'as'
- [hba] 'myself'
- [šbahth] 'goes'
- [sba] 'him/herself'
- [t'uhbil] 'beautiful'
- [ilbil] 'seen'
- [tahb] 'twenty'
- [ti'bal] 'meat'
- [ho'bel] 'San Cristóbal' (a city)
- [ma'ba] 'not'
- [ča'b] 'honey'
- [haye'b] 'how many'
- [tuth] 'small'
- [tulel] 'to harvest'
- [htath] 'my father'
- [čitam] 'pig'
- [path] 'back'
- [nahth] 'tall'
- [sith] 'fruit'
- [t'ut'] 'greedy'
- [t'anal] 'heaped'
- [t'ulel] 'to pour'
- [yut'il] 'inside'
- [naht'] 'long'
- [path] 'back'
- [pohph] 'mat'
- [spuy] 'his snail'
- [hpikh] '8000'
- [k'opoh] 'spoke'
- [sp'uy] 'squashed'
- [p'ihp'inel] 'to spread'
- [snop'] 'seized'
- [lap'ap'] 'sticky'
- [hp'itp'on] 'throbbing'
- The Tzeltal phoneme /b/ has two allophones, [b] and ['b].
Using the words above, which are representative of the contexts in which the
allophones occur, say what the complementary distribution of the allophones is.
['b] occurs after vowels; [b] occurs elsewhere.
- The Tzeltal phoneme /t/ has two allophones, [t] and [th]. Using the
words above, which are representative of the contexts in which the allophones occur,
say what the complementary distribution of the allophones is.
[th] occurs at the ends of words; [t] occurs elsewhere.
- Recall that the Spanish phonemes /b/ and /g/ each have two allophones,
stops ([b] and [g]) and approximants ([β] and γ) and that the stops are
used when the consonant begins a word after a pause.
But they are also used when these consonants follow a nasal consonant.
In fact if we look at a lot of Spanish words in context, we see that
the only nasal consonant that occurs before /b/ is [m], and the only
nasal consonant that occurs before /g/ is [ŋ].
In other words these sequences are possible: [mb], [ŋg], and these are
not: [mβ], [nb], [ng], [nγ], [ŋγ].
Some words ending in nasals even change their pronunciation to maintain
So consider the words un and con, which normally end in
a dental nasal [n̪].
When they are followed by /b/ or /g/, however, they take the form of [m]
or [ŋ], for example, un vaso [um'baso] 'a glass',
con gusto [koŋ'gusto] 'with pleasure'.
Explain what is going on, that is, why the nasal consonant changes and
why /b/ and /g/ are not realized as approximants after nasals.
Your explanation should be in terms of assimilation.
In addition to the voiced stops /b/, /d/, and /g/, the voiced
fricative /z/, and the nasals
/m/, /n/, and /ŋ/, Lingala has a set of prenasalized
voiced stops and fricatives, that is, stops that begin with nasalization.
I'll write them with a superscript nasal consonant symbol preceding
the stop symbol, for example, /mb/ for the
voiced bilabial stop beginning with nasalization and bilabial closure ([m]).
The prenasalized stops that occur in all dialects of Lingala are
- Say why you think
these are the only prenasalized stops and fricatives that
occur and not, for example, /mz/ or /ŋd/
The prenasalized stops and fricatives that
occur have the same place of articulation for the nasal part and
the stop or fricative part. This is easier to pronounce than
a combination with different places of articulation, such as
/mz/ (bilabial + alveolar).
Given the following words, say how you know that
(i) [b] and [mb] belong to separate phonemes,
(ii) [m] and [mb] belong to separate phonemes,
(iii) [n] and [nd] belong to separate phonemes,
(iv) [g] and [ŋg] belong to separate phonemes.
Recall that the best way to establish that two phones belong to
separate phonemes is to find a minimal pair for them.
([´] over a vowel marks a high tone; low tone is unmarked.)
- [kozimba] 'to trick'
- [mbáŋgu] 'fast'
- [béŋga] 'call!'
- [gúmbá] 'fold!'
- [kozima] 'to be extinguished'
- [núka] 'gather'
- [mbéŋga] 'vocation'
- [ndúka] 'dam'
- [ndáko] 'house'
- [koŋgala] 'to be wild'
- [ŋgúmbá] 'spleen'
(i) [béŋga] and [mbéŋga] are a minimal pair for [b] and [mb].
(ii) [kozima] and [kozimba] are a minimal pair for [m]
(iii) [núka] and [ndúka] are a minimal pair for
[n] and [nd].
(iv) [gúmbá] and [ŋgúmbá] are a minimal pair for
[g] and [ŋg].
- Old English had both short and long low front vowels, [æ]
From the following examples, say how you can know that these two phones
belong to different phonemes.
- [sæ:] 'sea'
- [θæ:r] 'there'
- [æ:t] 'food'
- [mæ:st] 'most'
- [græ:y] 'gray'
- [dræ:van] 'to drive'
- [klæ:ne] 'clean'
- [bæk] 'back'
- [æt] 'at'
- [wæter] 'water'
- [fæstan] 'fasten'
- [mæst] 'mast'
- [næyl] 'nail'
- [hwæt] 'what'
There are two minimal pairs for [æ] and [æ:]: [æt] / [æ:t] and [mæst] / [mæ:st]
- In some accents of southeastern England, the tense (long) vowels include
u:/ (a long rounded, high, central vowel), and
/ʌu/. From the following examples, say how you know these are separate
phonemes in this accent. Note: they do not correspond directly to
phonemes in other accents such as General American.
- moan [mʊun]
- tow [tʊu]
- nose [nʊuz]
- sole [sʊul]
- mown [mʌun]
- soul [sʌul]
- knows [nʌuz]
- toe [tʌu]
- news [n
- moon [m
- two [t
- soon [s
To show that the three phones are separate phonemes, we need to find a minimal pair (or near minimal pair) for each of the three possible pairs. But a minimal triple will also work because it's actually three minimal pairs. There are two minimal triples among the words: moan / mown / moon and tow / toe / two.
- In sign languages the main dimensions along which syllables differ
are handshape, location (the place on the body or in space where the sign
is made), movement (the motion of the articulators in space), and
orientation (the direction that the palm "points"). Given the following
ASL signs, say how you know that location and movement are
contrastive dimensions in ASL.
To show that location and movement are contrastive, we need a minimal pair (or near minimal pair) for each.
For location, the signs for 'dry' and 'summer' are a minimal pair; the two signs are the same except that one is produced near the forehead, the other near the bottom of the face.
For movement, the signs for 'airplane' and 'fly' are a minimal pair; the two signs are the same except that one uses two forward movements, the other uses one.
Below are some words from Argentine Spanish containing the phones [s]
and [h]. (Each has a different meaning, but the meanings are left off
because they might make it easy for students who know Spanish.)
Syllable boundaries are marked with a space, and [x] represents a
voiceless velar fricative.
Based on these words only, are these two sounds in complementary
or overlapping distribution in the language?
If the distributions are complementary, say what the contexts for each phone are.
If the distributions are overlapping, say how you know.
(Hint: [h] is the more restricted phone.)
- [ph] (aspirated, voiceless bilabial stop), [p] —
[ph]: at the beginning of a stressed syllable, [p] elsewhere
- [s], [š] — overlapping, sip/ship is a minimal pair
- [sa 'lar]
- ['ka sah]
- ['se so]
- ['swer te]
- ['swa reh]
- [gon 'sa leh]
- [dye si 'sye te]
- [rre 'swel to]
- ['xwi syo]
- ['ka xah]
- ['fyeh tah]
- ['xeh to]
Complementary distribution: [h] occurs at the ends
of syllables, [s] elsewhere.
(In fact [h] is an allophone of /s/ in this dialect and many other
dialects of Spanish.)
In Amharic, consonants can be simple, for example, [t] and [m], or long,
for example, [tt] and [mm].
Given the following words, say whether consonant length is a contrastive
dimension in Amharic, and explain how you know.
- [yɪmɛtal] 'he hits'
- [mɛtta] 'he hit'
- [mɛla] 'scheme'
- [tɛdɛrrɛgɛ] 'it was done'
- [bɛrr] 'door'
- [nɛgɛ] 'tomorrow'
- [mɛlla] 'it got full'
- [nɛgga] 'it dawned'
- [yɪmmɛttal] 'he got hit'
- [zɛr] 'seed'
- [k'ɛllɛlɛ] 'it got easy'
Consonant length is a contrastive dimension.
There is a minimal pair for consonant length, [mɛla] / [mɛlla].
(And there are two near minimal pairs, [yɪmɛtal] / [yɪmmɛttal] and
[nɛgɛ] / [nɛgga].)
Amharic /b/ has two allophones, [b] and [β] (a voiced bilabial fricative).
Given the following words, which are representative of words containing
this phoneme, say what the complementary distribution of these allophones
is. (Remember from the last problem that consonant length is contrastive in Amharic.)
- [laβ] 'sweat'
- [rɛhaβ] 'hunger'
- [nɛβɪr] 'leopard'
- [nɛbbɛre] 'was'
- [bɪrd] 'cold'
- [bal] 'husband'
- [bɛlla] 'he (it) ate'
- [tɛβɛlla] 'he (it) was eaten'
- [kɪβɪr] 'honor'
- [tɛβabbɛru] 'they were united'
- [aβro] 'together'
- [gɛβs] 'barley'
- [wɪβɛt] 'beauty'
- [tɛsɛbbɛrɛbbɛt] 'it was broken by it'
- [kɛβt] 'livestock'
- [t'ɪβɛβ] 'wisdom'
- [t'ɛbbaβ] 'narrow'
- [ambɛssa] 'lion'
- [arba] 'forty'
- [albɛlam] 'I don't eat'
[β] occurs after vowels unless it is long; otherwise [b] occurs.
- In Tzeltal /t/ (with the context-sensitive allophones [t] and [th]) and /t'/ ([t']) are separate phonemes.
Say how you can know this from the words in problem 4 in 4.8.1 above.
Minimal pair: [tulel] (o), [t'ulel] (w).
- In Tzeltal /p/ (with the context-sensitive allophones [p] and [ph]) and /p'/ ([p']) are separate phonemes.
Say how you can know this from the words in problem 4 in 4.8.1 above.
Minimal pair: [spuy] (ab), [sp'uy] (ae).