CHAPTER 4 : Optimality Theory Analysis
4.2 Adaptations in Violation of Mandarin Phonotactics
4.2.2 Phonotactic Violation: Following Interlanguage Phonology
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4.2.2 Phonotactic Violation: Following Interlanguage Phonology
39.29% (66/168) of the tokens in the Zhuyin/English transcription belong to interlanguage phonology. The sequences are grouped by velar, labial, and alveolar consonants and some certain vowels, such as /jʌ, e/ in Chapter 3.
The sequence of the velar /k, kh, k’, h/ and the vowel /i/ follows Korean phonology, mapping to [ki], [xi]. However, when the rhyme is /jʌ/, 100% (19/19) of the outputs follow neither Korean phonology nor Mandarin phonology, inducing [khjou]. An analysis is shown in tableau (45).
(45) /sam.khjʌ/ → [saŋ.khjou] (桑ㄎㄧㄡ) ‘swallow’
Input: /khjʌ/ IDENT[strident]ONSET IDENT-V[back] MAX[+high]ONSET
☞1 a. khjou
b. khi *! *(!)
c. tɕjou *! *(!)
d. tɕi *! *(!) *(!)
e. khɤ *!
In tableau (45), [kh]-to-[tɕ] mapping is successfully eliminated by IDENT[strident]ONSET
(45c-d). Also, [khi] in (45b) is ruled out by IDENT-V[back] because of the disparity in the back vowel, and [khɤ] in (45e) is ruled out by MAX[+high]ONSET since the input /j/ is deleted in the output. In the end, (45a) is chosen to be the most optimal output without other variants.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
In the case of the labial /p, ph, p’, m/ and the vowel /jʌ/ sequence, 55.56% (20/36) of the tokens show a phonotactic violation which belongs to interlanguage grammar. An analysis is shown in (46) below.
(46) /pʰjʌŋ.seŋ/ → [phjoŋ.san] (ㄅㄩㄥ三) ‘for life’
Input: /pʰjʌŋ / IDENT-V[back] MAX[+high]ONSET MAX-SCODA
☞1 a. phjoŋ
b. phi *! *(!) *
c. phən *! *(!)
d. phu *! *
Previously proposed constraints successfully rule out wrong outputs. In tableau (46), the deletion of the [-high, +back] vowel /ʌ/ is banned by IDENT-V[back] as in (46b-c). Also, the deletion of the glide /j/ is ruled out by MAX[+high]ONSET (46d). Lastly, (46a) does not violate any constraints and becomes the most optimal output following the interlanguage phonology.
Lastly, a phonotactic violation appears when an alveolar sound and the vowel /e/
are combined (77.14%, 27/35), following the interlanguage phonology. In the corpus, the input /se/ is mostly mapped to the output [sei], and partially mapped to [ʂei] and [se]. The source vowel /e/ is retained in all of the variations. In Mandarin phonotactics, the alveolar /s/ is not allowed before the vowel /e/. Therefore, MAX[e] is used above the cut-off line to preserve the vowel /e/, and to eliminate unfaithful vowels. *C[e] and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
IDENT[anterior]ONSET below the cut-off line ensure the variation. A syllable /se/ is given as an example in tableau (47).
In tableau (47), the deletion of the vowel /e/ are all ruled out in (47d-f). Also, *C[e] and IDENT[anterior]ONSET, previously posited above the cut-off line, are re-posited to be below the cut-off line to preserve the [se] variation as in (47c). Lastly, IDENT[anterior]ONSET is dominated by *C[e]. Thus, [ʂei] (47b) became the second most frequent variants when [se] (47c) is third most frequent variant.
The constraints related to the ranking of adaptations in violation of Mandarin phonotactics are illustrated (48):
(47) /se.ro/ → [sei.lou] (SAY 樓) ‘newly’
Input: se MAX[e] *C[e] IDENT[anterior]ONSET
☞1 a. sei
☞2 b. ʂei *
☞3 c. se *
d. su *!
e. sa *!
f. sɤ *!
‧
(48) Adaptation Violating Mandarin Phonotactics
IDENT[strident]ONSET, *[y], IDENT-V[back], MAX[+high]ONSET, IDENT-V[-low], IDENT-V[-high], MAX[e]
Cut-off MAX-SCODA,*C[o], *C[e]
|
IDENT[anterior]ONSET
Compared to the constraint ranking of the adaptations in conformity to Mandarin phonotactics, there are two changes in (48). First, the constraints IDENT[strident]ONSET,
*[y] and MAX[+high]ONSET are added in the ranking above the cut-off line.
IDENT[strident]ONSET is proposed to preserve the velar sound and to ban palatalization.
*[y] is proposed to eliminate the [+high, -back] vowel /y/, which is not allowed in the Korean grammar. MAX[+high]ONSET is proposed to prevent the deletion of /j/.
Second, *C[o], *C[e], and IDENT[anterior]ONSET are dominated below the cut-off line in the ranking. The majority of the occurrences of /o, e/ are diphthongized and banned to solely occur after a consonant in the adaptation in conformity to Mandarin phonotactics.
However, the inputs /no/ and /se/ are faithfully mapped to [no] and [se] in violation of Mandarin phonotactics. Thus, *C[o], *C[e] cannot function to eliminate the ill-formedness of Mandarin phonotactics. IDENT[anterior]ONSET rules out [s]-to-[ʂ] mapping.
4.3 Summary
In the previous section, the OT analysis of KM Kong-er adaptation was done in two parts: an analysis of the unmarked phonology which conforms to Mandarin
‧
following Korean phonology and that following interlanguage phonology. Under the frame of the ROE, a constraint above the cut-off line eliminates fatal candidates, and a constraint below the cut-off line preserves variations.In unmarked KM Kong-er adaptation, illicit consonants and vowels in Mandarin are ruled out by the constraints above the cut-off line, such as *C[e], *C[o], *[ʌ], *[ɯ],
*TENSE(C), CODA CONDITION. Other constraints above the cut-off line such as [+high, -back]CX, [IDENT[tense]&*[+aspirated]]ONSET, MAX-SONSET, MAX[e], IDENT-V[back], IDENT[anterior]ONSET are proposed to delete undesirable mapping, following the Mandarin phonotactics again. Candidates who survive those constraints become the output as either optimal or possible variations. Since the constraints do not conflict with each other, no hierarchy is observed above the cut-off line.
The constraints below the cut-off line decide the candidate ranking. The more a candidate violates high-ranked constraints, the more low-ranked it is in the candidate hierarchy. In the unmarked phonology, the conflict between the constraints above the cut-off line does not decide the rank of the constraint, but the frequency of the variation below the cut-off line does decide the hierarchy between the constraints.
In coda deletion, the dominant relation IDENT[nasal]CODA >> MAX-SCODA >>
IDENT[anterior]CODA successfully generate the candidate ranking for each coda type.
IDENT[nasal]CODA >> MAX-SCODA results in ‘coda deletion >> coda replacement’ for /p, t, k, l/, and ‘coda replacement >> coda deletion’ for /m/. IDENT[anterior]CODA regulates unfaithful output from /n, ŋ/ so that a ‘retainment >> replacement >> coda deletion’
hierarchy is generated.
‧
rounded vowel violates IDENT-V[round] so that ‘vowel lowering >> vowel rounding’ is generated.For marked phonology which is in violation of Mandarin phonotactics, three constraints above the cut-off line are proposed: namely IDENT[strident]ONSET, MAX[+high]ONSET and *[y]. IDENT[strident]ONSET functions to rule out [kj]-to-[tɕj]
mapping, which is highly ranked in Mandarin phonotactics. Since the velar stop /k/ and the alveolar affricate /tɕ/ are different in the [strident] feature, IDENT[strident]ONSET can delete the mapping. Also, IDENT-V[-low] and IDENT-V[-high]are used to preserve /o, ʌ/
and to prevent [nʌ]-to-[na] or [nʌ]-to-[nu] mapping. *[y] is activated in the adaptation following Korean phonology since Mandarin vowel /y/ is illicit in Korean phonotactics.
Lastly, *C[e] and *C[o] are dominated below the cut-off line since the marked phonology allows the vowels /o, e/ to solely occur and allows [+high, -back] vowels after a [+high, +back] consonant. The Hasse diagram of the two types of adaptations is presented in (52).