Language-to-music Mapping

(85b), the singing output is still [æktə], where each syllable is linked to one musical beat respectively. The structure of the prosodic word is removed in the singing output.

(86) Statistics of the linguistic output2 (σσ) to singing mapping

Linguistic output2 Singing output (q) Singing output (q q)

(VCCV)

The table in (86) shows that there are totally 21 linguistic output2 in the database.

When they are associated with one musical beat, 17, or 81%, of them are sung as two syllables. When they are associated with two beats, 18, or 85.7%, of them are still sung as two syllables.

4.3 Language-to-music Mapping

The model I proposed in (46) of Chapter 3 is applicable here. The monosyllabic input yields two linguistic outputs, which are parsed into prosodic words. Each of the outputs is assigned with one and two musical beats. Linguistic output1 is read with one syllable, where no vowel is inserted. Linguistic output2 is read with two syllables, where a vowel is inserted to resolve the coda cluster.

The mapping from the linguistic output to the musical input demonstrates a need for the perception grammar. The linguistic outputs are perceived as the musical inputs, and assigned with one and two musical beats. The prosodic word structure,

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

52

which may affect beat assignment, is shown in the musical input. When linguistic output1 is associated with one musical beat, the coda cluster is preserved and the syllable remains monosyllabic. When linguistic output1 is assigned with two musical beats, a vowel is inserted and the syllable becomes disyllabic. On the other hand, linguistic output2 remains disyllabic, regardless of the number of the musical beat assigned. Finally, the musical inputs are mapped to the musical outputs, where the prosodic word structure is removed.

4.4 L

I

-to-L

O

Mapping: Production Grammar

This section analyzes the linguistic input (LI) to the linguistic output (LO) mapping under the framework of Optimality Theory (Prince and Smolensky 1993/2004). There are two linguistic outputs. One is with vowel insertion and the other is without vowel insertion. Take /ækt/ ‘act,’ for example. The output of /ækt/ can be either (ækt) or (æktə). Since a word like (æ kt) violates the SSP, the constraint in (87) is thus relevant.

(87) SSP:

Assign one violation mark for every syllable whose coda does not rise in sonority toward the nucleus, or whose coda does not decrease in sonority from the nucleus.

This constraint is ranked relatively low so that (æ kt) can surface. Adding this constraint to the constraint rankings proposed in Chapter 3, the two output variants can be evaluated through the tableaux in (88-89).

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

53

(88) LI-to-LO mapping: output1, (ækt) /ækt/ ‘act’ ALIGN-E

(LEX, )

MAX-C DEP-V SSP *CC

 a. (ækt) * *

b. (æktə) *!

c. (æk) *!

d. ækt *! * *

In tableau (88), MAX-C and DEP-V are ranked higher than *CC so that (ækt) will not be eliminated. (88b) inserts a vowel and incurs a fatal violation of DEP-V,so it is ruled out. (88c) deletes a consonant, and thus is ruled out by MAX-C. The candidate in (88d) is not a prosodic word, fatally violating ALIGN-E (LEX, ). (88a) is thus selected as the optimal output, in sacrifice of the SSP.

(89) LI-to-LO mapping: output2, (æktə) /ækt/ ‘act’ ALIGN-E

(LEX, )

MAX-C *CC SSP DEP-V

a. (ækt) *! *

 b. (æktə) *

c. (æk) *!

d. ækt *! * *

As in (89), *CCis ranked higher than DEP-V. (89a) preserves the coda cluster [kt], which is ruled out by *CC.(89c) deletes [t], and thus is ruled out by MAX-C. (89d) is not a prosodic word, and thus is ruled out by ALIGN-E (LEX, ). (89b) thus emerges.

The constraint rankings for the linguistic input-to-output production grammar are summarized in (90).

‧

(90) Linguistic input to linguistic output production grammar

a. Output1: [ækt]→ (ækt) ALIGN-E(LEX, ), MAX-C, DEP-V >> SSP >> *CC

b. Output2: [ækt]→ (æktə) ALIGN-E(LEX, ), MAX-C, *CC>>SSP >> DEP-V

In either ranking above, the SSP is not crucial.

4.5 L

O

-to-M

I

Mapping: Perception Grammar

4.5.1 Lo

1

-to-M

I

Mapping

This section examines the monosyllabic linguistic output1 (LO1) to the musical input (MI) mapping. LO1 is perceived as the musical segmental input and is respectively assigned with one and two musical beats. I first discuss the case that LO1

is assigned with one musical beat. Given the constraints posited in Chapter 3 with the addition of the SSP, the coda cluster mapping is illustrated in tableau (91), where the bottom-ranked constraints are omitted due to the limited space.

(91) LO1-to-MI mapping: coda cluster (♩)

‧

violation of NOSTRAY. (91c) deletes a consonant to satisfy the SSP, but violates the higher-ranked MAX-C. (91d) is ruled out by DEP-V with an insertion of [ə].

ALIGN-R(♩, ) is inactive here because there is only one musical beat and one syllable.

Eventually, (91a) emerges.

However, this constraint ranking selects the wrong output in tableau (92).

(92) LO1-to-MI mapping: coda cluster (♩♩) indicated by the parenthesized right-head () symbol. (92c) is incorrectly selected as the optimal output, as indicated by the black right-head  symbol. In order to select the correct output, I posit the conjoined constraint in (93).

‧

Assign one violation mark for every syllable that contains complex coda and is shared by two musical beats, ♩♩.

The unconjoined NOSHARE-σ and *CCCODA are ranked at the bottom. When NOSHARE-σ and *CCCODA are conjoined, the constraint ranking is substantially promoted. The purpose of the local conjunction is to rules out the worst of the worst output, which is called the WOW effect (Green 1993; Smolemsky 1993, 1995;

Moreton & Smolensky 2002, among others). The conjoined NOSHARE-σ & *CCCODA is violated only when both NOSHARE-σ and *CCCODA are violated. This constraint

‧

enabled to surface. It should be noted that when a monosyllabic syllable with an onset cluster like (blu) is assigned with two musical beats, its surface form is still (blu); since it does not have a complex coda, the conjoined constraint NOSHARE-σ &

*CCCODA would be irrelevant.

4.5.2 Lo

2

-to-M

I

Mapping

This section discusses the mapping from linguistic output2 (LO2) to the musical input (MI). The disyllabic LO2 is linked to either one or two musical beats in MI. When

‧

violation of DEP-V. However, (æ ktə) in (95a) is selected as the optimal output in that there is already an inserted vowel, and it does not violate DEP-V. Both (95d) and (95e) are ruled out by SSP. (95b) has an unassociated syllable and an unassociated musical beat, and is ruled out by NOSTRAY. In (95c), [t] is deleted and fatally violates MAX-C.

When the LO2, (æ ktə), is assigned with two musical beats, it is still mapped to the disyllabic (æ ktə),where each syllable is linked to one musical beat, as in (96).

(96) LO2-to-MI mapping: coda cluster (♩♩) LO2: (æ ktə) MI:(æ ktə)

  ♩ ♩

In (96c), two musical beats are linked to a closed syllable with a complex coda, which is ruled out by the conjoined NOSHARE-σ & *CCCODA. (96b) is ruled out byNOSTRAY, and (96d) by MAX-C. Eventually, (96a) emerges as the optimal output.

(æ ktə)

‧

musical beat association but removes the prosodic structure such as a prosodic word.

Given the constraints posited in Chapter 3, tableau (97-99) show how these constraints compete with each other.

(97) MI-to-MO mapping: coda cluster (/♩)

‧

In (97-99), candidates (b) are ruled out by *ProsSt, as they preserve the prosodic word in the output, and candidates (c) by ID-ASSOC, since the association lines are changed.

It should be noted that when (æ kt) of LO1 and (æ ktə) of LO2 are assigned with two musical beats, both emerge as (æ ktə) in MI, i.e., as the input in (99).

4.7 Summary

The mapping of a coda cluster also yields two linguistic outputs: LO1 is monosyllabic and faithful, while LO2 is disyllabic, with a vowel inserted to resolve the coda cluster. I have added the SSP to the set of constraints posited in Chapter 3, and the constraint rankings are enriched as in (100-101).

(100) LI-to-LO1 mapping: ()

ALIGN-E(LEX, ) DEP-V MAX-C

SSP

*CC

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

61

(101) LI-to-LO2 mapping: ()

ALIGN-E(LEX, ) *CC MAX-C

SSP

DEP-V

In the LI-to-LO mapping, there are two constraint rankings yielding two linguistic outputs, but in the LO-to-MI perception grammar, there is only a single constraint ranking, as in (102), where I have observed the influence of SSP on the insertion of vowels in singing.

(102) LO-to-MI Mapping

NOSTRAY MAX-C NOSHARE-σ & *CCCODA

DEP-V

SSP

ALIGN-R(♩, )

NOSHARE-B NOSHARE-σ

As posited in Chapter 3, the constraints ID-ASSOC, *PROSST, and MAX-PROSST serve to map between the musical input (MI) to the musical output (MO). The musical output preserves the musical beat association but removes the structure of the prosodic word.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

62

Chapter 5

Language-to-Music Mapping: Foot and Musical Beat Assignment

5.1 Introduction

Previous studies (Fang and Su 2005, Li 2009) have found that Mandarin children’s song values the rhythmic correspondence between lyric and song, which facilitates Mandarin learning of children. In terms of the connection between lyric and song, this chapter examines the language-to-music mapping of rhythmic structures in Mandarin children’s songs with a focus on the foot.

The mapping schema provided in Chapter 3 is also applicable here, which is reproduced below.

(103) Language-to-music mapping: rhythm

The syntactic structure of the lyrics in the linguistic input yields the prosodic structure in the linguistic output through the production grammar. The linguistic output is then mapped to the musical input through the perception grammar. The prosodic structure

Linguistic input

Production grammar Linguistic output

Perception grammar

Musical input

Production grammar

Musical output

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

63

in the musical input serves to condition the musical beat assignment, but is removed in the musical output through the musical production grammar.

This study constructs a database of Mandarin children’s songs based on five lyric books that contain thirty-nine songs.⁴ These books are for children from zero to eight years old, and the lyrics are simple and close to children’s daily lives. The songs in the abovementioned books are new songs, but not traditional songs. The composing of the musical melodies is based on existing lyrics.

This chapter discusses three questions of the language-to-music mapping. First, how is a foot formed in the linguistic output? Second, how does the structure of the foot affect the assignment of musical beats? Finally, how does the musical structure emerge in the musical output?

5.2 L

I

-to-L

O

Mapping: Foot Formation

Along the lines of Shih (1986) and Hsiao (1991), I consider the structure of the prosodic foot is formed in the linguistic output on four conditions. First, a pair of syllables that are ICs constitutes a binary foot. Second, adjacent syllables are paired into a disyllabic foot. Third, unparsed monosyllable joins its adjacent foot to form a bigger foot.

Finally, two adjacent syllables that syntactically branch in the opposite direction cannot form a foot.

A couple of trisyllabic examples are given in (104).

4 The five books are zao an wan an 早安晚安,xiao yu di 小雨滴, xiao hou zi 小猴子, and 1234 dong dong ti cao 1234 動動體操. One traditional song is included in one of the books, and is excluded from the database because it is not clear whether the lyrics or the music is composed first.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

64

(104) (a) ‘Hold a knife.’ (b) ‘Cut a pomelo.’

In (104), dao-zi and you-zi form two binary feet first. The leftover monosyllables, na and qie, then join the existing binary feet to form trisyllabic feet respectively.

Consider the first two syllables in (105).

(105) ‘(He) loves to somersault when (he) has nothing to do all day long.’

As in (105), after the three pairs of ICs, zheng-tian, mei-shi, and gen-tou, form binary feet, the adjacent unparsed syllables, ai and fan, are paired into a disyllabic foot.

Unlike ai and fan, the adjacent syllables, li and you, in (106) cannot form a foot, since they branch in the opposite direction.

拿 刀 子 切 柚 子

na dao - zi qie you - zi

Hold knife cut pemelo

∣ ∣f ∣ ∣^f

∣ ∣^f ∣ ∣^f

整 天 沒 事 愛 翻 跟 頭

zheng-tian mei-shi ai fan gen-tou

whole day nothing-to-do love turn somersault

∣ ∣^f∣ ∣f ∣ ∣^f

∣ ∣^f

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

65

(106) ‘There is a big caterpillar in a big apple.’

In (106), ping-guo and mao-mao are ICs, and thus they are grouped into disyllabic feet respectively. At this point, li and you syntactically branch in the opposite direction so that they cannot form a foot; the unparsed you and da are then paired into a disyllabic foot.

The leftover monosyllables, da, li and chong, subsequently join the adjacent existing feet to trisyllabic and tetrasyllabic feet.

Foot constraints governing the foot formation are proposed in (107-110), based on the concepts developed in work by Shih (1986), Hsiao (1991), and Lin (2001).

(107) Align-E(IC, Ft):

Assign one violation mark for every pair of syllables that are ICs whose edges do not coincide with the edges of a foot.

(108) FtBin:

Assign one violation mark for every foot that contains more than two syllables.

(109)*MONO-FT:

Assign one violation mark for every monosyllabic foot in the output.

大 蘋果 裡 有 大 毛毛 蟲 da ping-guo li you da mao-mao chong

big apple inside have big caterpillar

∣ ∣f ∣ ∣f ∣ ∣f

∣ ∣f ∣ ∣^f

∣ ∣^f

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

66

(110) Parse-σ:

Assign one violation mark for every syllable that is not parsed into any foot in the output.

The constraints Parse-σ and *MONO-FT must be undominated to ensure that every syllable is parsed into a foot but not a monosyllabic foot. Align-E(IC, Ft) is also ranked at the top to match IC edges with foot edges. FtBin is ranked at the bottom, such that a trisyllabic or larger foot can occur. A partial constraint ranking is posited in (111).

(111) LI-to-LO partial constraint ranking:

Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin

Tableau (112) shows how this constraint ranking works.

(112) = (104)

Input: [na [dao-zi]NP]VP [qie [you-zi]NP]VP

拿 刀 子 切 柚 子 Hold knife cut pomelo ‘Hold a knife and cut a pomelo.’

Candidates:

a. (na (dao-zi)Ft)Ft (qie (you-zi)Ft)Ft

b.

na (dao-zi)

Ft)Ft (qie (you-zi)Ft)Ft

c. (na dao)Ft (-zi qie)Ft (you-zi)Ft

d. (na)Ft (dao-zi)Ft (qie (you-zi)Ft)Ft

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

67

LI-to-LO: Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin Parse-σ Align-E(IC, Ft) *MONO-FT FtBin

 a. **

b. *! *

c. *!

d. *! *

The footing pattern in (112b) is ruled out by Parse-σ, as na is left unfooted. The pair of the ICs, dao and zi, in (112c) is not aligned with any foot, incurring a fatal violation of Align-E(IC, Ft). The monosyllabic foot, na, in (112d) violates *MONO-FT, and thus is ruled out. As a consequence, (112a) is chosen as the optimal output.

The constraint ranking in (111) also correctly predicts that the fifth and sixth syllables in (113) form a disyllabic foot.

(113) = (105)

Input: [[zheng-tian]ADV[mei-shi]VP [ai [ fan [gen-tou]NP]VP]VP 整 天 沒 事 愛 翻 跟 頭

whole day nothing-to-do love turn somersault

‘(He) loves to somersault when (he) has nothing to do all day long.’

Candidates:

a. (zheng-tian)Ft (mei-shi)Ft (ai fan)Ft (gen-tou)Ft b. (zheng-tian)Ft (mei-shi)Ft ai (fan (gen-tou)Ft)Ft c. (zheng-tian)Ft (mei-shi)Ft (ai)Ft (fan (gen-tou)Ft)Ft d. (zheng-tian)Ft (mei-shi)Ft (ai)Ft (fan gen-tou)Ft

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

68

LI-to-LO: Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin Parse-σ Align-E(IC, Ft) *MONO-FT FtBin

 a.

b. *! *

c. *! *

d. * (!) * (!) *

Again, (113b) violates Parse-σ, as ai is not parsed into any foot, whereas ai in (113c-d) constitutes a foot alone and violates *MONO-FT. The left edge of the pair of the ICs, gen and tou in (113d) is not aligned with the left edge of a foot, and thus is ruled out by Align-E(IC, Ft). Finally, (113a) emerges.

The same constraint ranking, however, renders an incorrect prediction in tableau (114).

(114) = (106)

Input: [[da [ping-guo]NP]NP

li]

PP [you [da [[mao-mao] chong]NP]NP]VP, 大 蘋 果 裡 有 大 毛 毛 蟲

big apple inside have big caterpillar

‘There is a big caterpillar in a big apple.’

Candidates:

a. (((da (ping-guo)Ft)Ft

li)

Ft (you da)Ft ((mao-mao)Ft chong)Ft

b. (da (ping-guo)Ft)Ft (li you)Ft (da (mao-mao)Ft chong)Ft

c. (da (ping-guo)Ft

li)

Ft you (da ((mao-mao)Ft chong)Ft)Ft

d. (da ping-)Ft (guo li) Ft (you (da ((mao-mao)Ft chong)Ft)Ft

e. (da (ping-guo)Ft

li)

Ft (you (da ((mao-mao)Ft chong)Ft)Ft)Ft

f. ((da (ping-guo)Ft)Ft li)Ft (you)Ft (da ((mao-mao)Ft chong)Ft)Ft

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

69

LI-to-LO: Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin

Parse-σ Align-E(IC, Ft) *MONO-FT FtBin

() a. ***!

 b. **

c. *! ***

d. *! ***

e. ***!*

f. *! ***!*

The foot structure in (114b) erroneously wins over (114a) by one less violation of FtBin, as indicated by the black right-headed hand symbol . The real optimal output is (114a), as indicated by the parenthesized white right-headed hand symbol (). In order to obtain (114a), the constraint in (115) must be added to the partial constraint ranking provided in (111).

(115) NOSTRADDLE-Ft:

Assign a violation mark for every foot consisting of syllables that syntactically branch in the opposite direction.

NOSTRADDLE-Ft forbids syllables that syntactically branch in opposite direction to form a foot, which I refer to as a straddle foot. The enriched constraint ranking is provided in (116) and the top-ranking of NOSTRADDLE-Ft now successfully removes the straddling foot in (117b).

(116) LI-to-LO constraint ranking (enriched):

NOSTRADDLE-Ft, Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

70

(117) = (114)

LI-to-LO: NOSTRADDLE-Ft, Parse-σ, Align-E(IC, Ft), *MONO-FT >> FtBin NOSTRADDLE-Ft Parse-σ Align-E(IC, Ft) *MONO-FT FtBin

(114a). ***

(114b). *! **

(114c). *! ***

(114d). *! ***

(114e). ***!*

(114f). *! ***!*

When the linguistic output is perceived into the musical input, the foot structure plays a role in assigning musical beats, as will be discussed next.

5.3 L

O

-to-M

I

Mapping: Footing and Musical Beat Assignment

The perception grammar that maps the linguistic footing output to the musical input rhythm involves three questions. First, in what way can a musical beat be shared by syllables? Second, in what way can a syllable be shared by musical beats? Finally, how are musical beats confined by the domain of the foot? Before I enter the discussions, some musical basics are introduced in the following section.

5.3.1 Some Musical Basics

The names and durations of the musical notes are given in (118).

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

71

(118)

w

_{whole note}

h h

_{half note}

q q q q

quarter note

É É É É É É É É

eighth note

s s s s s s s s s s s s s s s s

sixteenth note

The whole note (

w

) is the top note, which is equal to two half notes (

h

). A half note is equal to two quarter notes (

q

). The duration of one quarter note is equal to two eighth notes (

É

). One eighth note is equal to two sixteenth notes (

É

). When there is a dot besides a note, the duration of that note becomes 1.5 times of itself. For example, if the length of

q

represents one beat,

j

indicates one and a half beat.

There are two kinds of time signatures in the database, namely 4/4 and 6/8. The time signature indicates the beat number in each measure and the note value that is equal to a musical beat.

(119) Time signature: 4/4

Music metric accent (primary accent: > secondary accent: >)

> >

' q q q q '

As in (119), the time signature is 4/4. The first digit is the beat number in each measure, whose edges are signaled by bar lines,

'

. The second digit indicates the note value that is equal to a beat. Therefore, 4/4 means that there are 4 beats in each measure and a quarter note is equal to a beat. In the 4/4 time signature, the default primary accent (>) is assigned to the first beat while the default secondary accent (^>) is assigned to the third beat.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

72

The time signature in (120) is 6/8, which means that there are six beats in each measure and an eighth note (

e

) is equals to a beat. In the 6/8 time signature, the first beat has the primary accent while the fourth beat has the secondary accent.

(120) Time signature: 6/8

Music metric accent (primary accent: > secondary accent: ^>)

> >

'e e e e e e '

When two notes are linked by a concave arc line (U), the duration is the combination of the two notes.

(121) Two musical beats

qU q

As in (121), the linked quarter-notes signal two musical beats when a quarter note is equal to a beat.

5.3.2 Musical Beat Assignment

The mapping of the lyric output rhythm to the musical input is keyed to the assignment of musical beats. Two relevant constraints are posited in (122-123).

(122)NOSHARE-(B5):

Assign one violation mark for every syllable that is shared by five or more musical beats.

‧ 國

立 政 治 大 學

‧

N a tio na

l C h engchi U ni ve rs it y

73

(123)NOSPLIT-IC(B):

Assign a violation mark for every pair of syllables that are syntactic ICs but are not associated with the same musical beat.

As previously proposed in Chapter 3, NOSTRAY stems from the notion of stray erasure, and bans any unassociated musical beat or syllable. NOSHARE-(B5) indicates the musical limit on syllable lengthening, and disallows a syllable to be shared by five or more than five musical beats. NOSPLIT-IC(B) suggests the tight connection between ICs, requiring a pair of ICs to share a musical beat. NOSTRAY and NOSHARE-(B5) are ranked above NOSPLIT-IC(B), as ICs do not always share a musical beat. NOSHARE-σ and NOSHARE-B, which are posited in Chapter 3, are ranked at the bottom, as a syllable is often linked to multiple musical beats to prevent stray elements and vice versa.

The partial constraint ranking is given in (124), and the tableau in (125) shows how this constraint ranking works.

(124) LO-to-MI partial constraint ranking

NOSTRAY,NOSHARE-(B5)>>NOSPLIT-IC(B),NOSHARE-B,NOSHARE-σ

(125)

LO: (zheng-tian)Ft (mei-shi)Ft (ai fan)Ft (gen-tou)Ft

整 天 沒 事 愛 翻 跟 頭 whole day nothing-to-do love turn somersault

‘(He) loves to somersault when (he) has nothing to do all day long.’

Music time signature: 6/8 (There are six beats in each measure.)

| e e e e e e | e e e e e e |

‧

The linguistic output of (125) emerges from tableau (113). The floating musical beats in (125b) incur fatal violations of NOSTRAY. (125c) is ruled out by NOSHARE-σ(B5), as fan is shared by five musical beats. Consequently, (125a) is selected as the optimal output, in spite of three violations of NOSPLIT-IC(B) and NOSHARE-σ.

The linguistic output selected in tableau (117) is mapped into music in (126).

(126)

LO: ((da (ping- guo)Ft)Ft li)Ft (you da)Ft ((mao-mao)Ft chong)Ft

大 蘋 果 裡 有 大 毛 毛 蟲 big apple inside have big caterpillar

‘There is a big caterpillar in a big apple.’

Music time signature: 4/4 (There are four beats in each measure.)

| q q q q | q q q q |

‧

There is a floating syllable in (126b) and a floating musical beat in (126e); hence both are ruled out by NOSTRAY. The pair of ICs, ping-guo, in (126d) is associated by two different musical beats, and thus is ruled by by NOSPLIT-IC(B). The real optimal output is (126a), but (126c) wrongly emerges as well, as indicated by the  symbol. To remove (126c), we need a constraint to prevent syllables that belong to different feet to share the same

‧

Assign one violation mark for every beat that is shared by syllables from different feet.

The constraint NOSTRADDLE-B(FT) is also top-ranked as shown in the enriched constraint ranking in (128). Tableau (126) can now be reanalyzed as tableau (129).

(128) LO-to-MI constraint ranking (enriched)

NOSTRADDLE-B(FT), NOSTRAY, NOSHARE-(B5)>> NOSPLIT-IC(B),NOSHARE-B, musical beat; hence it is ruled out by NoSTRADDLE-B(Ft).

5.4 M

I

-to-M

O

Mapping: Prosody Removal and Association Faithfulness

The production grammar that maps the musical input (MI) to the musical output (MO) preserves the musical beat association but removes the prosodic structure such as prosodic

NOSTRADDLE

‧

word. Consider the tableau in (130).

(130)

‘(He) loves to somersault when (he) has nothing to do all day long.’

Music time signature: 6/8 (There are six beats in each measure.)

| e e e e e e | e e e e e e |

The musical input of (130) emerges from tableau (125). The associated line in (130c) is different from that in the musical input, and (130c) thus is ruled out by ID-ASSOC. In

ID-ASSOC *PROSST MAX-PROSST

 a. *

b. *!

c. *! *

‧

The MI-to-MO mapping in (131) is evaluated by the same constraint ranking.

The MI-to-MO mapping in (131) is evaluated by the same constraint ranking.

在文檔中 英語與華語歌曲中語言與音樂之關係：音段、節奏與聲調 - 政大學術集成 (頁 59-0)