0(, 1(2' ,0,3 .-(! ,/ %,4'0(, / ./(*' &&(5 #

Chapter 5

1

2

3

The Sociophonetics of Prosodic Contours on NEG

4

in Three Language Communities: Teasing apart

5

Sociolinguistic and Phonetic Inuences on Speech

6 7

8

9

Malcah Yaeger-Dror, University of Arizona; Shoji Takano,

10

Hokusei Gakuen University; Tania Granadillo, University of

11

Western Ontario; Lauren Hall-Lew, Oxford University1

12

13

14

1. Introduction

15

16

Negatives provide cognitively critical information and are also interactively

17

signicant. The present study compares the prosodic realization of nega-

18

tives in three languages, and in two social settings for each language. The

19

study will provide evidence for three loci of prosodic variation in negatives

20

as they are used in amicable social interactions and in informative newscasts

21

in American English, Latin American Spanish,2 and Japanese. Comparative

22

evidence from adversarial interactions will be cited where relevant.

23

24

25

1.1 Language

26

27

Each of these three languages shows unique patterns for how prominence is

28

acoustically accomplished (Pierrehumbert and Beckman 1988; Hirst and di

29

Cristo 1998; Grabe et al. 2003; Jun 2005). Each has its own default negative

30

morphology with a given default syntactic position, and it is that most com-

31

mon form of negation that will be studied here. Rather than refer to each spe-

32

cic lexical item in this discussion, each language's `default' lexical negative

33

will be referred to as NEG.

34

35

36

1.2 Social situation

37

38

Within a given linguistic community, prosody varies radically with the social

39

situation. This chapter will discuss parallel results for the three languages in

40

PPrreesstotonn11stspt apgaegs.einsd.idndd133410

33/2/129/2/2001100 13:14:71:82:55P0MAM

134 Malcah Yaeger-Dror, Shoji Takano, and Lauren Hall-Lew

only two situations: friendly phone calls will be compared with recordings

1

of newscasts. Analyses of other situations can be found elsewhere (e.g., Yae-

2

ger-Dror 2002a, b; Yaeger-Dror, Hall-Lew, and Deckert 2002, 2003; Takano

3

2002, 2008; Kato 2004).

4

5

6

1.3 Culture

7

8

In different societies, prosodic prominence is manipulated in various ways,

9

even in apparently similar social situations. Some of these dissimilarieties

10

are purely linguistic (e.g., Grabe et al. 2003; Mennen 2006; Ladd et al. 2009),

11

while others appear to be culturally variable and may be dependent on soci-

12

etal norms of power and solidarity (Brown and Gilman 1960; Maclemore

13

1991; Watts 2003; Locher 2004). The present study argues that neither source

14

of variation should be ignored.

15

This study will permit cross-cultural and cross-linguistic comparisons,

16

showing that there are nontrivial language-specic and culture-specic

17

components. Cognitive, linguistic, situational, and cultural factors must

18

all be incorporated as variables for any analysis of the prosody of negation

19

strategies.

20

21

22

2. Review of the relevant literature

23

24

2.1 Parameters of prosodic variation

25

26

There are three primary phonetic parameters of prosodic variation which can

27

be mined for sociophonetic detail: loudness, measured acoustically as ampli-

28

tude (in decibals: dB), pitch variation, measured acoustically from a speaker's

29

fundamental frequency, or F (in HZ), and duration (where the duration of the 0

30

word or its linguistic subcomponents can be compared with the duration of

31

nearby equivalent tokens and is measurable in milliseconds--or msec). Figure

32

5.1 shows that all three are measurable using commonly available software:3

33

F0 is found on the lowest vertical axis"Pitchtrack" and "Amplitude" has its

34

own vertical axis immediately above it; "Duration" is measured along the

35

horizontal axis. Each of these parameters is manipulated to varying degrees

36

in different languages. Fortunately, in all three languages investigated here,

37

the primary perceptual and productive parameter for prominence is funda-

38

mental frequency (F0) and is measurable from the pitchtrack itself (Fagyal

39

and Yaeger, forthcoming).

40

PPrreesstotonn11stspt apgaegs.einsd.idndd134410

33/2/129/2/2001100 13:14:71:82:65P0MAM

The Sociophonetics of Prosodic Contours on NEG 135

1

2

3

4

5

6

7

8

9

10

11

Figure 5.1 Examples of Pitch (F0), amplitude, and duration measures.

12

13

14

Experimental studies have shown that for speakers of Standard American

15

English, amplitude generally appears to co-vary with fundamental frequency;

16

duration appears to be correlated with both sentential position and focal prom-

17

inence. While amplitude increments can be `perceived' as `accenting' a word

18

even in the absence of a fundamental frequency change, this is not common

19

even in a carefully read corpus (Cutler, Dahan, van Donselaar 1997).

20

In Japanese (as in English), experimental studies demonstrate that fun-

21

damental frequency plays the primary role in both production and percep-

22

tion of focal prominence (Pierrehumbert and Beckman 1988; Venditti 2005),

23

whereas amplitude and duration also participate as subsidiary parameters

24

(Sugitou 1982; Koori 1989a, b; Azuma 1992).

25

In Spanish (Navarro-Tom?s 1944; Sosa 1999; Face 2001, 2002; Estebas-

26

Vilaplana 2007) and other Romance languages as well (Di Cristo 1998; Dahan

27

and Bernard 1997) as well, focal prominence is produced primarily by vary-

28

ing fundamental frequency, while amplitude and durational prominence are

29

used primarily for other purposes.

30

In short, each of the three languages investigated here permits us to mea-

31

sure and code this primary parameter for prominence (F0) directly from the

32

pitchtrack, as shown on the example in Figure 5.1, taken from the rst Ken-

33

nedy/Nixon debate.

34

35

36

2.2 Cognition and prosodic salience

37

38

Bolinger (1978) proposed that prosodically emphasizing critical seman-

39

tic information is a cross-linguistic universal. Prosodic focal prominence

40

PPrreesstotonn11stspt apgaegs.einsd.idndd135410

33/2/129/2/2001100 13:14:71:82:65P0MAM

136 Malcah Yaeger-Dror, Shoji Takano, and Lauren Hall-Lew

maximizes the ability of conversational partners to focus attention on infor-

1

mation which is critical to mutual understanding (Cutler et al. 1997). The

2

assumed motivation for such prosodic salience will be referred to here as the

3

Cognitive Prominence Principle.

4

In addition, even within a single language dialects differ in their use of

5

prosodic prominence ( e.g., Beckman et al. 2002; Grabe et al 2002; Fagyal

6

2004; O'Rourke 2005; Thomas and Carter 2006; Mennen 2007; Estebas-Vila-

7

plana 2007; Ladd et al 2009).

8

9

10

2.3 Sentential position and prosody

11

12

Syntactic position within a sentence inuences prosodic options (Ladd 2008),

13

and it is possible to manipulate focus by altering such positions (e.g., Ochs,

14

Schegloff, Thompson 1996; Danieli et al. 2004; Couss? et al. 2004; Swerts

15

and Wijk 2005, inter alia). The unmarked sentence contour in most languages

16

permits an early prosodic peak with downstep narrowing the permissible F0

17

range later in the sentence. Many studies have documented that critical infor-

18

mation is more likely to be placed early in the sentence, and that material

19

presented early in the sentence is most likely to be prosodically prominent

20

(e.g., Cutler et al. 1997; Horne 2000; Jun 2005; Ladd 2008).

21

In theory, the closer the NEG is to the beginning of the sentence, the

22

greater the range and manipulability of prosodic prominence, so a speaker's

23

option to exploit the position of NEG to emphasize or neutralize its cognitive

24

salience is relevant to the discussion. Discussion of variation of placement to

25

manipulate prosodic prominence can be found in Horne (2000), Jun (2005)

26

and Takano (2008).

27

In declarative sentences, the unmarked placement for negatives analyzed

28

here--NEG--includes `verbal- no' for Spanish, not for English, nai for Japanese:

29

Spanish verbal-no occurs before the verb, near the beginning of the sentence;

30

English not immediately follows the English `AUX' verb, and precedes the main

31

verb, while nai occurs near the end of the sentence (Takano 2008; Jun 2005).

32

Even given that there is a strong preference for the unmarked position, it

33

is reasonable to assume that the likelihood of prominence in any given case is

34

mediated by the NEG's unmarked position in the sentence.

35

Considering both production and perception studies, Cutler et al. (1997)

36

conclude that "speakers seldom de-accent (critical) information, and if they

37

do, this hinders listeners." They show that while a prominent syntactic posi-

38

tion can be neutralized by the overriding signicance of other words in the

39

environment, focally informative words are unlikely to be reduced because

40

PPrreesstotonn11stspt apgaegs.einsd.idndd136410

33/2/129/2/2001100 13:14:71:82:85P0MAM

The Sociophonetics of Prosodic Contours on NEG 137

of their syntactic position. That is, words that carry critical information will

1

be prosodically prominent even if their syntactic position would minimize

2

prominence. Cutler et al.'s conclusion will be referred to as Cutler's Corol-

3

lary. Note that studies which support the corollary claim have been carried

4

out on both English (cf., op cit. and references therein) and French corpora

5

(Benguerel 1970; Dahan and Bernard 1997). A large segment of this chapter

6

is devoted to the analysis of ways in which negatives are either prosodically

7

prominent (supporting that claim) or not (possibly refuting the claim).

8

Consequently, the relative importance of the Cognitive Prominence Prin-

9

ciple and Cutler's Corollary with regard to actual NEG positions and prosody

10

in each of these languages will be discussed further in Section 3.4.

11

12

13

2.4 NEG and prosody

14

15

The point of departure for studies of negation and prosody was developed in

16

the work of Bolinger (1978), who claimed that cross-linguistically NEG will

17

receive "negative prominence." We have taken that to mean prominence that

18

would be represented in ToBI4 transcription with L*, and which would have

19

F0 no higher than nearby prosodically neutral words; analyses to date do not

20

support this claim.

21

O'Shaughnessy and Allen (1983) looked specically at negatives as carriers

22

of critical information. They elicited isolated sentences with negatives that car-

23

ried information which `focal prominence' is intended to highlight: they found

24

that NEG were almost categorically prominent which they attributed to their con-

25

veying cognitively critical information. While O'Shaughnessy did not charac-

26

terize this "prominence," the pitchtracks of the elicited sentences revealed that

27

overwhelmingly the NEG were either rising, rise-fall or high level--all variations

28

on the ToBI theme of H*, rather than the L* proposed by Bolinger (1978).

29

Subsequently, Hirschberg (1990, 1993) analyzed news reports read by

30

WBUR radio announcers (.

31

jsp?catalogId=LDC96S3; henceforth "BUR"); the newscasters were re-read-

32

ing National Public Radio stylized newscasts. Like O'Shaughnessy, she found

33

that the vast majority of prominences on NEG were H*. More recent studies

34

(Syrdal et al. 2001; Hirschberg 2000) present similar results; in fact, both

35

English not tokens (Hirschberg 1990, 1993) and French pas tokens (Morel

36

1995; Jun 2005) are reported as consistently pitch-raised in read speech, as

37

would be projected from the Cognitive Prominence Principle, although (con-

38

trary to Cutler's Corollary) French negatives inside relative clauses are not

39

necessarily prominent in isolated read sentences (Jun 2005).

40

PPrreesstotonn11stspt apgaegs.einsd.idndd137410

33/2/129/2/2001100 13:14:71:83:05P0MAM

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download