26
Prosody on the hands and face Sign Language & Linguistics 5:2 (2002), 105130. issn 13879316 / e-issn 1569996X© John Benjamins Publishing Company Evidence from American Sign Language Diane Brentari and Laurinda Crossley Purdue University The analysis in this paper deals with the prosodic cues that were present in a one- hour lecture by a native signer of American Sign Language (ASL). Special attention is paid to the interaction of the dominant hand (H1) and the nondominant hand (H2), as well as to facial expressions articulated on the lower face. In our corpus, we found that H1 and H2 interact in several prosodic contexts; we analyze four of them here: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical, and Forward-Referencing. Our main finding is that, while the spread of the nondominant hand (H2-Spread) is an important redundant cue to prosodic structure, the primary cue is on the lower face. Our findings also confirmed posi- tional cues and domain effects of H2-Spread in Prosodic Words and Phonological Phrases that were previously found in Israeli Sign Language. Keywords: sign language, phonology, prosody, nonmanual behavior, nondominant hand, prosodic word, parenthetical, forward referencing 1. Introduction In this paper we demonstrate that suprasegmental prosodic cues in American Sign Language (ASL) provide important information about prosodic constituency and about the meaning of utterances. This is one part of a larger study that investigates the ways in which various suprasegmental properties of American Sign Language individu- ally contribute to the meaning of utterances and to the criteria used to define prosodic constituency. These properties include not only the interaction of the two hands, but also eyegaze, lower- and upper-face nonmanuals, eyeblinks, and extended holds. Our work will also demonstrate how these properties systematically interact with one another in prosodic constituents.

Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face

</SECTION "opt"><SECTION "art" TITLE "Articles">

Sign Language & Linguistics 5:2 (2002), 105–130.

issn 1387–9316 / e-issn 1569–996X�©John Benjamins Publishing Company

<TARGET "bre" DOCINFO AUTHOR "Diane Brentari and Laurinda Crossley"TITLE "Prosody on the hands and face"SUBJECT "SL&L, Volume 5:2"KEYWORDS "sign language, phonology, prosody, nonmanual behavior, nondominant hand, Prosodic Word, parenthetical, forward referencing"SIZE HEIGHT "240"WIDTH "170"VOFFSET "4">

Evidence from American Sign Language

Diane Brentari and Laurinda CrossleyPurdue University

The analysis in this paper deals with the prosodic cues that were present in a one-hour lecture by a native signer of American Sign Language (ASL). Special attentionis paid to the interaction of the dominant hand (H1) and the nondominant hand(H2), as well as to facial expressions articulated on the lower face. In our corpus, wefound that H1 and H2 interact in several prosodic contexts; we analyze four of themhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase,Parenthetical, and Forward-Referencing. Our main finding is that, while the spreadof the nondominant hand (H2-Spread) is an important redundant cue to prosodicstructure, the primary cue is on the lower face. Our findings also confirmed posi-tional cues and domain effects of H2-Spread in Prosodic Words and PhonologicalPhrases that were previously found in Israeli Sign Language.

Keywords: sign language, phonology, prosody, nonmanual behavior, nondominanthand, prosodic word, parenthetical, forward referencing

1. Introduction

In this paper we demonstrate that suprasegmental prosodic cues in American SignLanguage (ASL) provide important information about prosodic constituency andabout the meaning of utterances. This is one part of a larger study that investigates theways in which various suprasegmental properties of American Sign Language individu-ally contribute to the meaning of utterances and to the criteria used to define prosodicconstituency. These properties include not only the interaction of the two hands, butalso eyegaze, lower- and upper-face nonmanuals, eyeblinks, and extended holds. Ourwork will also demonstrate how these properties systematically interact with oneanother in prosodic constituents.

Page 2: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

106 Diane Brentari and Laurinda Crossley

1.1 Background on suprasegmental prosodic cues in sign languages

In this section, we want to provide some background on a set of phenomena that areused to define prosodic categories in sign languages. We are assuming the prosodichierarchy of Nespor and Vogel (1986): utterance, intonational phrase (IP), phono-

<LINK "bre-r14">

logical phrase (PP), clitic group, prosodic word (PW), syllable (σ). In our study of ASL,most, but not all, of the elements used for identifying prosodic categories in this paperhave to do with the timing of elements that are already present in the signal forinformational purposes. Identifying their prosodic role is based on how far theseproperties spread (and in what direction) and with what they co-occur. The next sub-sections describe each of the prosodic behaviors important in our study.

1.1.1 Handshape, movement and place of articulationPrevious work on ASL reveals that the preferred PW has one movement or a limited rangeof movement combinations (Coulter 1982, Brentari, 1998), one set of selected fingers

<LINK "bre-r6"><LINK "bre-r4">

(Mandel, 1981, Sandler, 1989), and one place of articulation (Sandler, 1987, Brentari,

<LINK "bre-r18"><LINK "bre-r18"><LINK "bre-r4">

1998). While not every word conforms to these constraints, even compounds (Sandler,

<LINK "bre-r18">

1989; Brentari, 1990) and borrowed words that enter the lexicon through fingerspelling

<LINK "bre-r4">

(Battison 1978, Brentari & Padden, 2001) confirm that these constraints are active in

<LINK "bre-r1"><LINK "bre-r4">

shaping new words in the ASL lexicon. Nespor and Sandler (1999) have argued that in

<LINK "bre-r14">

non-final position, which they call the ‘weak’ position within a PP, handshapeassimilation is a marker of a PW combining a pronominal clitic and a verb.1

1.1.2 The interaction of the two handsThe two hands interact with each other while signing, but they are not unconstrained.Battison (1978) was the first study of the interaction of the two hands in single signs,

<LINK "bre-r1">

and there have been several studies of the distribution of the two hands in longerstrings, and with respect to various phonological rules (Frishberg 1985; Brentari 1990,

<LINK "bre-r10"><LINK "bre-r4">

1998, Nespor & Sandler 1999; Sandler 1999a,�b). Battison’s typology of signs, shown in

<LINK "bre-r14"><LINK "bre-r18">

(1), is assumed in this paper.

(1) Battison’s typology of signs (Battison 1978:28):

<LINK "bre-r1">

Type Ø: 1-handed signs articulated in free space without contact (e.g. preach)Type X: 1-handed signs that contact the body in any place except the opposite

hand (henceforth called H2; e.g. sour, government)Type 1: 2-handed signs in which both hands are active and perform identical

motor acts; the hands may or may not contact each other, may or may

1.�Nespor and Sandler (1999) and Sandler (1999a,�b) have analyzed Israeli Sign Language, but theirconclusions can serve as a set of working hypotheses for other sign languages, including ASL.

Page 3: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 107

not contact the body, and they may be in either a synchronous or alter-nating pattern of movement (e.g. which, car, restrain-feelings)

Type 2: 2-handed signs in which one hand is active and one hand is passive, butboth hands are specified for the same handshape (e.g. short/brief, sit)

Type 3: 2-handed signs in which one hand is active and one hand is passive,and the two hands have different handshapes (e.g. discuss, contact

(a person))

The hands have been assigned the role of dominant (H1) or nondominant (H2), basedon which hand (the left or the right) signs 1-handed signs and executes fingerspelledsigns. H2/H1 interaction is a advantageous place to begin to study prosodic effects inthe sign stream because it is readily observable and it appears to be more invariantacross signers, while some facial markers of prosodic constituency appear to be morevariable; Nespor and Sandler (1999) begin here. In larger units than the word, the

<LINK "bre-r14">

handshape of H2 (of all types of 2-handed signs) has been observed to spread regres-sively (backwards) or progressively (forwards) systematically in two ways in Israeli SignLanguage (ISL). One way, called “Coalescence”, is argued to mark a PW and occurs atthe end of an IP and spreads progressively from the trigger to a single target, which isan indexical, pronominal sign (Sandler 1999a,�b). The other way, called “H2 spread” is

<LINK "bre-r18">

forward or backward spreading in larger units, that are argued in Nespor and Sandler

<LINK "bre-r14">

(1999) and Sandler (1999a) to be within the domain of the PP (or the ‘intermediate

<LINK "bre-r18">

phrase’ in Pierrehumbert 1980). In this paper, the entire range of cases of H2-Spread,

<LINK "bre-r17">

both the PW Coalescence cases of Sandler (1999a,�b) and the PP H2-Spread cases of

<LINK "bre-r18">

Nespor and Sandler (1999), are referred to here as H2-Spread.

<LINK "bre-r14">

(2) Examples of Coalescence and H2-Spread in ISL2

a. Coalescence b. H2-Spread in PPs(from Sandler 1999a,�b) (from Nespor & Sandler 1999)

<LINK "bre-r18"><LINK "bre-r14">

H1: store index]pw]ip persuade study…]pp]ip

H2: store persuade

A phenomenon related to H2-Spread, called “dominance-shift” has also been dis-cussed in the literature (Frishberg 1985), but the assimilation of H2 in these instances

<LINK "bre-r10">

has heretofore not been addressed. Dominance shift occurs when H2 becomes theprimary signing hand for a string of signs, sometimes this occurs within a sentence, asshown in (3), or sometimes throughout a longer string, such as when the two handsare assigned specific characters in a narrative. It is a dominance shift because H2 is

2.�The transcription for our examples may include all of the following, as appropriate for itspresentation: Upper Face (UF), Lower Face (LF), H1 glosses, and H2 glosses (H2). Blinks areindicated ‘^’ in the H1 gloss, and H2-Spreading is shown by underlining from the target sign in thedirection of spreading, as far as the spreading continues.

Page 4: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

108 Diane Brentari and Laurinda Crossley

signing the 1-handed signs typically articulated by H1. When the dominance switchesoccur, handshape spreading may optionally occur as well, whereby the H1 handshapeis held in the signing space while H2 signs an independent phrase or longer string. In(3), the H1 handshape held is that of why, the interjected phrase uttered during thedominance switch is only-one-me. H2 is then held during the remainder of theutterance when dominance switches back.3 The parenthetical is only-one-me.

(3) Dominance Shift (from Frishberg 1985)

<LINK "bre-r10">

H1: army rush-at-us embarrassed why girl five boy five

H2: only-one-me

‘The soldiers came right at us. I was embarrassed because — I was alone — agirl with five guys.’

This example shows something else that is an interesting phenomenon related toH2-Spread; namely, that it is not only a phenomenon that occurs when 2-handed signsare in the sign stream. Under certain conditions, the handshapes of 1-handed signsspread as well; sometimes they act as a trigger of spreading in cases of switch domi-nance. The handshape of a 1-handed sign is held on H1 while H2 continues to sign.Moreover, some 2-handed signs can be realized as 1-handed forms, under conditionsof “Weak Drop” (Padden & Perlmutter 1987), a type of phonological deletion of H2.

<LINK "bre-r15">

The effects of Weak Drop could cause a 2-handed sign to behave like a 1-handed signin the sign stream (e.g., it might not block the spread of H2 because H2 is not obligatoryin such signs).

It has been reported in the ASL literature that the timing of H1 and H2 interactioncan have consequences for morphosyntactic constituency. Two kinds of cases involveclassifiers (Padden 1983) and determiners (Zimmer & Patschke 1990). In sentences

<LINK "bre-r15"><LINK "bre-r22">

containing classifier predicates, Padden (1983) has argued that there is a difference in

<LINK "bre-r15">

syntactic constituency in (4a) vs. (4b) that depends on the interaction of H1 and H2.Padden argues that (4a) is bi-clausal, while (4b) is monoclausal. In (4a) the classifierconstruction ‘flat surface’ is signed alone, without the simultaneous articulation ofmonkey; that is there is an interruption of the H1 sign stream while the classifierconstruction is produced on H2. In contrast in (4b) monkey is simultaneouslyarticulated with the classifier construction.

(4) Differences in syntactic constituency based on H2-Spread (from Padden

<LINK "bre-r15">

1988:217)a. H1: tablea monkey CL: ‘small animal’+ jumpa

H2: tablea CL: ‘flat surface‘There is a table and the monkey jumped onto it.’

3.�No additional nonmanual behaviors were transcribed in this utterance reported in Frishberg(1985), so we do not know where the blinks are or what nonmanual behaviors are involved.

Page 5: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 109

b. H1: tablea monkey CL: ‘small animal’+ jumpa

H2: tablea CL: ‘flat surface’‘The monkey jumped onto the table.’

c. H1: *tablea monkey CL: ‘small animal’+ jumpa

H2: *tablea CL: ‘flat surface’‘The monkey jumped onto the table.’

Moreover, H2-Spread across CL: ‘small animal’+jump is required, as shown by theungrammaticality of (4c), where spreading does not occur. In the cases discussed in ISL,the absence of Coalescence or H2-Spread does not make the utterance ungrammatical;H2-Spread is optional when used for prosodic purposes. We have confirmed thisdifference using elicitation data showing the presence and absence of H2-Spread in thesame prosodic constituent produces different grammaticality judgments, based onwhether the string contains an H2 that is a classifier construction or not. A full accountof these facts must await further investigation, but this type of data was our basis forkeeping morphological and phonological cases of H2 use in different data sets.

Another morphological use of spreading occurs when an indexical point used asa determiner is signed by H2 (Zimmer & Patschke 1990). The determiner is signed on

<LINK "bre-r22">

H2, while the H1 signs a nominal. An example of this appears in (5).4

(5) Determiner use of indexical pointUF: topicH1: most important what ^ attitude

H2: most important what ^ index

‘It’s the attitude that is most important.’

This phenomenon is very similar to Sandler’s Coalescence. Sandler suggests thatCoalescence occurs with a Type 1, 2-handed sign, followed by indexical point (locativeor pronominal) articulated on H1. In the NP+determiner cases, the indexical point isarticulated by H2. The difference between these two forms is in the hand that articu-lates the indexical point — H1 in ISL Coalescence, H2 in ASL determiners — and thefollowing subtle difference of timing. Sandler claims that in Coalescence, the indexicalpoint appears “half-way through the articulation of the Type 1, 2-handed sign, and[the indexical point] loses its syllabicity”. Zimmer and Patschke (1990) describe

<LINK "bre-r22">

determiners as occurring simultaneously with the noun.In both the classifier case in (4) and the determiner case in (5), H2 is doing more

grammatical work than marking prosodic constituency. In order to keep our data setfree of these effects, we focused on forms that could not be confused with a morpho-

4.�We should point out that not all determiners in ASL are simultaneous articulations of H1 andH2, but these are the ones that might be confused with Coalescence, so we are only describing thesehere.

Page 6: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

110 Diane Brentari and Laurinda Crossley

syntactic use of H2-spread. An example is given in (6) of such a case. sign-asl-

fluently and CAN are both Type 1, 2-handed signs, both signed in neutral space, andneither of them has additional affixal morphology associated with them.

(6) Purely phonology use of H2-SpreadUF: brow raise/conditionalH1: sign-ASL-fluently can ^ go-ahead

H2: sign-ASL-fluently ^ go-ahead

‘If (one) can sign fluently, (he/she) can go-ahead (and enter the community).

In this paper we consider phonological cases of what Sandler calls Coalescence, such asthose in (6). H2 is not a separate morpheme all of the cases we present; rather, H2 is onlysustaining the handshape of a previous sign. The determiner and clitic cases weredeleted from our data set because H2 is used as a separate morpheme in these cases.

1.1.3 Nonmanual suprasegmentalsThere is a small, but growing, body of work on timing effects and nonmanualbehaviors in sign languages, both of which are thought to be markers of prosodicstructure. These include lengthening effects, as well as lower face behaviors, eyeblinks,and torso leans.

Lengthening: Perlmutter (1992) has discussed the rule of Mora-Insertion in ASL,5

<LINK "bre-r16">

and Miller (1996) discusses this operation in Langue des signes quebecoises (LSQ).

<LINK "bre-r12">

Sandler (1999a) has discussed lengthening specifically with respect to H2-Spread in

<LINK "bre-r18">

ISL, and has claimed that lengthening occurs at the right edge of a PP.6

Lower Face: The timing of gestures on the mouth and lower face has been investi-gated in sign languages other than ASL. Bergman and Wallin (2001) argue that

<LINK "bre-r2">

Swedish SL has a set of noun classifiers based on the timing of the mouth gestures informs such as ‘man-CL:baby-c handshape’, in which a single mouth pattern extendsacross the noun and the classifier. This is used as evidence that the two morphologicalentities are part of the same constituent. Sandler (1999b) uses the spread of the mouth

<LINK "bre-r18">

pattern of the host to the clitic as supporting evidence that the two parts involved inCoalescence form a single word. Boyes Braem (2001) has analyzed Swiss German SL

<LINK "bre-r3">

(DSGS) mouth patterns in early and late learners and found that mouth patterns areimportant in phrasal constituents in this language. Early learners of DSGS extendmouth patterns across more than one sign, as shown in (7a–c).

5.�Perlmutter (1992) discusses this lengthening effect in terms of the word, but this domain wasshown to be incorrect in Miller (1996) for LSQ and in Brentari (1998) for ASL.

6.�Sandler (1999) claims that reduplication is a PP boundary marker as well, but in our data all ofthe instances of reduplication occur at the right edge of a PP that is also the rightward boundary ofan IP.

Page 7: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 111

(7) Examples from DSGS showing the extension of mouth patterns across largerconstituents than the word (from Boyes Braem 2001)

<LINK "bre-r3">

mouth: a. aufpa---ssen b. au-------to c. do-------ktorsigns: aufpassen ich auto CL:b-hs doktor personpl

English: ‘I pay attention.’ ‘Car, CL:vehicle go_by’ ‘doctors.’

Eyeblinks: Wilbur (1994) has argued that there are two possible types of eyeblinks used

<LINK "bre-r20">

for linguistic purposes — inhibited eyeblinks, which are a boundary marker at IP, andvoluntary eyeblinks that are a marker of emphasis and occur throughout a final lexicalsign. We found evidence of both in our data, but the ones that are relevant to ourdiscussion here are inhibited blinks. These are short blinks that occur either just afteror only slightly overlapping with the rightmost sign in an IP. The predicted locationfor these blinks is an “ungoverned maximal projection”. Because signing rate can effecthow many possible sites for such eyeblinks actually get them, such blinks were foundto be a prosodic, rather than a purely syntactic, phenomenon.

Torso Leans: Body leans have a variety of meanings, as shown in Wilbur and

<LINK "bre-r20">

Patschke (1998). These include (1) inclusion or exclusion, (2) contrastive focus, and(3) affirmation. In our data, we found the contrastive focus use, and when it wasobserved, this nonmanual marker co-occurred with eyeblinks.

We took as given the findings about eyeblinks at IPs and lengthening at PPs todetermine where the boundaries for these prosodic constituents occurred, indepen-dently from H1/H2 interactions. What we want to know is whether a manual marker— i.e. H2-Spread — would confirm this constituent structure. If this were the case, itwould advance our understanding of prosodic structure and confirm findings fromprevious work using a linguistic marker that has attracted considerably more attentionin the literature over a longer period of time and is more stable across signers. But,because we know about the differences in morphosyntactic structure that can arise dueto its differences in use and because H2-Spread is obligatory, rather than optional, insome morphosyntactic uses, (as (4) and (5) show) we must be careful not to conflatedata sets that might be associated with distinct phenomena.

1.1.4 Positional factors in sign language prosodic constituencyWe also want to determine whether, in our corpus, certain types of prosodic structuresco-occurred with specific positions in a phrase or sentence. We know that thisbehavior, well-documented in spoken languages (see, for example, Zec & Inkelas

<LINK "bre-r21">

1990), occurs in ASL. Sentence-final position attracts phonologically heavy items(Brentari 1998), morphologically heavy items (Fischer & Janis 1990) and focused items

<LINK "bre-r4"><LINK "bre-r8">

(Wilbur 1999, 2000). In addition, IP-final position is called ‘strong’ in Nespor and

<LINK "bre-r20"><LINK "bre-r14">

Sandler (1999). New evidence suggests that forms in IP-final position may includeredundant, inserted material to fill this ‘heavy’ position by double articulation — thatis, signing the form and then fingerspelling the same form ((8); Shay 2002). Sandler

<LINK "bre-r19"><LINK "bre-r18">

Page 8: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

112 Diane Brentari and Laurinda Crossley

(1999a) and Miller (1996) observe that in ISL and LSQ, respectively, reduplication is

<LINK "bre-r12">

PP boundary marker, but in ASL this double-articulation occurs at IP boundaries.

(8) Phonological expansion in heavy positionUF: brow raise/topicH1: [[carrot, potato, lettuce all]PP ^ [don’t-like [vegetables

v-e-g]PW]PP]IP

‘Carrots, potatoes, lettuce, all of them. I don’t like vegetables.’

The two forms vegetables and v-e-g are two variants of the same ASL word. If oneaccepts that the heavy position is a position that allows for more phonological material,a possible explanation for why these redundant 2-sign pairs appear together in thisposition, and do not appear in other positions, is that the redundant form is insertedto fill a larger than normal word-level template that is available in this position. Theability to handle phonological heavy forms would also be a possible explanation forwhy forms that are being lexicalized would appear in this position. Forms about toundergo the lexicalization process are larger than the typical ASL word; they are largerthan the preferred monosyllabic template of ASL.7

1.2 Hypotheses

First, we hypothesize H2-Spread will be correlated with one or more nonmanualmarkers for prosodic constituency. Second, we hypothesize that by looking at naturallyoccurring data we will find more types of H1/H2 interaction, and particularly ofH2-Spread, than have been reported in the literature previously using elicited data.

We will argue that in ASL H2-Spread is not the primary marker of these constitu-ents, but rather H2-Spread is a redundant marker of prosodic cues found on the face.The findings reported here confirm some of the results of Nespor and Sandler (1999)

<LINK "bre-r14">

and Sandler (1999a,�b). We also report on two cases of H2-Spread not previously

<LINK "bre-r18">

discussed in the literature, which contribute to utterance meaning.

2. Methodology

The primary corpus for this study consist of a 60-minute, videotaped lecture in anatural setting by a native signer to a mixed audience of Deaf people and hearing peoplewho are ASL students. We target four suprasegmental phenomena — H2-Spread,cheek tension (LFT), inhibited involuntary eyeblinks, and lengthening — defined in (9).

7.�See Coulter (1993) and Brentari (1998) for evidence that the preferred word structure in ASL ismonosyllabic.

Page 9: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 113

(9a–b) have been previously established as markers for PPs and IPs. The dominanthand (H1) for our signers was determined observing the hand typically used by a givensigner for fingerspelling and for the articulation of 1-handed signs.

(9) Definitions of the four suprasegmental properties targeteda. Eyeblinks: These are inhibited eyeblinks, as described in Section 1.1.3, first

argued to be an IP marker for ASL (Wilbur 1994).

<LINK "bre-r20">

b. Lengthening: This is the extended duration of the final timing unit (eithersegment or mora) and/or movement at a PP boundary (Perlmutter 1992;

<LINK "bre-r16">

Miller 1996).

<LINK "bre-r12">

c. H2-Spread: An H2 handshape from a sign articulated on 1 or 2 hands isassimilated either regressively (anticipatory H2-Spread) or progressively(progressive H2-Spread) in the signing space, while the other hand contin-ues with the utterance

d. Lower Face Tension (LFT): This nonmanual behavior involves general flex-ing of a group of muscles at the corners of the mouth, which takes placeduring mouth gestures of all types (Crossley 2001).

<LINK "bre-r7">

First, we transcribed our data with respect to eyeblinks to determine the IP boundaries,and then, within IPs, we used lengthening to determine the intermediate or PPboundaries. Second, we identified instances of H2-Spread or dominance switch. Thetotal number of instances of both of these phenomena in the 60-minute sample was166. Then, we divided the occurrences of H2-Spread into those in which H2 carrieddistinct morphological information and cases where it did not. We defined a morpho-logical or morphosyntactic instance of H2-Spread as one in which the handshapeand/or location of the trigger sign had a meaningful referent, either in space as personor spatial agreement, or as a classifier handshape, or both. We defined a prosodicinstance of H2-Spread as one in which the handshape and/or location of the non-dominant hand had no morphological or morphosyntactic purpose, so that these casescould be investigated solely on the bases of the phonological/prosodic use. Here we areaddressing only those instances of H2-Spread that did not have a morphological ormorphosyntactic role. There were 84 instances classified as phonological and 82instances classified as morphosyntactic H2-Spread in this 1-hr. lecture. Only thephonological cases will the subject of investigation here; the morphosyntactic ones willbe addressed in future work. Third, we described what the lower face was doing duringthe instances of H2-Spread or dominance switch.

After our analysis of this first corpus, we confirmed it by performing the sameprocedures described above on a shorter, second ASL corpus by a native signer, andfound evidence of all four categories described here. The second transcription wasmade of a published source; these were the explanatory narratives given by Ella MaeLentz on the video “Treasures” (1995) of her own poems (not the poems themselves).

Page 10: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

114 Diane Brentari and Laurinda Crossley

The signers in both of our tapes are females of approximately the same age, but fromdifferent part of the US — the first is from Chicago; the second is from the Bay Areain California. We used SignStream™ (Neidle & MacLaughlin 1998) to transcribe the

<LINK "bre-r13">

data, in order to identify precisely the timing of these behaviors both alone and withrespect to one another within utterances.

3. Analysis

We found the distribution of H2-Spread to be across a broad range of prosodicconstituents, but that it can be nonetheless analyzed systematically, since H2-Spreadinteracts with inhibited eyeblinks and with nonmanuals of the lower face — inparticular, with tension in the cheeks and corners of the mouth. We found that, withinan IP, H2-Spread has two possible domains, which we summarize in (10). Theirdistribution confirms the findings of Nespor and Sandler (1999) and Sandler

<LINK "bre-r14"><LINK "bre-r18">

(1999a,�b); we add further evidence for their analysis by proposing facial cues thatdifferentiate the two.

We conclude that the behavior of H2 alone cannot be used as a primary index ofany constituent, but rather that it is redundant to the timing of facial expression on thelower face nonmanuals and eyeblinks.

3.1 Analysis of P-Word (PW) vs. P-Phrase (PP) Domains

In the work here, we found that the timing of H2-Spread, blinks, and LFT are used todistinguish single PWs from multiple PWs in a PP. Lengthening effects are not muchhelp here, because, since PW cases appear in the final position of an IP, they alsoexhibit lengthening. As you will see below, H2-Spread is not the primary phonetic cue,but it provides an important redundant cue.

3.1.1 The Prosodic Word (Total 30; 36% of H1/H2 interaction)One Prosodic Word/Coalescence is the combination of two morphosyntactic elementsthat typically occur as free morphemes (i.e. independent words) into a single phono-logical word. The definition of the PW use of H2-Spread is given in (10).

(10) One (P)rosodic-Word/Coalescencea. There is no change in LFT value between the trigger sign and the target sign

of H2 spread.b. There are no nonmanual changes during the H2 spread (i.e. eyegaze, head

tilt, blinks).c. The trigger of assimilation is H2 (i.e. not a switch of hand dominance).

Page 11: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 115

It follows the findings of Sandler (1999a,�b) in the following ways:

<LINK "bre-r18">

d. It occurs at the right edge of a IP.e. There is a maximum of two signs involved.f. H2-Spread is progressive.

The arguments for calling these forms PWs in ASL — rather than some other prosodicunit — come from disyllabic 2-handed signs, including both compounds (e.g.sleep^sunrise = oversleep) and other disyllabic monomorphemic forms (e.g.lock). First, when compounds in ASL include a lower-face gesture, this single gestureextends across both stems of the compound. There are no compounds with twodistinct lower face gestures (11).8

(11) Compounds with a single lower face gesture (each compound is a separate form)LF: (u)----------- (r)---------- (r)-------------signs: good^night male^same female^same

LF: (closing)------ (u)--------------signs: male^marry female^marry.

Second, in disyllabic, 2-handed signs there can be at most one distinctive H2 specifica-tion; forms demonstrating this are given in (12). The forms in (12a) are mono-morphemic, 2-handed forms. Using Battison’s 1978 typology of 2-handed signs, where

<LINK "bre-r1">

Type 1 and Type 2 signs have the same handshape, and Type 3 have different hand-shapes, the forms in (12a) are Type 3 signs. No type 3 sign in ASL has 2 different H2handshapes, which led to a re-analysis of H2 as a word-level appendix (Brentari, 1998)

<LINK "bre-r4">

rather than a syllable coda as was argued in previous work (Brentari and Goldsmith,

<LINK "bre-r4">

1993). The forms in (12b) are disyllabic compounds and “+er” nominals. In eachform, at least one of the signs is a Type 1 sign, which by definition has a redundant H2specification; therefore, only 1 of the 2 H2-handshapes is distinctive. The forms in(12c) are compounds, which contain two Type 3 signs, but both of signs happen tohave the same H2-handshape; so, again, there is only one distinctive H2 handshape(both signs nude and zoom; check and read have a b-handshape). Finally, in (12d),are forms that should be legitimate compounds semantically, but they are ungrammat-ical forms. They are ungrammatical, we would argue, precisely because they have twodistinctive H2 specifications.

(12) Distribution of H2 in disyllabic, 2-handed signsa. disyllabic monomorphemic words: background, curriculum, program,

project

8.�A “gesture” can be a steady state gesture, such as liprounding, or a single, dynamic gesture, suchas mouth opening or closing. Whenever possible, we are using IPA symbols to indicate the lipposture in use.

Page 12: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

116 Diane Brentari and Laurinda Crossley

b. polymorphemic forms with two redundant specifications (compounds and“+er” nominals: jesus^book (“bible”), word^book (“dictionary”),teacher, money^behind, server

c. polymorphemic forms with one distinctive H2 specification compounds:check^read, nude^zoom

d. ungrammatical compounds with two distinctive h2 specification:*word^help (“spell checker”), *check^vote (“recount”), *money^

enough (“savings”)

Since all of the forms in (12a–c) are single words, and since the domain of their LFvalue and domain of H2-Spread is the same as those forms at the right edge of an IP,we conclude that the unit in these H2-Spread cases at the end of an IP is also theprosodic word. The direction of assimilation for H2-Spread is regressive in compounds(Liddell & Johnson 1986), but the direction of assimilation in lexicalized compoundsmay occur later in the lexicalization process.9

Sandler (1999a,�b) calls these cases Coalescence in ISL, to distinguish them from a

<LINK "bre-r18">

more general description of H2-Spread at the level of the PP, but she reports thatCoalescence is restricted in two ways that we did not find in ASL. First, Sandleranalyzes Coalescence as a type of cliticization, since in ISL it appears that only locativeand pronominal index forms can appear in this context. In ASL, this is not the case; avariety of grammatical categories are combined, not only noun + index forms. Second,Sandler extends the use of Battison’s typology of 2-handed signs to ISL, and theyreport that in ISL only Type 1 signs can trigger Coalescence. In the ASL corpus, whilethe trigger sign of Coalescence is more commonly a 1-handed sign, Type 1 sign, Type2 or Type 3 signs can also trigger the spread of H2. A range of examples of PW uses ofH2-Spread, which do not conform the ISL data, are given in (13). In all of theseexamples, the LFT value is across both signs in the 2-sign combination.

(13) 2-sign combinations, analyzed as prosodic words at the R-Edge of IPs, citedwith the way that they deviate from the findings reported for ISL (Sandler1999a,�b).a. H1: word what (word is not a Type 1 sign; what is not a pronoun.)

H2: what

b. H1: depend where (depend is not a Type 1 sign; W where is not a pro-noun.)H2: depend

c. H1: role have (role is not a Type 1 sign; have is not a pronoun.)H2: role

9.�It has been suggested by John Kingston (personal communication) that this is precisely the wayto tell a lexicalized compound from a PW.

Page 13: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 117

d. H1: expression face (face is not a pronoun.)H2: expression

e. H1: parenthesis indian (indian is not a pronoun.)H2: parenthesis

f. H1: cold shoulder (shoulder is not a pronoun.)H2: cold

g. H1: hmm where (hmm is not a 2-handed sign; where is not a pronoun.)H2: hmm

A sentence containing a PW case of H2-Spread in is given below in (14). The H2 of theType 1 sign cold spreads progressively to the next sign shoulder (Figure 1).

(14) PW: “cold shoulder”LF: (o)--------------H1: polite-small-talk feel quote cold shoulder ^H2: cold(s-hs)‘It’s just polite small talk; you feel like you are getting the cold shoulder.’

polite small-talk

quote

feel

shouldercold

Figure 1.

Page 14: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

118 Diane Brentari and Laurinda Crossley

3.1.2 More than One PW in a PP (Total 32; 38% of H1/H2 interaction)Turning to non-final position in an IP, we find that in ASL the LFT value changesbetween the trigger and target sign(s), and the direction of assimilation can beregressive or progressive. The criteria for this type of H2-Spread are given in (15).

(15) Criteria for determining H2-Spread with a domain of more than PWa. There is a change in LFT value, and possibly other nonmanuals, between

the trigger and the target signs of H2-Spread.b. The number of signs involved has a minimum of 2 signs but the maximum

is specified by higher order prosodic structure.c. The trigger of assimilation is H2 (i.e. not a switch of hand dominance).d. Following Nespor and Sandler (1999), the direction of assimilation may be

<LINK "bre-r14">

regressive or progressive.

This occurs in the sequence from and c-o-d-a in (16). In this example, the H2-Spreadis the 1-handshape on H2 from the first sign from which spreads to the fingerspelledform c-o-d-a (Figure 2).

(16) PP: “from c-o-d-a”LF: (a)--- (o)-----H1: from c-o-d-a different story ^H2: from(1hs)‘From a CODA (child of Deaf parents) it is a different story.’

Page 15: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 119

from c-o-d-a

different story

Figure 2.

The test for PW status is whether there is one LFT value or not; PWs have one LFTvalue, more than one PW will allow more than one LFT value. Also, the PW context ismore restrictive than that of two PWs in a PP in terms of both the direction ofassimilation and in the number of signs across which H2 may spread. In our corpusPWs have a domain of two signs (including trigger and target), while PPs have adomain as large as four signs. Following Nespor and Sandler (1999), we expected that

<LINK "bre-r14">

the spread would be to the left or right edge of the PP, unless a two-handed sign blocksit. This was generally correct, except for the forward-referencing cases described in thenext section.

In sum, we found that it was not reliable to look for H2 behavior alone to dis-ambiguate these units; the prosodic cues on the face were more consistent. We foundthat H2-Spread alone cannot predict anything unless seen in light of the LFT value and

Page 16: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

120 Diane Brentari and Laurinda Crossley

position in the IP. In fact, there are instances of the PW LFT marking in phrase-finalposition that involve only 1-handed signs such as the one given in (17). The signer istalking about differences between Deaf and hearing senses of humor, and she signs sad

(blink) watch happen hit watch j-a-y l-e-n-o… (blink) (English: It’s sad, when Ihappen to watch Jay Leno …) well watch [protracted] not funny, at the right-edgeof the IP has the same LFT value, and hence, under this analysis would count as asingle PW.

(17) 1-handed-sign combination of a PW: not-funny

LF: (r)---------H1: well watch[protracted] not funny ^‘Well, I watch it and it’s not funny [to me].’

In (18) we see an example of a 2-sign combination in which both signs are 1-handed— you-know indian. Despite the fact that this form is phrase-final (in this sentencethis two-sign combination is a “topic structure”, Padden 1983) there are two LF

<LINK "bre-r15">

gestures; i.e., the nonmanuals change between the trigger and target signs. Thisexample makes two points: first, not all 2-sign combinations at the right edge of an IPare a single PW, and, second, the way one can tell consistently between them is not bylooking at H2 behavior, but by looking at their nonmanual behavior.

(18) 1-handed-sign combination in a PP: you-know indian

UF: brow raise/topicLF: (o)----------- (a)------ (o)-- (a>p)H1: know-that indian ^ like old habit sign indian ^

‘You know that, as for the sign “Indian”, it’s like an old habit to sign ‘Indian’.’

To conclude, we confirmed the findings of Nespor and Sandler (1999) that there are two

<LINK "bre-r14">

uses of H2-Spread — a PW use and one in a larger PP domain. We found that in ASL,however, PW cases are not restricted in the types of signs that might be involved; wealso found the nonmanual and positional factors are more predictive than H2-Spread.

4. Analysis H2-Spread for Utterance Meaning

In this section we will analyze two cases of H2-Spread that contribute specific mean-ings to their utterances, following the work of Pierrehumbert and Hirschberg (1990).

<LINK "bre-r17">

The timing of H2-Spread and eyeblinks provides information about the prosodiccontours of the utterance that contribute to meaning. We have identified parentheticaluses of H2-Spread and forward referencing uses of H2-Spread. As expected, since theseare units that are larger than the prosodic word, the value for LFT changes during theexecution in all of these instances of H2-Spread. We found that these occur far less

Page 17: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 121

frequently than the cases discussed in the previous section. Taken together, PW and PPcases make up 74% of the total, while instances meaningful use of H1/H2 interactionare only 26% of the total.

4.0.1 Forward Referencing (Total 18; 21% of H1/H2 interaction)“Forward Referencing” is a way to show the relatedness of two adjacent IPs (Pierre-

<LINK "bre-r17">

humbert & Hirschberg 1990). H2-Spread crosses the right IP boundary into the firstsign of the next IP. In these cases, and in these cases it serves to mark a strong semanticconnection between the two IPs. Our analysis of forward-references follows that ofPierrehumbert and Hirschberg (1990), where “forward referencing” is defined as

<LINK "bre-r17">

signaling a close semantic link in a second IP by a prosodic cue in the first IP. This isshown in (19). In (19a) the two IPs are semantically linked, while in (19b) they are not.A high boundary tone (transcribed as a H%, shown in bold) indicates that the secondIP about to be uttered in (19a) is closely linked to the one just uttered.

(19) Forward referencing in English (from Pierrehumbert & Hirschberg 1990)

<LINK "bre-r17">

a. The train leaves at seven It’ll be on track fourH* H* H* L H% H* H* L L%

b. The train leaves at seven There’s a full moon tonight.H* H* H* L L% H* H* H* L L%

The ASL case, however, is much more like Swedish accent patterns as described inBruce (1977), where intonational markers can be systematically shifted within a phrase

<LINK "bre-r5">

and need not line up cleanly with the predicted stressed syllables which usually areresponsible for carrying such accents. This type of misalignment is analogous to theASL case; that is, it is precisely because the H2 handshape shifts one word to the rightof the normal boundary of forward referencing that these cases mark themselves outas special. The criteria for forward referencing in ASL are given in (20), examples aregiven in (21)–(22). Because forward referencing is a category of H2-Spread that hasnot been documented before, we include two examples. One is from our primarycorpus, and one is from our secondary corpus.

(20) Criteria for a “forward referencing” instance of H2-Spreada. The trigger is in the first of two IPs, determined by an eyeblink, and the

target is the first sign of the second IP in the sequence.10

b. The trigger is H2 (i.e. no dominance switch)c. The direction of assimilation is progressive.

10.�The limit we place on the spreading of H2 in these cases is tentative. The spreading stops at anemphatic 2-handed sign in all of the cases we have from our data. More examples of this would needto be collected to confirm this limit.

Page 18: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

122 Diane Brentari and Laurinda Crossley

(21) Forward-referencing: “distribute-all-over (blink) index1sg” (Figure 3)Body lean: (left)--------------------------------- (right)----------------H1: index3SG distribute-all-over ^ index1SG darn ^H2: distribute (5hs)‘She distributed (the advertisement) all-over. (There I was,) “Darn!”’ [It had anerror.]

index3SG distribute

all-over blink

index1SG darn

Figure 3.

Page 19: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 123

(22) Forward referencing (ex. from Lentz 1995): “language (blink) wow” (Figure 4)

<LINK "bre-r11">

H1: english asl different[int] language ^ wow mind-blowout ^H2: english different[int] language mind-blowout

‘English and ASL are really different languages; I was really blown away.’

english ASL different

languages (1st position)languages (2nd position) [blink]

wow mind blow-out

Figure 4.

Note that in both examples, the opportunity to continue the assimilation of H2 furtherinto the second IP is possible. In (21) and (22), darn and mind-blowout, respective-ly, the second sign of the second IP is a Type 1 sign that could easily have surfaced asa 1-handed variant via the operation of Weak Drop (Padden & Perlmutter 1987)

<LINK "bre-r15">

thereby allowing H2 to spread, but it did not. This shows that there may really be alimit on H2-Spread for forward referencing (cf. note 10).

Page 20: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

124 Diane Brentari and Laurinda Crossley

4.0.2 Parentheticals (Total 4; 5% H1/H2 interaction)The second type of meaningful use of H1/H2 interaction, and the fourth type we found inour corpus, was the parenthetical remark. During a signed utterance, a comment that isof secondary importance is interjected into the immediate topic of conversation. It ismarked by a “dominance shift” (Frishberg 1985); that is, the nondominant hand, H2,

<LINK "bre-r10">

signs the parenthetical remark and the handshape held is that of the typical dominanthand, H1. The parenthetical portion forms its own IP within the matrix sentence. Theunits on either side of the parenthetical unit also form IPs. The signer will changedominance to indicate an abrupt shift in the main flow of a conversation topic, whilecontinuing to sign the remaining utterance with their dominant hand. Frishberg foundthat “in that case, the DR[dominance relation] set off material as background to themain force of the narrative”. The example taken from Frishberg (1985) is repeated here

<LINK "bre-r10">

in (23). The criteria for an ASL parenthetical are given in (24).

(23) Dominance Shift (from Frishberg 1985; repeated from (3))

<LINK "bre-r10">

H1: army rush-at-us embarrassed why girl five boy five

H2: only-one-me

‘The soldiers came right at us. I was embarrassed because — I was alone — agirl with five guys.’

(24) Criteria for a “parenthetical” instance of H2-Spreada. Dominance switch (Frishberg 1978);

<LINK "bre-r10">

b. The handshape that spreads is the typical H1-handshape;c. Inhibited eyeblinks at the left and right edges of the remark;d. The direction of assimilation is progressive;e. The strings that frame the parenthetical must be isomorphic to syntactic

constituents, or, minimally, they have the same subject and the same pointof view.

In our corpus, parentheticals were not marked only by a dominance shift, but also bythe spread of (the usual) H1 across the parenthetical remark signed by (the usual) H2for a given signer. Inhibited eyeblinks in parentheticals occur at the left and right edgesof the string, marking it as an independent IP embedded within a matrix sentence.Nespor & Vogel (1986) argue that parentheticals can be identified “when a root sentence

<LINK "bre-r14">

has an intervening obligatory IP”. The clauses that frame the parenthetical in ASL had thesame subject and the same point of view. We were unable to say definitely whether thetwo clauses framing the parenthetical are a single sentence or not, but the behavior andfunction of parentheticals was consistent and systematic in the use of dominance shiftand H2-Spread, subject, and point of view. An example is shown in (25), in whichmisunderstood is signed by both hands, then index1SG send 20 (‘I sent her the 20th’)is signed by H2, which we analyze as parenthetical to the rest of the utterance, misunder-

stood … think 28 (‘She misunderstood … she thought the 28th’) (see Figure 5).

Page 21: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 125

(25) Parenthetical: “index1SG send 20”11

H1: misunderstood ^ ^ think 28H2: misunderstood index1SG send 20‘(She) misunderstood. I sent the 20th. (She) thought the 28th.’

misunderstood blink index1SG

send (1st position) send3SG (2nd position) 20

blink think3SG 28

Figure 5.

11.�Recall that the blinks mark all IPs, not just the boundaries of a parenthetical.

Page 22: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

126 Diane Brentari and Laurinda Crossley

5. Discussion and Conclusion

Our main finding was that, while H2-Spread can determine nothing on its own. InASL, it has too many different (albeit systematic) uses to definitively determine anyprosodic category. In fact our data supports a result where the facial cues are primary,and H2-Spread is an associated, redundant behavior, as the one handed examples in(12) and (13) have shown. The relevant nonmanual prosodic cues are present whetherthe sentence contains 2-handed signs or not. However, H2-Spread is a manual, salientcue to prosodic structure, one that is expected to have cross-linguistic significance,since the nonmanual behaviors are more variable and could possibly make cross-linguistic comparisons more difficult. The cues on the face and hands must beexamined individually, and then the temporal relationships among them uncovered,if we are to find the prosodic patterns in sign languages.

5.1 The Elusive Phonological Phrase

Nespor and Sandler (1999) attribute the instances of H2-Spread that are not Coales-

<LINK "bre-r14">

cence to the domain of the PP, since they have found no instances of spreading beyondthe PP in ISL and have found cases of where H2-Spread does not extend to the PPboundary due to the presence of a 2-handed sign that would block such spreading.12

Sandler (1999) also reports the following independent PP boundary markers in ISL: (a)slower movement and (b) lengthening of the final hold to be PP markers in ISL (c)optional iteration/reduplication.

In ASL, we have found only lengthening to be a reliable cue for PP boundaries.Repetitions/iterations of form, or even the cases of double-articulation of the sameword with different variants (e.g. vegetables+v-e-g) occur consistently at IPboundaries in ASL. While our data unequivocally supports a PW domain for thespecific type of H2-Spread referred to by Sandler as Coalescence in ISL, it is lessdefinite in its support of a PP domain for H2-Spread. We have cases of spread beyondthe PP boundary and also cases that fail to spread to the PP boundary when expected;the forward referencing data are evidence of this. The H2 handshape spreads across anIP boundary, and then stops after the first sign. If one considers these cases to have aspecial status because of their discourse-level meaning, these cases may only be only anapparent set of counter-examples.

We conclude that, alone, H2-Spread is not a reliable marker for any constituent inASL, and that the nonmanual behaviors on the face are more important in this regard.

12.�It is not entirely clear what this means, since some 2-handed signs (notably Type 1 signs) areallowed to drop H2 and would then be available for H2-Spread.

Page 23: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 127

If our results are correct, we can propose a set of criteria for establishing PWs, PPs, andIPs independent of H2, given in (26).

(26) Criteria for PW, PP, and IP prosodic constituents in ASL

lower face/LFT lengthening blinks

P wordP PhraseI Phrase

no changechangeNA

noyesyes

nonoyes

5.2 Theoretical implications

This work has important implications for diachronic change, as well as for the study ofprosodic structure in signed and spoken language. Here we see that PW formationexhibits important positional and lower face cues. If there can only be one LF value inPW, then this nonmanual aspect of the phonology conforms to the set of constraintson the other parameters of ASL; namely, as a form becomes lexicalized, there will beone contrastive value per parameter (i.e. handshape, movement and place of articula-tion). These data show that in considering the processes involved in lexicalization insign languages, therefore, one must consider nonmanual behaviors, as well as thosemanual behaviors, such as those discussed in Brentari and Padden (2001). One might

<LINK "bre-r4">

speculate that as polymorphemic forms become lexicalized as single words, thetendency is toward single values for all phonological parameters, both manual andnonmanual. This tendency toward a single specification has been well documented forthe manual properties of handshape (Friedman 1978), for place of articulation

<LINK "bre-r9">

(Sandler 1987), for H2 (Brentari 1998) and for movements (Coulter 1982).

<LINK "bre-r4"><LINK "bre-r6">

This work also highlights where one might expect to find such forms-in-transitionin prosodic structure in ASL — i.e. at the right edge of an IP. As discussed in Sec-tion 1.1.4, the right-most sign of a sentence is the site for focus and for morpho-phonological weight. A short sentence or clausal unit can be an IP. A question forfuture research might be to investigate the position of forms, which are not yet fullylexicalized, but rather exist in the peripheral strata of the ASL lexicon with respect toword-level constraints, such as the sign + fingerspelled compounds discussed inBrentari and Padden (2001). One might speculate that these forms would exhibit the

<LINK "bre-r4">

same distribution within utterances or behaviors on the lower face similar as the PWcases analyzed here. Two-sign combinations that appear in this position may stillcurrently be nonce forms, but they are candidates for further lexicalization operations,such as compounding or nativization of a fingerspelled form (Battison 1978; Brentari

<LINK "bre-r1"><LINK "bre-r4">

& Padden 2001). When focus is not a factor, the LFT behavior and positional use maybe early indicators that lexicalization may have begun.

Page 24: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

128 Diane Brentari and Laurinda Crossley

Our work on sign language can also contribute methodologically to studies ofprosody in spoken languages. First, it emphasizes that a prosodic constituent may beidentifiable by a constellation of behaviors, and the timing relationships among thesebehaviors are important clues to structure. Second, even though pauses, tunes, andtone sequences have been considered the primary means of identifying prosodicconstituents in English, for example (Pierrehumbert 1980, and the generation of work

<LINK "bre-r17">

thereafter), the body may also be important in signaling prosodic constituency. Acompositional approach to prosodic analysis need not be an analysis of solely sequen-tial acoustic behaviors (such as tone sequences), but can include nonverbal behaviorsoperating in parallel. Eyeblinks and facial postures, such as head tilts and eyegaze, aswell as manual gestures in spoken languages may be useful in work on prosody inspoken languages, as well.

Acknowledgments

We would like to acknowledge Drucilla Ronchen for her assistance on this project. This material isbased upon work supported by the National Science Foundation under grant no. 9905848.

References

Battison, R. (1978). Lexical borrowing in American Sign Language. Silver Spring, Md.: Linstok Press.

<DEST "bre-r1">

Bergman, B. & L. Wallin. (2001). “The discourse function of noun classifiers in Swedish Sign

<DEST "bre-r2">

Language.” In V. Dively, M. Metzger, S. Taub & A. Baer (eds.), Sign Languages: Discoveries frominternational research, pp.45–61. Washington, DC: Gallaudet University Press.

Boyes Braem, P. (2001). “Functions of the mouthing component in Swiss German Sign Language.”

<DEST "bre-r3">

In D. Brentari (ed.), Foreign vocabulary in sign languages, pp.1–45. Mahwah, NJ: LawrenceErlbaum Associates.

Brentari, D. (1998). A prosodic model of sign language phonology. Cambridge, MA: MIT press.

<DEST "bre-r4">

Brentari, D. (1990) Theoretical foundations of American Sign Language phonology, Doctoral disserta-tion, University of Chicago. [Published 1993, University of Chicago Occasional Papers inLinguistics, Chicago, Illinois.]

Brentari, D. & C. Padden (2001). “A language with multiple origins: Native and foreign vocabularyin American Sign Language.” In D. Brentari (ed.), Foreign vocabulary in sign language: Acrosslinguistic investigation of word formation, pp.87–119. Mahwah, NJ: Lawrence ErlbaumAssociates.

Brentari, D. & J. Goldsmith (1993). “Secondary licensing and the nondominant hand in ASLphonology.” In G. Coulter (ed.), Phonetics and phonology, Vol. 3: Current issues in ASLphonology, pp.19–41. New York: Academic Press.

Bruce, G. (1977). Swedish word accents in sentence perspective. Doctoral dissertation, Lund University.

<DEST "bre-r5">

Coulter, G. (1982). “On the nature of ASL as a monosyllabic language.” Presented at the annual

<DEST "bre-r6">

meeting of the Linguistic Society of America, San Diego, California.

Page 25: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

Prosody on the hands and face 129

Crossley, L. (2001). Sign language prosody on the face and hands: H2-Spread and prosodic constituents.

<DEST "bre-r7">

MA thesis, Purdue University.Fischer, S. & W. Janis (1990). “Verb sandwiches in American Sign Language.” In S. Prillwitz & T.

<DEST "bre-r8">

Vollhaber (eds.), Current trends in European sign language research, pp.279–294. Hamburg:Signum Verlag.

Friedman, L. (1976). Phonology of a soundless language: phonological structure of American Sign

<DEST "bre-r9">

Language. Doctoral dissertation, University of California, Berkeley, California.Frishberg, N. (1985). “Dominance relations and discourse structures.” In W. Stokoe & V. Volterra

<DEST "bre-r10">

(eds.), SLR’83: Sign language research, pp.79–90. Linstok Press, Inc. and CNR, Rome.Lentz, E. (1995). Treasure. In Motion Press. Berkeley, CA.

<DEST "bre-r11">

Miller, C. (1996). Phonologie de la langue des signes québécoise: Structure simultanée et axe temporel.

<DEST "bre-r12">

Doctoral dissertation, Université du Québec á Montreal.Neidle, C. & D. MacLaughlin. (1998). “SignStream: A tool for linguistic research on signed languag-

<DEST "bre-r13">

es.” Sign Language & Linguistics 1(1): 111–114.Nespor, M. & I. Vogel (1986). Prosodic phonology. Dordrecht: Foris Publications.

<DEST "bre-r14">

Nespor, M. & W. Sandler (1999). “Prosody in Israeli Sign Language.” Language and Speech 42:143–176.

Padden, C. (1983) Interaction of morphology and syntax in American Sign Language. Doctoral

<DEST "bre-r15">

dissertation, University of California, San Diego. [Published 1988, Garland Press, New York.]Padden, C. & D. Perlmutter (1987). “American Sign Language and the architecture of phonological

theory.” Natural Language and Linguistic Theory 5: 335–375.Perlmutter, D. (1992) “Sonority and syllable structure in American Sign Language.” Linguistic

<DEST "bre-r16">

Inquiry 23: 407–442.Pierrehumbert, J.B. (1980). The phonology and phonetics of English intonation. Doctoral dissertation,

<DEST "bre-r17">

MIT.Pierrehumbert, J.B.& J. Hirschberg (1990). “The meaning of intonational contours in discourse.” In

P. Cohen, J. Morgan & M. Pollack (eds.), Intentions in communication, pp.271–311. Cambridge,MA: MIT press.

Sandler, W. (1999a). “The medium is the message: Prosodic interpretation of linguistic content in

<DEST "bre-r18">

sign language.” Sign Language and Linguistics 2: 187–216.Sandler, W. (1999b). “Cliticization and prosodic words in a sign language.” In T. Hall & U.

Kleinhenz (eds.), Studies on the phonological word, pp.223–254. Amsterdam: John Benjamins.Sandler, W. (l989). Phonological Representation of the Sign. Dordrecht: Foris.Sandler, W. (1987). “Assimilation and feature hierarchy in American Sign Language.” Papers from the

23rd annual Meeting of the Chicago Linguistic Society, Vol. 2. Parasession on autosegmental andmetrical phonology, pp.266–278. Chicago: Chicago Linguistic Society.

Shay, R. (2002). Grammaticalization and lexicalization in ASL fingerspelling. MA thesis, Purdue

<DEST "bre-r19">

University.Wilbur, R. (2000). “Phonological and prosodic layering of nonmanuals in American Sign Language.”

<DEST "bre-r20">

In H. Lane & K. Emmorey (eds.), The signs of language revisited: Festschrift for Ursula Bellugi andEdward Klima, pp.213–241. Hillsdale, NJ: Lawrence Erlbaum.

Wilbur, R. (1999). “Stress in ASL: Empirical evidence and linguistic issues.” Language and Speech 42(2–3): 229–250.

Wilbur, R. (1994). “Eyeblinks and ASL phrase structure.” Sign Language Studies 84: 221–240.Wilbur, R.B. & C.G. Patschke (1998). “Body leans and the marking of contrast in American Sign

Language.” Journal of Pragmatics 30: 275–303.

Page 26: Prosody on the hands and face - Purdue Universitywilbur/Wilbur EALing/Brentari and Crossley 2002.pdfhere: Single Prosodic Word, Multiple Prosodic Words in an Intermediate Phrase, Parenthetical,

130 Diane Brentari and Laurinda Crossley

Zec, D. & S. Inkelas (1990) “Prosodically constrained syntax.” In S. Inkelas & D. Zec (eds.), The

<DEST "bre-r21">

phonology-syntax connection, pp.365–378. Chicago: University of Chicago Press.Zimmer, J. & C.G. Patschke. (1990). “Determiners in ASL.” In C. Lucas (ed.), Sign language research:

<DEST "bre-r22">

Theoretical issues. Washington, DC: Gallaudet University Press.

Authors’ addresses

Dr. Diane BrentariPurdue UniversityLinguistics ProgramHeavilon Hall500 Oval DriveW. LAFAYETTE IN 47907–2038USA

</TARGET "bre">