The Grammar of Headshake:
Sentential Negation in German Sign Language
Roland Pfau (Universiteit van Amsterdam)
1 Introduction
Negation in natural languages comes in many different shapes and extensive research
has been done on typological variation in the realization of sentential negation
(e.g. Horn 1978; Dahl 1979; Payne 1985). To the best of my knowledge, however,
the available studies are dealing with spoken language negation only. Given
that sign languages are natural languages and given that – despite the
use of a different language modality – they have been shown to share many
interesting properties with spoken languages (see e.g. Sandler (2000) for phonology,
Aronoff et al. (2000) for morphology/morphosyntax, and Neidle et al. (2000)
for syntax), the question arises: Do sign languages fit into the proposed typological
scheme?
In the following, I shall focus on German Sign Language (Deutsche Gebärdensprache:
DGS). Sentential negation in DGS (as well as in other sign languages) is particularly
intriguing because it involves the combination of a manual and a non-manual
signal, namely the manual negation sign NICHT ‘not’ and a headshake
which is associated with the predicate of the sentence. Despite this peculiarity,
I am going to show that DGS fits very well into the typological scheme proposed
for spoken languages and I will therefore conclude that the proposed typology
is modality-independent.
2 A typology of spoken language negation
According to the comprehensive typological studies by Dahl (1979) and Payne
(1985), there are three ways of expressing sentential negation in spoken languages:
negative particles, negative affixes, and negative auxiliaries.
Independent negative particles find use e.g. in Dutch. In Dutch matrix clauses,
the particle niet ‘not’ usually appears in sentence-final position
(1b).
(1) a. Ik begrijp het antwoord b.
Ik begrijp het antwoord niet
I understand the answer
I
understand the answer not
‘I understand the answer.’
‘I
don’t understand the answer.’
Turkish, on the other hand, belongs to the class of languages with affixal/morphological
negation. As can be seen in (2b), the negative suffix –mu attaches to the
verb stem.
(2) a. Almanya-da otur-uyor-um b.
Almanya-da otur-mu-yor-um
Germany-LOC live-PRES-1.SG
Germany-LOC
live-NEG-PRES-1.SG
‘I
live in Germany.’ ‘I
don’t live in Germany.’
Negative auxiliaries shall not concern us here. What is of importance, however,
is the fact that in some languages, negation comes in two parts. This phenomenon
is usually referred to as ‘split negation’. The best-known language
of this type is probably French where the two Neg particles ne and pas embrace
either the modal or the full verb (3b).
(3) a. Il lit le livre b.
Il ne lit pas le livre
he read.3.SG ART book
he
NEG read.3.SG NEG ART book
‘He reads the
book.’ ‘He
doesn’t read the book.’
Another widespread option for the realization of split negation is the combination
of a negative particle with a negative affix. In Ewe, a Western Sudanic language
spoken in Togo, for instance, a negative prefix attaches to the verb stem and
a negative particle appears in sentence-final position (4b) (Bole-Richard 1983:307).
(4) a. Kç$kú sà-nà sìgâ b.
Kç$kú mú-sà-nà sìgâ ò
Kokou sell-HAB
cigarette Kokou
NEG-sell-HAB cigarette NEG
‘Kokou sells
cigarettes.’
‘Kokou does not sell cigarettes.’
The above examples make clear that natural language negation comes in many different
shapes, i.e. as an independent particle or an affix, as simple or as split negation.
In the next section, I am going to consider if and how DGS negation fits into
this picture.
3 Sentential negation in German Sign Language
As mentioned before, sentential negation in DGS is particularly interesting
because it involves a manual sign as well as a non-manual element, i.e. a negative
headshake. In this section, I will propose a modality-independent account for
negation marking in DGS, which involves affixation of a featural affix.
3.1 The data
DGS is a SOV language. The manual negation element NICHT ‘not’ is
one of the very few signs that may follow the verb. As is illustrated by the
example in (5b), NICHT is optional and appears in sentence-final position. The
non-manual element is an obligatory side-to-side headshake (hs) which is performed
simultaneously with the verb sign. Similar to many spoken languages, DGS does
not have a copula verb. Consequently, in sentences with adjectival or nominal
predicates, the negative headshake accompanies the adjective or the noun (5d).
hs
hs
(5) a. FRAU BLUME KAUF b.
FRAU BLUME KAUF (NICHT)
woman flower buy woman
flower buy.NEG (NEG)
‘The woman
buys a flower.’ ‘The
woman does not buy a flower.’
hs
hs
c. MEIN BRUDER ARZT d.
MEIN BRUDER ARZT (NICHT)
my brother doctor
my
brother doctor.NEG (NEG)
‘My brother is
a doctor.’ ‘My
brother is not a doctor.’
The combination of an optional manual element with a non-manual component is
a recurrent pattern in the expression of sentential negation found across many
sign languages. The phonological form as well as the position of the manual
element within the clause, however, may differ. For instance, in American Sign
Language (ASL), which – according to Neidle et al. (2000) – has a
basic SVO word order, the word order in a negative clause is S-Neg-V-O while
in DGS it is S-O-V-Neg (cf. Pfau (2002) for a syntactic account of the differences
between ASL and DGS). Moreover, the exact location and the stretch of non-manual
marking may obey language-specific constraints. Finally, as Zeshan (2002) points
out, the realization of the non-manual grammatical marker is influenced by culture-specific
factors. In particular, in the Eastern Mediterranean area (e.g. Greek SL, Turkish
SL, and Lebanese SL), instead of a headshake, a negative headnod may find use.
I assume that the DGS sign NICHT is a particle. The fact that it appears in
sentence-final position reminds of the Ewe example in (4b); the fact that it
is optional reminds of the situation in colloquial French where the Neg element
ne is usually dropped. But what about the headshake? Apparently, this non-manual
element confronts us with a situation that is at odds with the typological patterns
presented before.
3.2 Prosodic alteration by featural affixation
In a number of studies, it has been convincingly shown that affective non-manual
markers (which, of course, may also accompany spoken utterances) are clearly
distinct from linguistic non-manual markers with respect to scope/timing (Baker-Shenk
1983), intensity/continuity (Reilly & Anderson 2002) and hemispheric localisation
(Corina et al. 1999).
Recently, sign language research has begun to seriously investigate the claim
that many instances of non-manual markings, in particular facial articulations,
may best be understood as fulfilling the role of intonation. That is, facial
articulation may serve a prosodic function in sign language and can be compared
to intonational contours in spoken language (Sandler 1999; Nespor & Sandler
1999; Wilbur 1999, 2000).
Following this line of research, I propose to analyze the negative headshake
associated with the predicate as a prosodic alteration imposed on a base form.
The negative headshake which is realized simultaneously with the manual string
can be analyzed as an autosegment, which behaves in a way similar to tonal prosodies
in tone languages. As is well known, tonal prosodies, too, may not only serve
to distinguish otherwise identical lexical items; they may also have a morphological
function by supplying e.g. aspectual or agreement information.
The prosodic change that accompanies negation in DGS can be accounted for in
a straightforward way if we think of the non-manual Neg element as a featural
affix which gets associated with a base form. I adopt the term ‘featural
affixation’ from Akinlabi (1996) who presents intriguing spoken language
data in which free (floating) features function as morphemes. The most commonly
found cases are those which involve a tone change but Akinlabi also presents
data in which non-tonal features function as grammatical morphemes (e.g. [+nasal]
expressing the category of first person in the Brazilian language Terena).
In very much the same way, the feature [headshake] in DGS can be interpreted
as an autosegment representing the Neg morpheme. As in spoken languages, the
featural affix must be associated with a base (here: with the predicate) in
order to be prosodically licensed, that is, the headshake must be accompanied
by manual material (but cf. Dively (2001) for ‘nonhanded signs’ in
ASL). This association is illustrated in (6). Note that a sign is usually taken
to consist of a sequence of locations (L) and movements (M), with LML constituting
the maximal syllable (Sandler 1989; Perlmutter 1992).
(6) Negation Negation
|
|
[headshake] [headshake]
[L M L]Pred [L M L]Pred
To sum up: The above discussion of the properties of sentential negation in
DGS shows that, from a typological point of view, DGS can be compared to languages
such as Ewe in that it shows split negation with one Neg element being a particle
and the other one an affix. In contrast to Ewe, however, the negative affix
is featural in nature and behaves in a way similar to tonal prosodies and other
featural affixes in spoken languages.
4 Prosodic alterations in spoken language negation
In this section, I am going to present examples from two spoken languages, which
I take to parallel the DGS examples presented before in that negative marking
– be it simple or split negation – is accompanied by a prosodic change.
In Gã, a Western Sudanic language spoken in Ghana, the realization of
negation on the verb crucially depends on the tense specification of the sentence.
In the perfect tense, the low-toned negative suffix -kò is used and moreover,
a tone change from low to high is observed in the verb stem (7b). Even more
intriguing, however, is the past tense pattern. In the past tense, there is
no visible negative suffix. It is only the shape of the verbal stem that is
altered by lengthening the final vowel and by raising its tone (7d) (Ablorh-Odjidja
1968:60ff).
(7) a. Mí-yè níì mómó b.
Mí-yé-kò nóko#
1.SG.PERF-eat meal
already 1.SG.PERF-eat-NEG
something
‘I have already eaten my
meal.’ ‘I
have not eaten anything.’
c. Mí-gbè gbèé
ko# d.
Mí-gbée gbèé ko#
1.SG.PAST-kill
dog ART 1.SG.PAST-kill.NEG
dog ART
‘I killed a
dog.’ ‘I
did not kill a dog.’
Obviously, negation is realized by a prosodic/phonological change only in the
Gã past tense; no particle or visible affix is involved. A similar pattern
is found with some verbs in Ógbrû, a Kwa language spoken in the
Southern Ivory Coast. According to Mboua (1999), the negative marker in Ógbrû
is a discontinuous morpheme characterized by a high tone featural affix and
the negative particle mú, which is subject to vowel harmony. In (8b),
the high tone gets associated with the aspectual morpheme that intervenes between
the subject and the verb while the particle follows the verb. Note, however,
that the negative particle never appears in sentences with monosyllabic verbs.
This is due to a tonal constraint which prevents the appearance of three successive
high tones. Consequently, in (8d) – as in the Gã example (7d) above
– negation is realized by a prosodic change only (Mboua 1999:15f).
(8) a. Kirî ò búkù òkókò
Kéré
ASP ask.for.RES banana
‘Kéré
has asked for the banana.’
b. Kirî ó búkù
mú òkókò
Kéré
ASP.NEG ask.for.RES NEG banana
‘Kéré
has not asked for the banana.’
c. Kirî à pá
òkókò
Kéré
ASP buy.RES banana
‘Kéré
has bought bananas.’
d. Kirî á pá
(*mˆ€) òkókò
Kéré
ASP.NEG buy.RES (NEG) banana
‘Kéré
has not bought bananas.’
The above examples exemplify that prosodic modifications are also observed in
some spoken languages in the context of negation (cf. Pfau (2002) for further
examples). From a typological point of view, I take the Gã and Ógbrû
examples to be as close to the DGS examples as one can get when comparing languages
in different modalities. Remember that the negative particle in DGS is optional.
When the particle is used, the DGS examples parallel the Ógbrû
example in (8b), i.e. we observe split negation with one Neg element being a
negative particle and the other one being a featural affix triggering a prosodic
change. When the particle is dropped, the situation is similar to the one in
the Gã past tense (7d) and the Ógbrû example (8d) in that
negation is realized by a prosodic modification alone.
5 Accounting for non-manual spreading
Things get more complicated when we take into account that sentential negation
in DGS may also be expressed in a slightly different way. In the DGS examples
(5bd) above, the negative headshake was indicated as being associated with the
predicate sign only (note that the sign NICHT is lexically marked for a headshake).
It is, however, possible for the headshake to spread onto neighboring constituents,
for example onto the direct object BLUME ‘flower’, as in (9a). Note
that spreading of the headshake is syntactically constrained; it is not possible
for the headshake to spread onto parts of phrases, as is illustrated by the
ungrammaticality of example (9b) in which the headshake is associated with the
adjective ROT ‘red’ only.
hs hs
(9) a. FRAU BLUME KAUF b.
* FRAU BLUME ROT KAUF
woman flower
buy.NEG woman
flower red buy.NEG
‘The woman does not
buy a flower.’ ‘The woman doesn’t buy
a red flower.’
Please note that at this point, I am not certain about how far exactly the negative
headshake can spread. In earlier work, I assumed that - in contrast to what
has been reported for other sign languages – the headshake cannot spread
over the subject NP (i.e. it cannot spread over the entire sentence). Here I
would like to tentatively claim that the spreading domain is the phonological
phrase domain and that subject NPs are topics and may therefore constitute a
phonological phrase of their own (cf. Nespor & Vogel 1986; Sandler 1999).
In (9a), a prosodic feature associated with the verb has obviously spread onto
a neighboring constituent. Since I have claimed before that the negative headshake
behaves in a way similar to tonal prosodies in tone languages, I shall now consider
if prosodic features in spoken languages can also spread across word boundaries.
In fact, they can. In the literature, the relevant phenomenon is ususally referred
to as external tone sandhi. Here, I will only give one representative example
from Tsonga, a Bantu language spoken in Mozambique and South Africa. Baumbach
(1987) observes various instances in which a high tone preceding a word with
only low tones spreads onto all syllables of this word except for the last one.
One particularly interesting case is that of an object with low tones only following
a high tone verb. In (10bd), the first two syllables of the objects xìkòxà
‘old woman’ and nhwànyànà ‘girl’ receive
high tones due to progressive high tone spreading (sites of tone change are
underlined).
(10) a. xìkòxà
b. Vá pfúná xíkóxà.
old.woman
they
help old.woman
‘They
help the old woman.’
c. nhwànyànà d.
Ú rhándzá nhwányánà.
girl
he likes girl
‘He
likes the girl.’
I therefore claim that optional spreading of the headshake in DGS – as
exemplified by (9a) – is an instance of prosodic feature sandhi comparable
to external tone sandhi in spoken languages. Still, there are important differences.
Note that in spoken languages, every tone-bearing unit must bear a tone and
every tone must be associated with a tone-bearing unit; that is, no vowel can
be articulated without a certain tone value. Due to this restriction, spreading
of tone requires repeated delinking or change of a tone feature (Odden 1995).
But this does not hold for the sign language data, since skeletal positions
(the L and M slots in (6) above) in DGS are for the most part not inherently
associated with a prosodic feature, say, a headshake. For this reason, the spreading
of the non-manual in DGS does not imply delinking or feature change; rather,
a feature is added to the featural description of a sign.
6 Conclusion
In this article, I have investigated in how far the patterns of sentential negation
in German Sign Language can be captured by a typological scheme that has been
proposed for spoken language negation. By comparing DGS data to selected data
from various spoken languages, I have shown that DGS – although it is expressed
in the visual-gestural modality – fits very well into the typology developped
for spoken languages, i.e. languages in the vocal-auditory modality. Therefore,
I take the typological scheme to be modality-independent.
First of all, DGS has split negation. One Neg element is an optional, sentence-final
particle, the other one is an affix that is associated with the verb/predicate
morphology. Secondly, in contrast to what we observe in languages such as Turkish
and Ewe, the affix does not have phonological content but rather is featural
in nature. The relevant featural affix is the feature [headshake], which –
in order to be prosodically licensed – must be associated with a base.
I have claimed that this featural affix triggers a prosodic change comparable
to a tone change in tone languages. Interestingly, similar negation patterns,
i.e. the combination of a negative particle with a featural affix, are observed
in spoken languages such as Gã and Ógbrû. Thirdly, I have
shown that the prosodic component in DGS, i.e. the negative headshake, is capable
of spreading. I have suggested to analyze this spreading process as a sandhi
phenomenon comparable to external tone sandhi in tone languages.
Note finally that although I have mentioned above that we find quite similar
negation patterns across a number of different sign languages (i.e. a combination
of an optional manual with an obligatory non-manual element), it is not necessarily
the case that all these sign languages are typologically the same. Actually,
the similarity may only be a superficial one. For ASL, for instance, it has
been claimed that it has simple particle negation and that the headshake is
part of the lexical entry of the manual Neg sign (Neidle et al. 2000).
References
Ablorh-Odjidja, J.R. (1968). Ga for beginners. Accra: Waterville Publishing.
Akinlabi, A. (1996). Featural affixation. Journal of Linguistics 32, 239-289.
Aronoff, M., I. Meir & W. Sandler (2000). Universal and particular aspects
of sign language morphology. In: Grohmann, K.K. & C. Struijke (eds.), University
of Maryland Working Papers in Linguistics 10, 1-33.
Baker-Shenk, C. (1983). A microanalysis of the non-manual components of questions
in American Sign Language. Doctoral dissertation, University of California,
Berkeley.
Baumbach, E.J.M. (1987). Analytical Tsonga grammar. Pretoria: University of
South Africa.
Bole-Richard, R. (1983). Systématique phonologique et grammaticale d’un
parler Ewe: Le Gen-Mina du Sud-Togo et Sud-Bénin. Paris: L’Harmattan.
Corina, D.P., U. Bellugi & J. Reilly (1999). Neuropsychological studies
of linguistic and affective facial expressions in deaf signers. Language and
Speech 42 (2-3), 307-331.
Dahl, Ö. (1979). Typology of sentence negation. Linguistics 17, 79-106.
Dively, V. (2001). Signs without hands: Nonhanded signs in American Sign Language.
In: Dively, V., M. Metzger, S. Taub & A.M. Baer (eds.), Signed languages:
Discoveries from international research. Washington: Gallaudet University Press,
62-73.
Horn, L.R. (1978). Some aspects of negation. In: Greenberg, J.H., C.A. Ferguson
& E.A. Moravcsik (eds.), Universals of human language. Vol. 4: Syntax. Palo
Alto: Stanford University Press, 127-210.
Mboua, C. (1999). Le système aspecto-modal de l’Ógbrû:
de la description des marqueurs préverbaux à la structure de la
phrase en Ógbrû. M.A. thesis, Université de Genève.
Neidle, C., J. Kegl, D. MacLaughlin, B. Bahan & R.G. Lee (2000). The syntax
of American Sign Language. Functional categories and hierarchical structure.
Cambridge, MA: MIT Press.
Nespor, M. & W. Sandler (1999). Prosodic phonology in Israeli Sign Language.
Language & Speech 42, 143-176.
Odden, D. (1995). Tone: African languages. In: Goldsmith, J.A. (ed.), The handbook
of phonological theory. Cambridge, MA: Blackwell, 444-475.
Payne, J.R. (1985). Negation. In: Shopen, T. (ed.), Language typology and syntactic
description. Vol.1: Clause structure. Cambridge: Cambridge University Press,
197-242.
Perlmutter, D.M. (1992). Sonority and syllable structure in American Sign Language.
Linguistic Inquiry 23, 407-442.
Pfau, R. (2002). Applying morphosyntactic and phonological readjustment rules
in natural language negation. In: Meier, R.P., K.A. Cormier & D.G. Quinto-Pozos
(eds.). Modality and structure in signed and spoken languages. Cambridge: Cambridge
University Press, 263-295.
Reilly, J. & D. Anderson (2002). FACES: The acquisition of non-manual morphology
in ASL. In: Morgan, G. & B. Woll (eds.), Directions in sign language acquisition.
Amsterdam: Benjamins, 159-181.
Sandler, W. (1989). Phonological representation of the sign: Linearity and nonlinearity
in American Sign Language. Dordrecht: Foris.
Sandler, W. (1999). Prosody in two natural language modalities. Language &
Speech 42, 127-142.
Sandler, W. (2000). One phonology or two? Sign language and phonological theory.
In: Cheng, L. & R. Sybesma (eds.), The first Glot International state-of-the-article
book. The latest in linguistics. Berlin: Mouton de Gruyter, 349-383.
Wilbur, R.B. (1999). Stress in ASL: Empirical evidence and linguistic issues.
Language & Speech 42, 229-251.
Wilbur, R.B. (2000). Phonological and prosodic layering of nonmanuals in American
Sign Language. In: Emmorey, K. & H. Lane (eds.), The signs of language revisited:
An anthology to honor Ursula Bellugi and Edward Klima. Mahwah, NJ: Lawrence
Erlbaum, 215-244.
Zeshan, U. (2002), Hand, head, and face – negative constructions in sign
languages. Ms., La Trobe University, Melbourne.