Chapter 7 Music-to-Language Mapping: Tone and Term
7.7 L I -to-L O Mapping: Type 2 Operation
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
116
In tableau (195), the children fail to associate the linguistic input with the actual term at the lexical level such both candidates violate ASSOC-TERM. TONOTACTICS favors (195a) over (195b), so tai11 yang55 emerges as the initial output, which then serves as the postlexical input and activates ASSOC-TERM again, as in tableau (196).
(196) LI-to-LO mapping: final output
Initial output/postlexical input: tai11 yang55 Final output: tai53 yang35
At the postlexical level, ASSOC-TERM successfully selects (196a) as the optimal output, where the NP, tai53 yang35, emerges. The two-step operation of ASSOC-TERM
is developed from Kiparsky’s (2010, 2015) Stratal OT, which considers that each linguistic level has an individual grammar. In this spirit, I observe that ASSOC-TERM
operate separately at lexical and postlexical levels.
7.7 L
I-to-L
OMapping: Type 2 Operation
In the type 2 operation, the linguistic input is successfully associated to a familiar term at the lexical level, and thus an actual word is directly produced.
Tableau (197) demonstrates the production grammar from the linguistic input to the linguistic final output.
ASSOC-TERM TONOTACTICS ID-T
a. tai53 yang35 太陽 ‘sun’
**
b. tai11 yang55 *!
c. tai33 yang55 *(!) * (!) *
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
117
(197) LI-to-LO mapping: final output Linguistic input: tai11 yang55 Final input: tai53 yang35
(197a) is the only term that is associated and thus is selected as the optimal output, regardless that fact that it violates ID-T. (197b-c) are unassociated terms, and thus are ruled out by ASSOC-TERM.
7.8 Summary
This chapter has examined closely how the children map the singing outputs to their linguistic outputs. I have proposed a model, as in (175) that distinguishes the perception grammar from the production grammar. In the perception grammar, I have posited principles to account for the correspondence of musical pitch, register, and distance in the MO-to-LI mapping, which is governed by a set of constraints, as summarized by the Hasse diagram in (198).
(198) MO-to-LI perception grammar: pitch-tone correspondence ID-REG ID-CTR NoDIST=1 MAX-DIST
NoDIST=4
ASSOC-TERM TONOTACTICS ID-T
a. tai53 yang35 太陽 ‘sun’
**
b. tai11 yang55 *!
c. tai33 yang55 *(!) * (!) *
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
118
In the production grammar, two types of operations are observed. In the type 1 operation, term association fails at the lexical level and an initial output is produced.
The initial output then serves as the postlexical input, which is successfully associated and yields the final output. In the type 2 operation, the linguistic input is successfully associated to a familiar term at the lexical level, and thus an actual word is directly produced. A set of constraints are posited to govern the production grammar, which are ranked as in (199).
(199) LI-to-LO production grammar: term association ASSOC-TERM TONOTACTICS
ID-T
To summarize, the MO-to-LIperception grammar values the faithfulness relation between the musical pitch and the linguistic tone. The LI-to-LO production grammar requires term association at the lexical and postlexical levels.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
119
Chapter 8
Music-to-Language Mapping: Tone and Markedness
8.1 Introduction
This chapter continues to examine how children map the musical pitches to linguistic tones in the singing of Mandarin songs, with a focus on the failure of term association. Precisely, I raise the question that when the children fail to associate the tonal strings to the words or phrases in their mental lexicon, how the final linguistic output is produced. In particular, I look at the role of markedness constraints in the tonal production.
8.2 Data Design
The data collected from the preschool children also contain words or phrases that the children are not familiar to, such as technical terms, idioms, and other less frequent words. Some examples are given in (200).
(200) Examples of the unfamiliar disyllabic words/phrases Unfamiliar words/phrases
shu35 jia53 ‘ransom’ 贖價 ju53 fa21 ‘syntax’句法 li35 li35 ‘flourish’ 離離 you53 you21 ‘and have’ 又有
The syllables are sung by a female native Mandarin speaker with electric keyboard.
Each syllable is set to one musical pitch, whose duration is one beat. The singer is
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
120
well-trained in music and singing so that the musical pitches are produced accurately.
The children are asked to tell the possible lyrics that are sung by the singer. The unfamiliar words and phrases are sung with different musical pitch contours. The following are some examples.
(201) Examples of the singing words/phrases Music contour Singing output
a. Falling
yin6 shi
#4 ‘diet’ 飲食
b. Rising ji jiao#
1 ‘haggle’ 計較
c. Level
li5 li 5 ‘flourish’ 離離
The musical output in (201a) displays a falling contour, while that in (201b) shows rising contour. The pitch contour in (201c) is level.
8.3 Musical Pitch to Linguistic Tone Mapping
This section discusses the mapping from the musical pitch to the linguistic tone in unfamiliar terms, without association of the perceived tonal strings to their mental lexicon. In this case, an initial output is produced. The fact that the initial tonal output is totally faithful to the musical pitch output supports the argument that the initial tonal output is the same as the perceived linguistic tonal input. The emergence of the final linguistic output is then keyed to the interaction of faithfulness and markedness constraints. I have proposed the model in (202) to account for the unassociated term mapping.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
121
(202) Musical pitch to linguistic tone mapping: unassociated terms Musical pitch output
Linguistic tonal input (lexical) Term assoc. fails
Initial output/postlexical input (postlexical)
No mark. violation Term assoc. Mark. violation fails.
Faithful output Unfaithful Final output Type 3a Type 3b
In the failure of term association at the lexical level, an initial output is produced and serves as the postlexical input. The model in (202) indicates that when term association also fails at the postlexical level, two types of production operations are in order, here referred to as type 3a and type 3b. In the type 3a operation, the postlexical input is faithfully mapped as the final output, with no crucial violation of any markedness constraint. In the type 3b operation, the postlexical input crucially violates some markedness constraints and is subject to segmental adjustments, such that the emerged final output is not faithful to the postlexical input. The table in (203) presents some statistics.
‧
(203) Statistics of the unassociated terms Tokens Percentage
Type 3a 105 73.9%
Type 3b 37 26.1%
Total 142 100%
There are 142 tokens of the unassociated terms in the database. 105 (73.9%) of them are produced faithfully to the initial output (postlexical input), which pertains to the type 3a operation. 37 (26.1%) of them are unfaithful to the initial output (postlexical input), which pertains to the type 3b operation.
8.3.1 Type 3a Operation
Based on the singing, some of the children may produce an initial output which faithfully yields the final output. The following are some examples.
(204) Examples of the type 3a operation
Musical output Initial tonal output Final tonal output
a. si6 zhou6‧
In (204a), the musical output, si6 zhou6, which is identical to the initial tonal output
si55 zhou55, is then faithfully mapped as the final output. It should be noted that in
(204f), yu1 san5, ‘umbrella’, is supposed to be a word that children recognized.However, since they fail to associate yu1 san5 to their mental lexicon, it is regarded as an unfamiliar word.
Consider the pitch distance in (205).
(205) Musical pitch distance and tonal distance
Musical output Linguistic tonal input
a. si6 zhou6 ‘surroundings’ 四周 Dist: 0
In (205a-b), the musical pitch contour is level. Mandarin has a high tone, 55, and a low tone, 21, which is like 11. As discussed in Chapter 7, the music register may affect the mapping of level tones. In (205a), the music register of si6 zhou6 is [+upper], which corresponds to the [+upper] tone register of si55 zhou55. Since the children fail to associate the linguistic input, si55 zhou55, to their mental lexicon at
‧
the lexical and the postlexical level, si55 zhou55 is produced as the initial output and eventually as the final output. The mapping in (205b) is operated in the same way.
Given the pitch distance mapping principle in (172) of Chapter 7, the musical pitch distance of yin6 shi#
4 in (205c), is 1.5, and the corresponding tonal distance is 2,
which is preserved in the linguistic tonal input, yin55 shi33. The disyllabic tonal contour of the tonal input, yin55 shi33, is falling, which is identical to the pitch contour of the musical output. Since the children fail to associate the linguistic input,yin55 shi33, to their mental lexicon at the lexical and postlexical level, yin55 shi33 is
produced as the initial output, which is then faithfully mapped as the final output. The pitch distance in (205d, e) is preserved in the same way.8.3.2 Type 3b Operation
When the children fail to associate the initial tonal outputs to their mental lexicon, the initial tonal outputs may incur crucial violations of some markedness constraints and be subject to surface well-formedness. Consequently, the final outputs may not be faithful to the tonal inputs. The unassociated terms with the unfaithful outputs are shown in table (206).
(206) Examples of the type 3b operation
Musical output Linguistic tonal input Final tonal output
a. you1 you1 ‘and have’ 又有‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
125
c. mei#
4 mao
#4 ‘eyebrows’ 眉毛
Dist: 0Reg: [+upper]
mei55 mao55
Dist: 0Reg: [+upper]
mei35 mao55
In (206a), the music register of you1 you1 is [-upper], which corresponds to the [-upper] tonal register of the linguistic input, you11 you11. However, it violates the OCP (Leben 1973, Goldsmith 1976), which disallows adjacent low tones in Mandarin.
Consequently, the final output, you35 you21 is not faithful to the tonal input. (206b) is operated in the same way. In (206c), mei#
4 mao
#4 ‘eyebrows’, is supposed to be a
word that children are familiar with. Nevertheless, since mei#4 mao
#4 fails to be
associated to the children’s mental lexicon, it is regarded as an unfamiliar word. In (206c), the music register of mei#4 mao
#4 is [+upper], which corresponds to the
[+upper] tone register of mei55 mao55. However, the syllable-tone combination ofmei55 does not exist in Mandarin, and is thus excluded. Consequently, the final output, mei35 mao55 is not faithful to the tonal input.
8.4 M
O-to-L
IMapping: Pitch Perception
The perception grammar that maps the musical output (MO) to the linguistic input (LI) is the same as that discussed in section 7.5. A falling musical pitch like 6-#
4 may
be mapped as the tonal strings 55-33 or 33-11; when the musical pitch distance is 1.5, its corresponding tonal distance is 2, and the pitch distance is preserved. A rising musical pitch like
-#1 may be mapped as the tonal strings 11-33 or 33-55; when the
musical pitch distance is 1, its corresponding tone distance is 2, and this pitch distance is preserved. A level musical pitch like 4-4 may be mapped as the tonal strings 55-55 or 33-33; when the musical pitch distance is 0, its corresponding tone distance is also‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
126
0, and this pitch distance is preserved. The constraint ranking that governs the MO-to-LI mapping is proposed in (189) of Chapter 7, and reproduced in (207).
(207) MO-to-LI constraint ranking
ID-REG,ID-CTR,No-DIST=1,MAX-DIST >>No-DIST=4
ID-CTR requires faithful correspondence of pitch contour, and ID-REG requires faithful correspondence of register. MAX-DIST requires preservation of pitch distance.
No-DIST=1 and No-DIST=4 lay down restrictions on the shape of the disyllabic contour. ID-REG,ID-CTR,No-DIST=1andMAX-DIST must be ranked above No-DIST=4 to preserve the contour shape and the register, but to avoid mappings like [#
6-
#1 →
55-44], [1-1 → 55-55],etc.8.5 L
I-to-L
OMapping: Type 3a Operation
As proposed in (202), the schema of musical pitch to linguistic tone suggests two types of production operation. This section discusses type 3a operation, in which term association fails at the lexical level and postlexical level while the final output is faithful to the linguistic tonal input. Two constraints are relevant, ASSOC-TERM and ID-T. The partial constraint ranking is posited in (208).
(208) LI-to-LO partial constraint ranking:
ASSOC-TERM >>ID-T
Tableau (209) shows how these constraint ranking works.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
127
(209) LI-to-LO production grammar: faithful output Linguistic input: yin55 shi33 ‘diet’ 飲食 Final output: yin55 shi35
ASSOC-TERM ID-T
a. yin55 shi35 * *
b. yin33 shi55 * *!*
c. yin35 shi55 * *!*
In tableau (209), the terms are unassociated at either lexical or postlexical level, and the constraint ASSOC-TERM is violated by each of the candidates. The candidate in (209a) wins over the others by one less violation of ID-T.
8.6 L
I-to-L
OMapping: Type 3b Operation
In the type 3b operation, the final output is unfaithful to the linguistic input. The markedness constraints rule out the faithful output and select the output that conforms to surface well-formedness. Three constraints are relevant for this mapping, as in (210-212).
(210) OCP-LOW:
Assign one violation mark for every pairs of low tones.
(211) TONOTACTICS:
Assign one violation mark for every tone such as 33 and 44 that does not exist in Mandarin.
(212)ID-T-R:
Assign one violation mark for every rightmost tone that is not identical to its correspondent in the input.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
128
OCP-LOW prohibits consecutive low tones, and forces one of the tones to change.
ID-T-R preserves the right tone and allows the left tone to change. The constraint TONOTACTICS bans any tone that does not exist in Mandarin, and must be ranked higher than ID-T. The enriched constraint ranking is posited in (213).
(213) LI-to-LO constraint ranking (enriched):
ASSOC-TERM,TONOTACTICS,OCP-LOW >>ID-T, ID-T-R;
Tableau (214) shows how this constraint works.
(214) LI-to-LO production grammar: unfaithful output
Initial output/Postlexical input: li11 li11 ‘flourish’ 離離 Final output: li35 li11
ASSOC-TERM TONOTACTICS OCP-LOW ID-T ID-T-R
a. li35 li11 * *
b. li11 li11 * *!
c. li11 li35 * * *!
d. li33 li11 * *! *
The markedness constraints rule out (214b, d) respectively, while (214c) is ruled out by the faithfulness constraint, ID-T-R. As a consequence, (214a) emerges as the optimal output, where li35 li21, is different from the postlexical input, li11 li11.
Another example lies in the phonotactics. For instance, the musical output, mei5
mao5, perceived as mei55 mao55 in the linguistic input, which, in the failure of
lexical term association, is faithfully mapped as the initial output, which in turns serves as the postlexical input. In the failure of postlexical term association, the postlexical input, mei55 mao55, is eventually mei35 mao55, which is an unfaithful output. This unfaithful mapping is attributed to the fact that the syllable-tone‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
129
combination of mei55 is illegal in the Mandarin output. The constraint PHONOTACTICS, as in (215), is thus needed for the following discussions.
(215) PHONOTACTICS:
Assign one violation mark for every illegal syllable-tone combination in the output.
PHONOTACTICS should be undominated, and the partial constraint ranking is expended as (216).
(216) LI-to-LO constraint ranking (enriched):
ASSOC-TERM,TONOTACTICS, PHONOTACTICS,OCP-LOW >>ID-T, ID-T-R
Tableau (217) demonstrates the constraint interaction.
(217) LI-to-LO production grammar:
Initial output/Postlexical input: mei55 mao55 ‘eyebrows’ 眉毛 Final output: mei35 mao55
ASSOC- TERM TONOTACTICS PHONOTACTICS ID-T
a. mei35 mao55 * *
b. mei55 mao55 * *!
Since the children fail to associate the term to their mental lexicon, the constraint ASSOC-term is violated by both candidates. The syllable-tone combination of mei55 in (217b) violates PHONOTACTICS. (217a) thus surfaces and an unfaithful output is selected.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
130
8.7 Summary
This chapter has addressed a third type of operation in the music-to-language mapping. The perception grammar that maps MO to LI is the same as that in type 1 and type 2, as discussed in Chapter 7. This third type of operation further consists of two subtypes, type 3a and type 3b, in both of which term association fails at the lexical and postlexical levels. The mapping from the postlexical input to the final output is keyed to the interactions of faithfulness and markedness constraints. I have posited a set of constraints, which are ranked as in (218).
(218) LI-to-LO production grammar: term association
ASSOC-TERM TONOTACTICS OCP-LOW PHONOTACTICS
ID-T ID-T-R
The MO-to-LI perception grammar values the faithfulness relation between the musical pitch and the linguistic tone. The LI-to-LO production grammar requires term association at the lexical and postlexical levels. If term association fails, the final linguistic output will be faithful to the postlexical input; however, when surface markedness constraints are violated, an unfaithful output will emerge.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
131
Chapter 9
Conclusion and Further Issue
9.1 Conclusion
This study probes into the connection between language and music through the perception and production grammar in English and Mandarin songs. The interaction between language and music is revealed in three aspects, namely, segment, rhythm, and tone.
Chapters 3 and 4 compare the linguistic mapping and the language-to-music mapping in onset and coda clusters through Mandarin accented English. The linguistic mapping in onset clusters and coda clusters involves the interaction between segmental faithfulness and markedness, as well as prosodic alignment, which yields two output variants: the monosyllabic LO1and the disyllabic LO2. At this stage, the SSP is not crucial in either constraint ranking. Consider the Hasse diagrams reproduced in (219).
(219)LI-to-LO mapping in onset and coda clusters (a) LI-to-LO1 mapping: ()
ALIGN-E (LEX, ) DEP-V MAX-C
SSP
*CC
(b) LI-to-LO2 mapping: ()
ALIGN-E(LEX, ) *CC MAX-C
SSP
DEP-V
(220a) exhibits the LO-to-MI perception grammar in onset clusters, while (220b) presents that in coda clusters.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
132
(220)
(a) LO-to-MI mapping in onset clusters NOSTRAY MAX-C
DEP-V
ALIGN-R(♩, )
NOSHARE-B NOSHARE-σ
(b) LO-to-MI mapping in coda clusters NOSTRAY MAX-C NOSHARE-σ & *CCCODA
DEP-V
SSP
ALIGN-R(♩, )
NOSHARE-B NOSHARE-σ
In (220b), the conjoined constraint, NOSHARE-σ & *CCCODA,is dominant and SSP is observed to affect the insertion of vowels in singing.
The MI-to-MO production grammar in onset and coda clusters requires faithfulness of syllable-beat association and the removal of prosodic structure in the musical output as reproduced in the Hasse diagram in (221).
(221) MI-to-MO production grammar
ID-ASSOC *PROSST
MAX-PROSST
Chapters 5 and 6 examine the language-to-music mapping with the focus on the composing of Mandarin children’s song. In Chapter 5, I have examined how the foot and IP structures affect the assignment of musical beats. The foot is formed based on the syntactic structures, in which immediate constituent (IC) and branching direction
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
133
are keyed to deciding the output. The constraint ranking for foot formation is reproduced in the Hasse diagram in (222).
(222) LI-to-LO production grammar
NOSTRADDLE-FT PARSE-ALIGN-E(IC,FT)*MONO-FT
FTBIN
The LO-to-MI perception grammar of the syllable-to-beat assignment of foot is exhibited in Hasse diagram reproduced in (223).
(223) LO-to-MI perception grammar
NOSTRAY NOSTRADDLE-B(FT) NOSHARE-(B5)
NOSPLIT-IC(B) NOSHARE-B NOSHARE-σ
As in (223) and in (220) proposed previously, both NOSHARE-B and NOSHARE-σ are ranked at the bottom, as a syllable is often linked to multiple musical beats to prevent stray elements and vice versa.
Chapter 6 continues to examine the mapping of rhythmic structures focalizing on the intonational phrase (IP). The IPs in the lyric output are defined by the punctuation marks, which is captured by ALIGN-R(PM, IP). The mapping of the IP to the musical input is crucial to the measure-to-IP alignment and the IP-final musical lengthening.
The LO-to-MI perception grammar is governed by two sets of constraints as reproduced
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
134
in the Hasse diagrams in (224) and (225).
(224) LO-to-MI perception grammar: measure-to-IP alignment ALIGN-R(SONG,IP) ALIGN-L(MN, IP)
ALIGN-R(M2N, IP)
ALIGN-L(IP, MN) ALIGN-R(IP,M2N)
(225) LO-to-MI perception grammar: musical beat assignment NOSTRAY *n-1 n]IP NOSHARE-(B5) NOSPLIT-IC(B)
NOSPLIT-IC(B)
The mapping between the musical input (MI) and the musical output (MO) in Chapter 5-6 is governed by three constraints, ID-ASSOC,*PROSST and MAX-PROSST. As previously posited in Chapter 3-4, the musical output preserves the musical beat association but removes the structure of the prosodic structures.
In Chapter 7-8, the music-to-language mapping is examined through children’s perception of the singing of Mandarin songs. Singing words or phrases are mapped to the linguistic input forms with similar tonal values through the perception grammar.
The perception grammar is accounted by the correspondence of musical pitch, register, and distance in the MO-to-LI mapping, which is governed by a set of constraints reproduced in the Hasse diagram in (226).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
135
(226) MO-to-LI perception grammar: pitch-tone correspondence ID-REG ID-CTR NoDIST=1 MAX-DIST
NoDIST=4
The children associate what they have heard to their mental lexicon through the LI-to-LO production grammar. Four types of operations are observed. In the type 1 and type 2 operations, term association is achieved at some points and the final outputs are actual words or phrases as governed by the constraint ranking reproduced in the Hasse diagram in (227).
(227) LI-to-LO production grammar: term association ASSOC-TERM TONOTACTICS
ID-T
The top-ranked ASSOC-TERM selects the associated term whose tone may be different from the linguistic input.
In the type 3a and 3b operations, term association fails at both lexical and postlexical level. If term association fails, the final linguistic output will be faithful to the postlexical input; however, when surface markedness constraints are violated, an unfaithful output will emerge. The relevant Hasse diagram is reproduced in (228).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
136
(228) LI-to-LO production grammar: term association
ASSOC-TERM TONOTACTICS OCP-LOW PHONOTACTICS
ID-T ID-T-R
Taken together, Chapters 3 to 8 demonstrate segmental changes, rhythmic alignment, beat assignment, and linguistic tonal changes in the mapping between language and music. From the perspective of segment and rhythm, the language-to-music mapping shows that the perception grammar plays an important role. Tonal changes in the music-to-language mapping also support that there is an independent perception grammar.
9.2 Further issue
In Chapters 7 and 8, I have examined how disyllabic musical pitches are transformed into linguistic tones. Each syllable is associated with one musical beat with equal musical prominence. This study has also observed the musical pitch to linguistic
In Chapters 7 and 8, I have examined how disyllabic musical pitches are transformed into linguistic tones. Each syllable is associated with one musical beat with equal musical prominence. This study has also observed the musical pitch to linguistic