PHONETICS AND PHONOLOGY PHONEMIC ANALYSIS: SOLUTIONS - R. H. Mannell 1) Artificial (nb. line up the sounds logically, and make sure that potentially phonetically similar sounds are adjacent to each other). p t k b d g m n i { a u O = (ie. stress) Compare [pb] [td] [kg] [{a] [ao] stress Sounds not compared [m] [n] [i] [u] Comparisons and Conclusions [pb] CD [p] only word-initial [b] only word-medial [td] CD [t] only word-initial [d] only word-medial [kg] CD [k] only word-initial [g] only word-medial [{a] CAE 6 versus 8; 11 versus 13; 9 versus 10 [ao] CIE 12 versus 13 stress - contrastive 2 versus 3 Phonemes (You must include the following list in any answer) /p/ [b] word-medial [p] word-initial /t/ [d] word-medial [t] word-initial /k/ [g] word-medial [k] word-initial /m/ [m] /n/ [n] /i/ [i] /{/ [{] /a/ [a] /O/ [O] /u/ [u] /=/ [=] 1
2) Artificial p t i u b d e o m n l a R Suspicious pairs [p/b], [t/d], [l/r], [i/e], [u/o], (and perhaps [e/a], [o/a]) Non-suspicious sounds [m], [n] a) Stress is always on first syllable, therefore non-contrasting b) Vowels Minimal pairs (CIE) [a] vs [o] 2 [=dira] vs 3 [=diro] [a] vs [e] 10 [=nata] vs 11 [=nate] no minimal pairs (or CAE) for [i] vs [e] and [u] vs [o] Complementary distribution [i] vs [e] [i] always syllable 1 (stressed), [e] always syllable 2 (unstressed) [u] vs [o] [u] always syllable 1 (stressed), [o] always syllable 2 (unstressed) c) Consonants Minimal pairs (CIE) [p] vs [b] 7 [=libo] vs 8 [=lipo] CAE [t] vs [d] 1 [=bito] vs 6 [=lido] (and also symmetrical cf p/b) Complementary distribution [l] vs [R] [l] always initial (ie. syllable 1), [R] always medial (ie. syllable 2) d) Conclusions /i/ ---> [i] in (stressed) syllable 1 [e] in (unstressed) syllable 2 /u/ ---> [u] in (stressed) syllable 1 [o] in (unstressed) syllable 2 /a/ ---> [a] /p/ ---> [p] /b/ ---> [b] /t/ ---> [t] /d/ ---> [d] /l/ ---> [l] initial position (syllable 1) [R] medial or VCV position (syllable 2) /m/ ---> [m] /n/ ---> [n] 2
3) Artificial s S i u z Z n J N a Suspicious pairs [s/s], [z/z], [s/z], [S,Z], [n/j/n] Non-suspicious sounds [i], [a], [u] a) Stress is always on the first syllable, therefore non-contrasting b) Vowels [i], [a] and [u] are phonetically dissimilar c) Consonants No minimal pairs or CAE Complementary distribution [s] vs [z] [s] always initial (ie. stressed syllable 1), [z] medial (VCV) [S] vs [Z] [S] always initial (ie. stressed syllable 1), [Z] medial (VCV) [s] vs [S] 1 & 2 [=su...], 9 & 10 [=sa...], 5 & 6 [=Si...] ie. [s] preceding [u] and [a], [S] preceding [i] [z] vs [Z] 2 [... zu], 7 & 11 [... za], 4 & 8 [... Zi] ie. [z] preceding [u] and [a], [Z] preceding [i] [n] vs [J] vs [N] 3,4,7 & 9 [... na...], 1,3 & 5 [... Ji...], 6.8.10 & 11 [... Nu...] d) Some general rules [alveolar] ---> [post-alveolar] /_ [i] [-voice] ---> [+voice] / V_V [n] ---> [N] / _ [u] e) Conclusions /i/ ---> [i] /a/ ---> [a] /u/ ---> [u] /s/ ---> [s] / #_ {[u] [a]} ---> [z] / V_ {[u] [a]} ---> [S] / #_ [i] ---> [Z] / V_ [i] /n/ ---> [n] / _ [a] ---> [J] / _ [i] ---> [N] / _ [u] 3
4. Spanish (simplified) p t k i u b d g e o T s x B D G a m n J l º Suspicious Pairs [i/e], [e/a], [a/o], [o/u], [p/b], [t/d], [k/g], [T/D], [x/g], [b/b], [t/t], [d/d], [k/x], [g/g], [n/j], [l/º] Non-suspicious sounds [m], [s] a) Stress is not specified b) Vowels [u] vs [o] 22 [mudo] vs 21 [modo] CAE [i] vs [e] 20 [Tima] vs 29 [Tena] [a] vs [e] 16 [ºama] vs 17 [Jame] [a] vs [o] 2 [donada] vs 24 [TeJiDo] c) Consonants a) Voice contrasts [p] vs [b] 9 [baxa] vs 10 [paxa] [t] vs [d] 6 [tonada] vs 2 [donada] [k] vs [g] 4 [kama] vs 3 [gama] [T] vs [D] 20 [moto] vs 22 [modo] [x] vs [G] 10 [paxa] vs 11[paGa] b) Place contrasts [l] vs [º] 25 [solo] vs 26 [soºo] c) Manner (stop/fricative) contrasts [t] vs [T] 7 [tapa] vs 19 [Tapa] [k] vs [x] 30 [paka] vs 10 [paxa] CAE b) Place contrasts [n] vs [J] 23 [TeniTa] vs [TeJiDa] Complementary Distribution c) Manner (stop/fricative) contrasts [b] vs [B] [b] always initial, [B] always medial (VCV) [d] vs [D] [d] always initial, [D] always medial (VCV) [g] vs [G] [g] always initial, [G] always medial (VCV) d) Some general rules {[+voice] [stop]} ---> [fricative] / V_V 4
e) Conclusions /i/ ---> [i] /e/ ---> [e] /a/ ---> [a] /o/ ---> [o] /u/ ---> [u] /p/ ---> [p] /b/ ---> [b] initially ---> [B] medially (VCV) /t/ ---> [t] /d/ ---> [d] initially ---> [D] medially (VCV) /T/ ---> [T] /k/ ---> [k] /g/ ---> [g] initially ---> [G] medially (VCV) /x/ ---> [x] /s/ ---> [s] /m/ ---> [m] /n/ ---> [n] /J/ ---> [J] /l/ ---> [l] /º/ ---> [º] 5
5) Artificial p t k i u b d g e o s V z a m n N Suspicious pairs [p/b], [t/d], [k/g], [s/z], [n/n], [i/e], [u/o], [V,a], [e,v], [o,v] Non-suspicious sounds [m] a) Stress is not specified b) Vowels None Contrast in Analogous Environment (CAE) Nothing that is immediately obvious Complementary Distribution i) Adjacent phone context Preceding context Following Context /i/ p, b, t, k, n n, #, g, z, m /e/ k, t, z, g n, s, k, p, t, N /a/ s, d, m,p, k, s b, #, z, d /V/ n, p, d, m, k, s, t N, s, t, n /u/ n, k, d, z, t, p #, d, b, g /o/ s, b, m, p, k n, p, s, N The clearest trend evident in this table is that /i, a, u/ can be word final (#), whilst /e, V, o/ are never seen to be word final. A second (but much less readily seen) pattern is that when /i, a, u/ are followed by a stop or fricative, that stop or fricative is voiced, whilst when /e, V, o/ are followed by a stop or fricative, that stop or fricative is voiceless. Further, /N/ only follows /e, V, o/ whilst /n/ can be seen to follow vowels of either group. This data suggests two classes of vowel and most likely implies three vowel phonemes each with two allophones. At this point, however, the analysis is inconclusive. Whilst following /N/, stop/fricative-voicing or word boundary (#) contexts are consistently related to vowel groups, /n/ can follow vowels of either group and so following phone and word-boundary contexts will not consistently predict the choice of vowels. It may be that both vowel identity and consonant distribution are conditioned by some other aspect of word structure. 6
ii) Position in word There are four word structures:- CVCV, CVCCV, CVCVC, CVCCVC The following table shows the vowels in the position marked "*" C*CV CVC* C*CCV CVCC* C*CVC CVC*C C*CCVC CVCC*C i, a, u i, a, u e, V, o i, a, u i, a, u e, V, o e, V, o e, V, o When examining the distribution of the vowels, it is found that [i], [u], [a] occur in the word positions marked in the following list by "*", whilst the vowels [e], [V], [o] only occur in the word positions marked by "@":- C*C*, C@CC*, C*C@C, C@CC@C This is a clear example of the conditioning of vowel phone selection by word position. At present, however, the rules would be rather cumbersome as they would require the specification of four word contexts for each vowel class. The rules may, however, be simplified by reference to syllable structure. iii) Syllable structure ("$" has been used here to mark syllable boundaries) Since we cannot be 100% sure of the position of syllable boundaries word medially, it is only possible to unambiguously determine legal syllable onsets word initially and syllable rhymes word finally. Such patterns can be said to be the unambiguously attested syllable onset and rhyme forms for the language being examined and are a very reliable guide to possible word medial syllable structures. The only unambiguously attested syllable onset that can be seen word initially is #CV..., whilst two types of syllable rhyme can be seen word finally,...v# and...vc#. A combination of the one legal onset with the two legal rhymes gives two possible syllable structures:- $CV$ and $CVC$ (ie. an open and a closed syllable, both with an initial consonant). The only syllable boundary analysis of the above four word types that results in unambiguously attested syllable structures are as follows:- $CV$CV$, $CVC$CV$, $CV$CVC$, $CVC$CVC$ and the resulting vowel distributions (see above) are:- $C*$C*$, $C@C$C*$, $C*$C@C$, $C@C$C@C$ It can be readily seen from this diagram that [i, a, u] always occur in CV syllables whilst [e, V, o] always occur in CVC syllables. (This result further confirms the syllable analysis as the two solutions combine to form the simplest analysis of this language). iv) Conclusions It is clear from the above that there are two classes of vowels that are in complementary distribution based on syllable structure. Based on considerations of phonetic similarity, it is clear which pairs of sounds belong to the same phonemes (ie. [i,e], [a,v], [u,o] ). As there is no clear indication in each case as to which allophone is the most widely distributed (they both appear in the same number of contexts) then the selection of which symbol to use for the phoneme has to be based upon some other criterion. It is common in such cases as this to 7
select to most peripheral of each pair and use its symbol as the phoneme symbol. eg. [i] is more peripheral than [e] being higher and generally more fronted. Similarly [u] is more peripheral than [o] being higher and more retracted. [a] is not as clearly more peripheral than [V] but is chosen as the phoneme symbol because it belongs to the same class as the previously chosen "i" and "u" symbols. This gives:- /i/ = [i,e], /a/ = [a,v], /u/ = [u,o] v) [e, V], [o, V]? It was suggested above that these two pairs might be sufficiently phonetically similar to allow the possibility the they might form allophonic pairs. There are a number of reasons why this is not so for the present language. Firstly, they are each much more phonetically similar to the sounds they are paired with in section iv, above, and the evidence very strongly indicates their being in the allophonic relationships shown in that section. Is it possible for the conclusions indicated in iv, above, to be correct and at the same time for there to be a possibility that the pairs [e, V] and [o, V] also be allophones of the same phoneme? The answer is NO. This would place all of the vowels into the same phoneme or alternatively some phones would simultaneously be allophones of more than one phoneme. Neither of these two conclusions is possible (or at least they are extremely unlikely). In any case, since we now have determined the syllable structure of the language we can now identify two CAEs:- [e] vs [V] 8 [kenpa] vs 16 [moskvn] ($ken$ vs $kvn$) [o] vs [V] 15 [timvs] vs 16 [moskvn] ($mvs$ vs $mos$) These two CAEs confirm the conclusion that [e, V] and [o, V] do not constitute allophonic pairs, but it must be noted that these two CAEs are not essential to that conclusion. c) Consonants None Contrast in Analogous Environment (CAE) None Complementary Distribution i) Consonant voicing [p, b], [t, d], [k, g], [s, z] It was noted in section b)i) above that stop and fricative voicing seemed to be related in some way to the preceding vowel, with the stop or fricative always being voiced when the preceding vowel was from the [i, a, u] class. It was shown later that the selection of alternative vowel classes depended upon the syllable type, so it is likely that the selection of stop and fricative voicing is related in some way to syllable type. Since the vowels [i, a, u] are only found in CV syllables then the following stop or fricative must be the onset of the following syllable. When the other three vowels [e, V, o] precede the stop or fricative word-medially there is a CC cluster. An examination of the contexts in which the voiced ("*") and voiceless ("@") stops and fricatives occur in the words results in the following distribution:- $@V$*V$, $@V@$@V@, $@V$*V@$, $@V@$@V@$ It is clear that the voiced stops and fricatives can occur in the initial position of both CV and CVC syllables for certain word structures and 8
not for others. It appears that syllable structure is not the key to this structure. If the syllable boundaries are moved then it can more clearly be seen that the voiced stops and fricatives only occur in a VCV context. The general rule is:- [-voice] --> [+voice] / V_V ii) [n, N] It was noted in section b)i) above that [N] seemed to be conditioned in some way by the preceding vowel, only occurring when the preceding vowel was from the [e, V, o] class (ie. in a CVC syllable). It can also readily be seen that [N] only ever occurs in an initial CVC syllable. However, [n] can also occur in that position (eg. [kenpa]) so if complementary distribution is to be demonstrated then it must be conditioned by the adjacent sounds. As [N] can occur following any of the three possible preceding sounds [e, V, o] then it must be conditioned by the following sound. It can be readily seen that the following sound is always [k]. The general rule is therefore:- {[nasal] [consonantal] [alveolar]} --> [velar] / _ [velar] and the specific rule is [n] --> [N] / _ [k] or [n] --> [N] / _ [velar] d) Conclusions /i/ ---> [i] / $C_$ ---> [e] / $C_C$ /a/ ---> [a] / $C_$ ---> [V] / $C_C$ /u/ ---> [u] / $C_$ ---> [o] / $C_C$ /p/ ---> [b] / V_V ---> [p] elsewhere /t/ ---> [d] / V_V ---> [t] elsewhere /k/ ---> [g] / V_V ---> [k] elsewhere /s/ ---> [z] / V_V ---> [s] elsewhere /m/ ---> [m] /n/ ---> [N] / _ [velar] [n] elsewhere 9
6) Artificial b d g i u p t k e o P T x E O s l R Suspicious Pairs [p/b], [p/p], [t/d], [t/t], [k/g], [k/x], [l/r], [i/e], [e/e], [u/o], [o/o] Non-suspicious Sounds [s] a) Stress is not specified b) Vowels [i] vs [e] 1 [pige] vs 13 [pege] [e] vs [E] 20 [steri] vs 5 [steri] [u] vs [o] 14 [kupsi] vs 4 [kopsi] [o] vs [O] 17 [skopse] vs 6 [skopse] c) Consonants None Contrast in Analogous Environment (CAE) None Complementary Environment i) Syllable analysis There are two unambiguously attested syllable onsets #CV.. and #CCV... but there is only one unambiguously attested syllable rhyme...v#, and so the attested syllable types are CV and CCV. ii) Word types There are four word types derivable from the attested syllable types and all four (and no more) are found in the data:- CVCV, CVCCV, CCVCV, CCVCCV This results in the following consonant locations #CV, #CCV, VCV, VCCV The consonant phones that can be found in the various word positions are:- #CV #*V p, t, k, l #CCV #*CV s #C*V p, t, k VCV V*V b, d, g, R VCCV V*CV s, P, T, x VC*V p, t, k, s It is clear that [b, d, g, R] are conditioned by a V_V context. The phones that they are paired with allophonically are determined by considerations of phonetic similarity (shown in the list of "suspicious 10
pairs"). The fricatives [P, T, x] are also shown to be in complementary distribution with [p, t, k] but the rule:- [stop] ---> [fricative] / V_CV does not make much sense phonetically. What would cause a stop to become a fricative in this context? It is desirable to examine the detailed context more closely. iii) Adjacent phone context The three fricatives [P, T, x] are even further restricted to the following context:- V_sV The general rule can now be restated as:- [stop] ---> [fricative] / _ [fricative] which is a simple manner of articulation assimilation. iv) General rules [-voice] ---> [+voice] / V_V or even more generally, [-voice] ---> [+voice] / [+voice] _ [+voice] [stop] ---> [fricative] / _ [fricative] d) Conclusions /i/ ---> [i] /e/ ---> [e] /E/ ---> [E] /u/ ---> [u] /o/ ---> [o] /O/ ---> [O] /p/ ---> [b] / V_V ---> [P] / _ s ---> [p] elsewhere /t/ ---> [d] / V_V ---> [T] / _ s ---> [t] elsewhere /k/ ---> [g] / V_V ---> [x] / _ s ---> [k] elsewhere /l/ ---> [R] / V_V ---> [l] elsewhere /s/ ---> [s] 11
7) Korean (simplified) p t k i u s S e @ o m n N E l a R Suspicious Pairs [s/s], [n/n], [l/r], [i/e], [e/e], [E/a], [a/o], [u/o], [i/@], [e/@], [E/@], [a/@], [o/@], [u/@] Non-suspicious sounds [p], [t], [k], [m] a) Stress is not specified b) Vowels Minimal Pairs There is an almost complete paradigm of vowels in a single environment 1 [sun], 2 [son], 3 [san], 4 [s@n], 5 [sen], 7 [sen] This proves that the vowels [u, o, a, @, e, E] belong to different phonemes. There are neither minimal pairs nor obvious CAEs separating [i] from [e] or [@]. We already know that [e] and [@] belong to separate phonemes and so [i] cannot be in allophonic variation with both [e] and [@] as that would place them in the same phoneme. The three possible hypotheses are:- 1) /i/ = [i], /e/ = [e], /@/ = [@] 2) /i/ = [i,e], /@/ = [@] 3) /i/ = [i,@], /e/ = [e] Contrast in Analogous Environment (CAE) None are presently obvious (wait for results of consonant analysis) c) Consonants [n] vs [N] 23 [man] vs 24 [man] Contrast in Analogous Environment (CAE) None Complementary Distribution i. [l] vs [R] It is easily seen that [R] only occurs in V_V context whilst [l] occurs in all other contexts. ii. [s] vs [S] [S] can be clearly seen to only occur when the following vowel is an [i] whilst [s] occurs in all other contexts. d) [i] revisited [i] can be found in numerous consonantal contexts and so it is not conditioned by [S] but on the contrary, /s/ ---> [S] / _ i Since it is quite clear that [S] is an allophone of /s/ then it is reasonable to treat the pairs:- 6 [Sin] vs 5 [sen] and 6 [Sin] vs 4 [s@n] as contrasts in analogous environment (CAE) and so accept the hypothesis that /i/ = [i], /e/ = [e], /@/ = [@] (ie. 3 phonemes) 12
e) Conclusions /i/ ---> [i] /e/ ---> [e] /E/ ---> [E] /a/ ---> [a] /o/ ---> [o] /u/ ---> [u] /@/ ---> [@] /p/ ---> [p] /t/ ---> [t] /k/ ---> [k] /s/ ---> [S] / _ [i] ---> [s] elsewhere /m/ ---> [m] /n/ ---> [n] /N/ ---> [N] /l/ ---> [R] / V_V ---> [l] elsewhere 13
8) Artificial p t k i u b d g B D G a s z Z n N l R Suspicious Pairs [p/b], [b/b], [t/d], [d/d], [k/g], [g/g], [s/z], [z/z], [n/n], [l/r] Non-suspicious sounds [i], [a], [u] a) Stress is always on the first syllable, therefore non-contrasting. b) Vowels All three are phonetically dissimilar and therefore are separate phonemes. c) Consonants [b] vs [B] 16 [=labup] vs 1 [=labup] [d] vs [D] 28 [=sadal] vs 3 [=sadal] [g] vs [G] 18 [=dugik] vs 5 [=dugik] Contrast in Analogous Environment (CAE) None Complementary Distribution i. Word and syllable types. There are only two word types:- CVCVC and CVCCVC There is only one kind of unambiguously attested syllable onset #CV... and only one unambiguously attested syllable offset...vc# which when combined result in only one syllable type, CVC. The word type CVCVC is clearly not analysable as consisting of two CVC syllables and so either CV$CVC or CVC$VC would result. The maximum onset principle is invoked to select the CV$CVC form and to therefore hypothesise the existence of a word initial CV syllable (generally far more common in the world=s languages than VC). This is an example of a language in which a word onset and offset analysis does not result in a syllable structure hypothesis which accounts for all word types. ii. Consonant positions. There are five possible consonant positions with actual consonant distribution listed alongside:- #CV b, d, g s l n VCV b, d, g B, D, G VCCV b, d, g B, D, G Z n, N VCCV b, d, g z R VC# p, t, k s l n The above consonant position analysis results in clear complementary distributions (CD) for [b, d, g] vs [p, t, k], and for [s] vs [z] vs [Z]. 14
These results do not give a clear CD for [n] vs [N] an so further analysis is required here. iii. Adjacent phone context It is readily clear from the data that [N] always precedes [g] and that the familiar velar assimilation of an alveolar nasal has occurred. {[nasal] [alveolar]} ---> [velar] / _ [velar] d) Conclusions /i/ ---> [i] /a/ ---> [a] /u/ ---> [u] /b/ ---> [p] / _# ---> [b] elsewhere /B/ ---> [B] /d/ ---> [t] / _# ---> [d] elsewhere /D/ ---> [D] /g/ ---> [k] / _# ---> [g] elsewhere /G/ ---> [G] /s/ ---> [Z] / V_C (or [Z] / V_ [r], see notes below) ---> [z] / C_V ---> [s] elsewhere /l/ ---> [R] / C_V ---> [l] elsewhere /n/ ---> [N] / _ [N] ---> [n] elsewhere Notes: 1) The voiced stop allophones are less restricted in their distribution than are the voiceless allophones and so the phoneme has been given the voiced symbol. 2) The rules for the stops, /s/ and /l/ appear to be idiosyncratic language-specific positional rules. The following rules are nevertheless evident:- {[stop] [-voice]} ---> [+voice] / _ [+voice] This explains voicing both word initially and medially (where all consonants are voiced). {[fricative] [-voice]} ---> [+voice] / [+voice] _ [+voice] This prevents voicing both word initially and word finally. The phonetic conditioning that selects [Z] or [z] is not immediately obvious until it is realised that [Z] only ever occurs preceding [R] (unfortunately there is no [z] in the same word position but with some other consonantal context). It seems likely, however, that the following rule may more reasonably determine the selection of [Z]:- {[fricative] [alveolar]} ---> [post-alveolar] / _ [post-alveolar] Which translates to the more specific [s] ---> [Z] / V_ [R] This, of course, assumes that [R] is pronounced as a post-alveolar in this language. 15