2.1 Signed language typology

Zeshan (2004) proposed a typology of negation in signed languages, based on a sample of thirty-eight different SLs, including Auslan. Zeshan found two main types of SLs with respect to negation: i) manual dominant SLs in which negation was conveyed primarily by manual negative signs, even though hs often co-occurred with these signs; ii) nonmanual dominant SLs in which negation was primarily conveyed by head movements and perhaps facial expressions during the production of the negated constituent, even though the nonmanuals could also sometimes be accompanied by manual negative signs.

The hs co-occurred with the negated constituent, if no manual negative sign was also present, or with a manual negating sign, if it was present. It could also spread to other constituents in the clause. See Table 1 for Zeshan’s summary of the characteristics of each type.

From Table 1 one would assume that a prototypical nonmanual dominant SL would always use hs during a negated clause, with or without a manual negative sign being present; and that a prototypical manual dominant one would never allow negation by hs alone. However, the distribution in the data reported by Zeshan (2004, 2006) suggests these prototypes are logical possibilities, rather than realities. Zeshan reports that Swedish SL and Finnish SL come close to the nonmanual dominant prototype, but nonetheless manual-only negation is reported as possible in each (Savolainen 2006; Bergman 1995). Japanese SL and Kata Kolok, a village SL on the island of Bali (Indonesia), were the only two SLs in which hs-only negation was reported not to be possible.

Zeshan (2006: 43) concedes that “some sign languages may display a mixed system where neither manual nor nonmanual negation can be seen as primary [my stress]”. The notion of primary mode of expression of negation suggests that some SLs could be more or less nonmanual dominant or manual dominant than others. However, if hs-only negation is possible in a SL, then according to the criteria in Table 1, it is nonmanual dominant, even if this is extremely rare. Consequently, there is a tension between primary and possible means of negation within the typology. Empirical usage-based data on negation in each SL thus becomes very important in evaluating the reality and usefulness of this typological categorisation.

Zeshan (2004: 52) concludes that SLs appear to differ in five main ways from (standard) clause negation as described in the typological literature on spoken languages (Dahl 1979, 2010; Payne 1985):

1. Intonation (i.e. headshaking) can be used as a grammatical marker of negation in almost all SLs, whereas spoken languages rarely use a specific intonation contour for negation, and even if there is a tone associated with negation, reports suggest it is nonetheless always in the presence of a negative morpheme (Payne 1985: 228; Dryer 2013a). (Zeshan explicitly equates head movements, and not just facial expressions, with intonation in spoken languages, p.6.)

2. Morphological negation in SLs commonly involves suffixes or internal modification, but there are no examples of negative prefixes in SLs. Spoken languages, in contrast, rarely make use of stem internal modification for negation and use both negative prefixes and suffixes, with the former being much more common (Dryer 2013a).

3. Across the thirty-eight SLs surveyed, negative words or particles appear more frequently in post-predicate and clause-final position, whereas in spoken languages the pre-verbal position predominates (Dryer 2013b).

4. Double or multiple manual negative elements in a single negated clause appear to be far more common in SLs than in spoken languages. If we accept hs as a grammatical negative marker, then doubling is also manifested by its co-occurrence with a manual grammatical negative sign. The combination of a negative morpheme with a grammatical intonational element is uncommon in spoken languages. (Headshakes are not considered grammatical markers in these languages.) However, one should note that double particle constructions, such as the ne…pas construction in French, are common in spoken languages (Dryer 2013c) but to date unattested in SLs.

5. The form of negative signs and particles in SLs are often iconic and/or appear to be related to or derived from co-speech gestures used in surrounding communities. Headshaking is a universal nonmanual gesture of naysaying, even if tilting the head backwards is also found in a culture. ^[2] Among manual gestures, the ipsilateral sideways movement of a raised hand palm facing outward, or rocking side to side, is found in many cultures (Kendon 2002). Zeshan observes that it is very similar to at least one of the manual negative sign(s) found in most SLs. In spoken languages negation markers appear arbitrary, with similar word forms not being found across unrelated languages.

Finally, Zeshan also concludes the facial expressions are not, unlike head movements, grammatical markers of negation, but remain gestural in nature because of “their optionality in clauses, the absence of clear-cut rules for their use in clauses, and their use in conjunction with individual signs.” (Zeshan 2004: 14). It would follow from this that if in a given SL the use of hs in negation was shown not to be obligatory, and/or its domain (established through spreading patterns) was unable to be shown to be linguistically constrained, then hs may not, just like facial expressions, be a grammatical marker of negation in that language.

2.2 Headshaking in spoken and signed languages compared

If we are to understand the role of hs in SL negation, we also need to have a clearer idea of how hs functions in face-to-face communication, spoken or signed. This includes contexts where there appears to be no overt grammatical negation.

Head movements have long been studied as a feature of language in face-to-face communication (Jakobson 1972; Goodwin 1980; Schegloff 1987; McClave 2000). Building on earlier researchers, Kendon (2002) analysed the use of headshake in co-speech gesture. He showed that headshakes were not just a non-verbal equivalent of saying no (though they obviously could be), but could separately contribute to the overall meaning of a spoken utterance in a number of ways. This was important because it underlined how co-speech gestures were not simply a substitute for spoken words—they could make their own non-redundant contribution to an utterance. As Kendon put it: “Why should we use gesture if they are just ‘words’ in another form?” (2002: 148).

Importantly, he stressed that all uses of headshake were unified by an underlying theme of negation even if there was clearly no grammatical negation of an otherwise positive co-occurring utterance. This was in contrast to some of the earlier researchers on head movements who had maintained that some uses of the headshake were clearly unrelated to the semantic theme negation because they could appear during an utterance that did not overtly involve grammatical negation.

The theme of negation was illustrated in the eight major ways Kendon observed the headshake being used in his data (Table 2). ^[3]

Researchers have recognised that some of the uses of headshake in Table 2 also occur in SLs in similar environments (Sutton-Spence & Woll 1999; Pfau 2015; Zeshan 2004, 2006). However, SL researchers also claim that there is a profound difference in headshakes used during negation in these languages: they have grammaticalised and become formal markers of negation. They are linguistic in SLs, not gestures as they are in spoken languages.

This is based on two claims. First, in SLs the hs alone can be the sole operator that negates a clause. Strictly speaking, this ability of headshake alone to negate a clause appears not to be covered in Kendon’s list of eight types, though it is close to (iv). For example, imagine a mother talking to her teenage daughter (once again the square brackets indicate the period of headshaking):

Even though the pragmatic effect on the meaning conveyed in example (1) can be that of a simple negative assertion (second paraphrase), in spoken languages the utterance is still regarded as grammatically positive. In SLs, on the other hand, the headshake is said to be able to change the polarity of the clause: all things being equal, i.e. none of the uses (i)-(viii) from Table 2 applying, the addition of the headshake makes an otherwise positive clause a grammatically negative one. This type of headshake (ix), apparently only possible in SLs, can alone make a clause grammatically negative. There are constructions of this type in Auslan: ^[4]

Second, the spreading of headshakes within the negated clause is linguistically constrained in SLs, whereas in spoken languages hs placement is said to be variable and strongly associated to particular lexical items (Andrén 2014). Take a possible invented SL utterance superficially like the type (iii) English example in Table 2:

The hs may or may not spread from the equivalent of the negative particle or negative verb to the subject or to the object, only as each SL language allows. Zeshan (2004) and Zeshan (2006 including several chapters therein) give examples of hs site and spreading patterns in different SLs in support of the claim that they vary from SL to SL and are subject to language-specific syntactic constraints. Pfau (2015) and Oomen & Pfau (2017) augment Zeshan (2006) with data using more examples and additional SLs.

3.1. The use of nonmanuals, especially head movements, in negation in SLs

Auslan, BSL and NZSL have all been reported as using hs, both as the only negating element in a clause (12) and as co-occurring with a manual negative sign (13) (Johnston 1989). ^[6]

With respect to the domain or site of the headshake, Sutton-Spence and Woll also point out that in negated clauses with a headshake, the headshake does not extend over the topic, a fact also noted for NZSL (Collins-Ahlgren 1989; McKee 2006). This is attested in Auslan (there is no overt nonmanual marking over the topic):

Sutton-Spence & Woll (1999) found that a clause-final headshake, without any manual negative sign, was possible in BSL, as it also is in Auslan, see (14).

Most importantly for this study, Sutton-Spence and Woll also point out that a hs need not negate a clause. It can “express emotions such as regret, frustration, disbelief, or sorrow” about the content of a clause that remains grammatically positive (p. 94), as can happen in Auslan too:

With respect to other nonmanuals in negation, such as facial expressions, Sutton-Spence & Woll (1999) report that facial expressions do not alone negate clauses, rather they express related negative feelings, such as displeasure, disgust, and dislike. The negation itself is made either by headshaking and/or by the use of a manual negative sign (p. 73). This is similar to the conclusion Zeshan draws on the role of facial expressions and negation in SLs generally (see Section 2.1):

3.2. The type of morphological negation found in SLs

Signs negated by a final negative suffix-like element; those negated with an internal modification of the sign; and those that have irregular suppletive forms have been reported for Auslan (Johnston 1989) (Figure 2). These morphologically negative verbs render the clause they appear in grammatically negative.

Figure 2:

Morphological negation in Auslan: will-not appends the negative up-turned open handshape (possibly a reduced one-handed form of do-not) to the sign will; see-not appends/blends the negative sign not to the sign see (there is assimilation of some of the extended fingers of the ‘okay’ handshape of not into the handshape of see at the beginning of the new affixed/blended sign); want-not reverses the direction of want; and finally can and can-not have no regular relationship.

3.3. The position of manual negators in SLs

Manual negators were reported to occur both before (13) and after a negated verb (17) in Auslan (Johnston 1989) but the post-verbal position, which is often also clause-final, was said to be preferred. This was also later reported for BSL (Sutton-Spence & Woll 1999) and NZSL (McKee 2006). McKee suggested that the pre-predicate position is probably influenced by English grammar, i.e. Signed English (p. 86).

3.4. The common use of double or multiple negative elements in SLs

Sutton-Spence & Woll (1999) also note the possibility of more than one manual negation sign appearing in a single clause, often for emphasis:

If we accept that hs is an independent formal negating element in negated clauses, then (18) is actually an example of triple negation.

Johnston & Schembri (2007) observed that negative auxiliaries, such as can-not, were also often doubled in negated clauses, with the double usually in a clause-final position:

3.5. The gestural origins of manual and nonmanual negative signs found in SLs

The obvious case of headshakes aside, the signs nothing, no-way and ban in Auslan (see Figure 1) and many similar signs in other SLs appear to have their origins in a commonly found negative gesture described by Kendon (2002).

In these five aspects, therefore, Auslan has been described in ways that suggest it has all the features of negation that Zeshan finds typical of SLs which make them distinct from spoken languages.

Only McKee (2006) suggests what might be the typical or unmarked way to express negation: “In relaxed discourse particularly, nonmanual negation alone is typical, and articulation of the negation headshake may be reduced to a single sideways movement […]” (p. 81). And “Predicates containing a negation sign—or at least the sign itself—tend to be marked by a negative headshake, but not invariably” (p. 84). Indeed, negative imperatives in Auslan were the only constructions in which hs was reported not to occur during negation:

On the basis of these reports, the categorisation of NZSL as nonmanual dominant for negation (Pfau 2015), from either primary or possible perspectives on the typology, seems reasonable. Given that all of the examples of negated clauses in Auslan discussed above contain hs (except for the negative imperative), one might well conclude that Auslan too is nonmanual dominant for negation.

4.1 Source of data

The study dataset consisted of 24.7 hours of video taken from the Auslan Corpus. This was edited into 413 separate clips linked to ELAN annotation files, henceforth referred to as texts. All the texts were segmented into signs and glossed. A large subset of these texts had additional annotations as the result of previous studies, e.g. free translations in English, grammatical class, mouth gestures, and clause boundaries. The annotation schema used in the corpus is described in Johnston (2010) and in the Auslan Corpus Annotation Guidelines (Johnston 2017a). The ELAN ^[7] annotation software used to create these texts allows for the precise time-alignment of annotations with the corresponding video sources on multiple user-specifiable tiers (Wittenburg et al. 2006). Only those annotation practices that are directly relevant to understanding the examples in this paper and how quantitative data were extracted are discussed here. At the conclusion of the study-specific annotation process described below, the number of clauses identified in the entire reference dataset had risen from 9,000 to 12,661 of which 1,672 were negated clauses. ^[8] A subset of 89 texts was comprehensively annotated from beginning to end on multiple tiers for various linguistically relevant features. In these texts there were 6,322 clauses of which 144 were negated.

4.2 Study-specific data preparation and annotation

To conduct this study, it was necessary to be able to identify cases of clause negation in Auslan. Given the nature of the corpus and the existing annotations, there were only three ways in which to locate potential examples of negation in an ELAN annotation file: i) search the glossing tier, ii) search the translation tier, or iii) search the head movement tier. The data preparation described here ensured that these searches would yield meaningful results.

4.2.1 Searching the glossing tier

I searched for the glosses of signs known to be associated with negation in Auslan, determined if the clause they occurred in involved negation, and identified the type of head movement(s) that co-occurred with the negated clauses.

It was necessary to inspect each token of potential neg-related signs to determine whether it was being used as a clause negator because in Auslan a single sign form (identified with a unique gloss) can often function in more than one way, e.g. as a noun or a verb. For example, ban can function as a neg-particle that negates a verb and its clause (7), or it could function as a lexical verb that does not negate the clause it occurs in, as in (21):

(21)

pro→3

ban

sign

‘It (the oral school) banned sign language.’

ADC_c4a,B=00:10:34.087,E=00:10:35.495

Similarly, not, nothing, no-way, and ban can all also be used as naysayers, i.e. stand-alone negative expressions meaning “no” (henceforth neg-interact signs); and nothing may function as a negative pronoun (neg-pro). In these other uses the signs are not clause negators.

The gesture-like palm up sign was also included in these searches because palm-up has been associated with negation in some SLs (Oomen & Pfau 2017) (Figure 3).

Figure 3:

The gesture-like palm-up sign.

Occurrences were tagged if they were associated with negation, produced with a hs and, importantly, whether the palm-up with hs was the only negating element in the clause, as in (22).

(22)

			hs
fs:but	same	pro→1=pro→3	palm-up

‘But they are not like me.’ (lit: ‘But similar me (and) them well-no.’)

ADC_c4a,B=00:10:34.087,E=00:10:35.495

When a neg-related sign was located, the aligned video for each was viewed, the relevant clause identified and each constituent sign tagged for the presence or absence of head movement, as follows: headshake (hs), one strong turn of the head (hs1), or, much less frequently, tilting-back (tilt-back), wobbling (wobble), or a side to side motion of the head alone or the head and the torso together (side-to-side) ^[9]; nodding (nod), and one strong nod (nod1). Nodding was included in the annotation schema because a pilot study (Filipczak et al. 2015) had identified a number of cases where nodding co-occurred with negated clauses.

Any sign in the clause that did not display any of these movements was tagged as having no headshake (nhs) to clearly signal that the clause had been investigated for head movement, and to enable searches for clauses with neg-related signs that did not have a hs over them or anywhere else in the clause. This is important for identifying spreading patterns. Headshakes that were observed to occur when no manual sign was being performed were also annotated over a place-holder gloss annotation on the glossing tier when this occurred, if they were deemed part of the clause and not independent fragmentary interjections. The placeholder gloss had the prefix nms (for ‘nonmanual sign’) thus: nms:hs, see (14). Each identified clause was given a free translation and a literal or close translation into written English, if this had not already been done at that point in the annotation file in previous studies.

It is obvious that a search for the glosses for neg-related signs will only yield clauses that are negated with manual signs, because otherwise there would be no gloss to match the search criteria.

5.1 Headshaking in negated clauses

Almost all (98%) of the grammatically negative clauses in the corpus included a neg-related sign and of these 61% overall also included a hs during, at minimum, the production of that sign (Figure 4). In other words, only 1.7% were negated nonmanually with hs-only, as in (2).

Figure 4:

Presence of HS in negated clauses using different types of neg-related sign (N=1,672), in ascending order of percentage of each type with co-occurring HS. For obvious reasons, the three columns on the right, which are sub-types of HS-only negation, have 100% presence of HS.

From Figure 4 one can also see that in a very small number of clauses, mouthing conveyed negation without any neg-related sign being produced, but there was always also a hs in such clauses. If one includes these clause tokens as exemplars of nonmanual only negation types, the overall count increases to just 2%. ^[10] One should not forget that mouthing is, nonetheless, extremely frequent in Auslan for all classes of signs, including neg-related signs (Johnston et al. 2016).

Equally rarely, one sees that palm-up with a hs (pu+ hs) can function as the sole negating element in a clause. It could be argued that such combinations are essentially tokens of hs-only negation (palm-up not being a lexical sign). Be that as it may, the low numbers mean that even if one added these tokens to the tally of nonmanual only ways of negation they would barely reach 2.9% of all negated clauses. Thus, whichever way one regards pu+hs, nonmanual hs-only negation is rare in Auslan.

The rarity of hs -only negation was somewhat surprising to the author, who is a hearing native signer. (Recall the impression created in the literature on Auslan, BSL and NZSL reported on earlier.) It is possible that text-type could have been a factor in this result because many, but not all, were narrative retells. It is also possible that since translations had previously been added to only about half of the texts in the dataset (i.e. in terms of total duration of the files), even though all of them were glossed, a number of hs-only negated clauses may have been overlooked in the files with translation gaps. If these texts had also been scanned from beginning to end, more hs-only negated clauses might have been identified.

To test these two possibilities ten texts that had incomplete free translations were examined from beginning to end. Eight of these were interviews. In addition, to increase the representation of conversational texts in the dataset, recordings of eight conversational pairs (sixteen texts) were also examined. They had not previously been included in the study dataset because they had not been annotated in any way, i.e. without even glossing. Each conversation lasted about fifteen minutes, representing a further 2 hours of texts (increasing the total reference dataset to 26.7 hours). Though hundreds of headshakes were found in these texts, only a handful of additional hs-only negated clauses were identified. There can be little doubt that hs-only negation is relatively rare in Auslan. There appears no reason to believe that scanning the entire dataset would significantly alter these proportions. ^[11]

5.2 Headshaking in non-negated clauses

When the subset of 89 texts that had comprehensive annotations were systematically scanned, almost 200 non-negated clauses with headshakes, but only 5 hs-only negated clauses, were found. This means there were 40 non-negating hs to every one hs-only negator. Recall also that these 89 texts contained in total 144 negated clauses. Since 65% of manually negated clauses (see Figure 4) are also accompanied by a hs this means that hs is associated with approximately 94 instances of clause negation, compared with approximately 250 instances where it is not. A hs occurring during a clause did not negate a clause in the overwhelming majority of cases, but rather it made its own semantic contribution to the utterance. Clearly, therefore, hs is used more frequently outside of grammatical clause negation than within it.

The relatively high frequency of hs in non-negated clauses was, again, not surprising as it, too, has long been recognised to be a very common behaviour in face-to-face multimodal human communication, spoken or signed, as discussed above (Section 2.2). As Kendon’s headshake usage types describe, negative semantics underlie the uses of the headshaking he identifies, irrespective as to whether the headshakes are produced during negated clauses or not. They tend to express a negative response, direct or indirect, to a question, statement or presupposition, or reflect a negative appraisal by the speaker about what they are saying, see examples (31) to (36). (The implied negation appears in bold. Types (i)-(iii) are not exemplified as they are self-evident)

Type iv. (additional meaning, e.g. disapproval)

Type v. (implied negative meaning in simultaneous wording/signing)

Type vi. (exceptionless statement)

Type vii. (superlative or intensified)

Type viii. (meta-comment, e.g. self-correction, doubt)

There can be little doubt, therefore, that signers use headshakes in these environments, with respect to both form and function, in much the same way as the hearing people they interact with in their surrounding communities.

However, there even appears to be another type (contrastive) to be added to the list:

Type x. headshake indicates a contrastive clause, see also (15)

The vast majority of non-clause-negating headshakes are able to be assigned to one of these types or occur in clauses which are a response to an immediately previously stated or implied proposition of the signer or the interlocutor, i.e. a response frame. How does this dynamic interact with manually negated clauses? Let us now return to the results in the light of this.

5.3 Headshaking and the response frame

Figure 5 presents the data in Figure 4 again, but in terms of the different contexts in which the negated clauses appear. In Figure 5 all neg-negator signs or clauses that had been tagged as response, reprise, contrastive, or imperative (or simply left as negator) have then been grouped together accordingly. This allows us to begin to see the impact of the context on the likelihood that headshaking will also be present during the production of a manually negated clause.

Figure 5:

Number of negated clauses according to semantic and pragmatic context type in ascending order of percentage of each type with co-occurring HS (N=1,672).

One can see from Figure 5 that the rate of co-articulation of hs with apparently straightforward manual negation reduces from 65% to 51%. More telling, the rate of hs increases significantly (up to 89%) when the utterance is part of a response frame rather than merely the assertion of a negative state of affairs (which the manual negation is achieving anyway in virtually all Auslan negated clauses). Two of the clause internal response frames that were identified—one with either a topic, e.g. (25), and one to a pre-posed rhetorical question, e.g. (26)—have already been noted in the literature as environments in which clause-final negation (both manual and nonmanual) is frequently found in SLs (Pfau 2015).

The reprise context, itself, can also be construed within the response frame: reprises tend to occur not only after the first neg-negator sign (by definition) but they also tend to be clause-final and in many cases, are realised not as a simple repetition of the first neg-negator but as a neg-particle. These neg-particles (not, nothing, no-way) are all also used alone as straightforward negative responses (neg-interact) in the language. Moreover, during the reprise element the signer tends to fix their gaze on the interlocutor (or return it to the interlocutor if it had been elsewhere), thus suggesting it also has a strong interactive role in the discourse.

Of course, many of the neg-negator clauses in Figure 5 are examples of Kendon’s type (iii) for headshake usage—headshakes straightforwardly associated with a negative expression. However, we still do not know the extent to which the other headshakes that accompany manually negated clauses in Auslan may actually be instances of the other types of Kendon’s categories. Only the most easily located examples of response frames were identified. Other negative appraisals that make their own semantic contribution, in ways shown above, undoubtedly exist. Take the following example in which the contrastive nature of the hs in the second clause is only obvious in the context of the previous clause, see also (15). Without any translation and without an overt signed but in the contrastive clause, such instances are difficult to locate.

Detailed textual analysis of all negated clauses is needed to reveal any additional semantic contribution of the hs, if present. So we cannot assume that they simply echo, reinforce or mark the manual negation. Many would do more than this.

The small set of hs-only negated clauses is revealing in this regard. Of these 27 clauses, only 3 are not in some kind of response frame, as in (2). Of the 24, 11 are in internal response frames which, by definition, mean they are also clause-final, see (25) and (26); and 6 are direct responses to the interlocutor and consisted of a single pointing sign, usually to the signer (i.e. ‘not me’), but also elsewhere (‘not you’, or ‘not him/her’, or ‘not this/that’). So even hs-only negation underlines the interactive nature of the headshaking. Many also have a negative facial expression (mouth gesture)—communicating rejection or displeasure—which again underlines the interactive nature of the headshaking (cf Bergman 1995).

The 3 non-response frame hs-only negated clauses also have no relevant mouth gesture, so are the only clauses in the dataset which one could say unequivocally are definitely HS-only negated clauses. Interestingly, two of these are imperatives (38) and are thus counterexamples for the claim made in Johnston (1989) on negative imperatives (20).

5.4 Headshaking and the position of neg-negator signs

The special status of hs in SLs as a marker of negation (or the obligatory marker of negation in some SLs) is also said to be manifest in language-specific grammatical constraints (position within clause, spreading of hs over particular syntactic domains and not others).

neg-negators were found both before (4) and after the verb (39), with a preference for before the verb at 62% of tokens (Table 3). This was the opposite of the preference previously claimed for Auslan (and BSL and NZSL) (Section 2.2).

It would be disingenuous not to acknowledge that this is also the position of negators in English, the ambient spoken language. It is also the most common position for negators cross-linguistically (Dryer 2013b). However, in contradistinction to English, negators also occur after the predicate without also being a reprise element (20% of tokens). This flexibility of position is not surprising given the variability in the order of other constituents that has been observed in Auslan, e.g. verbs and their core arguments, or modifiers and their nominal or verbal heads (for Auslan see Johnston 2014a, 2017b; Johnston et al. 2015).

As mentioned above, the negative reprise element was often not simply a repetition of the first neg-negator that occurred in the clause; the majority of reprise tokens were one of the neg-particles no-way, nothing, or not, and in this role they are often clause-final and often attract a hs (Table 4).

This can result in the phenomenon where there is actually no hs over the primary initial neg-negator sign, but there is one over the reprise element (yet again underlining the interactive role of hs) (40).

In fact, no-way is always at the clause extremities, mostly final but also initial, unless it is part of a self-correction, and is rarely signed without a hs in any position. A marked topic is also found in many of these post-predicate constructions (41).

In sum, the position of neg-negators relative to the predicate is not fixed, though there is a strong tendency to precede it. When they follow the predicate and are also clause-final reprise elements, they very frequently attract a co-occurring hs.

5.5 The spreading of headshaking within clauses

I have dealt with the site of hs with respect to neg-related signs (almost one third have no hs), and the possibility of clauses being negated by hs alone (very few are); but the spreading behaviour of headshaking is another factor relevant to determining its role in negation in Auslan. Here I only make generalisations based on hs in manually negated clauses because it has already been noted that almost all hs-only negated clauses were internal responses to topic-like elements. (We already know that the hs in these cases is clause-final and does not spread over the topic-like element. Thus the numbers of the other hs-only negated clauses are too low to make any meaningful generalisations about site and spreading in them.)

At the most general level, two-thirds of manually negated clauses had hs or nhs throughout the entire clause. This is compatible with the notion that headshaking behaviour is strongly related to the status of the clause as a turn within the discourse (including the self-directed responses), and not just to the presence of a neg-related sign. However, this may also be related to the fact that the average simple Auslan clause is quite short. Just over half of all clauses consist of two or fewer signs, and almost three quarters of three or fewer signs (Johnston 2014a). Moreover, we know that at least one overtly manually expressed argument in a clause is very likely to be realised by a pronominal pointing sign. Pronominal pointing signs are usually unstressed and very brief (Johnston 2013a). Because the neg-negator may be positioned on either side of the negated verb, pronominal pointing signs are thus likely to be articulated while the hs is beginning (for the neg-negator or neg-verb that follows) or ending (for the neg-negator or neg-verb that precedes). ^[12] The relative short length of the average simple clause and the likelihood of at least one argument being realised by a pronominal pointing sign means that many clauses have a hs, nhs or nod throughout the entire clause.

Nonetheless, there remain many different attested spreading patterns. In order to describe these, let us add Auslan to the comparison of six other SLs as described by Oomen & Pfau (2017) (Table 5).

There are attested Auslan examples of all negation features (iii) to (viii) in Table 5:

(iii) not clause-final (i.e. one of three neg-particles, glossed as no, no-way or nothing in the Auslan corpus) see (39) and also:

(iv) hs only on not, see (39) and (41).

(v) hs only on predicate:

(vi) hs spread onto object:

(vii) hs spread onto subject:

(viii) Negative Concord, see e.g. (18)

As we have seen, Auslan almost always uses manual neg-negator in clause negation, yet it is displaying some features believed not to be associated with manual dominant negation according to Oomen & Pfau (2017). Indeed, almost every possible permutation of hs spreading cited in Zeshan (2004, 2006), Pfau (2015), and Oomen & Pfau (2017) is attested in the corpus. Space and word limits prevent us from listing them all here, but see the Supplementary Materials for all of the examples cited in Oomen & Pfau with an Auslan Corpus time stamp of at least one equivalent for each.

Overall the data suggest that in Auslan the domain of spreading of hs over constituents of the negated clause does not appear to be constrained by specific syntactic constructions. It is not the case that hs spreads only over some types of constituents but not others, e.g. necessarily over the neg-negator and its negated predicate, and/or over the ‘subject’ or ‘object’ arguments with or without regard to their position relative to the neg-negator and the negated predicate, or the neg-verb.

5.6 The spreading of headshaking across clauses and/or negative transport

Like intra-clause hs spreading patterns, inter-clause spreading has been assumed to show that SL hs is linguistic, rather than gestural (Padden 1988; van Gijn 2004; Pfau 2015; Oomen & Pfau 2017). Spreading can occur in clause complexes (between two or more clauses in a dependency relationship) or in complex clauses (in which one clause is embedded within another as an argument or complement of a verb in the matrix clause, i.e. sentential complementation). There are two different issues here: the one regarding hs spreading from a negated matrix clause to a complement clause (which is not negated by the spread hs); the other regarding negative transport (i.e. negative raising) in which negation marking moves from the embedded clause to the matrix clause.

With respect to hs spreading from a matrix to a complement clause, the focus of Padden’s (1988) observations of ASL and van Gijn’s (2004) of Sign Language of The Netherlands (NGT) is that the hs marks the syntactic subordination of the complement to the matrix clause. This issue will not be pursued further here because spreading in these environments would be no different than the spreading in simple clauses, as already described, i.e. it is driven primarily by pragmatic and semantic factors whose scope can unproblematically cover the entire utterance unit, and is thus not evidence of formal marking of syntactic subordination.

With respect to hs moving from complement clause to matrix clause, the focus of Pfau’s (2015) and Oomen & Pfau’s (2017) discussion, similar observations can be made. However, if one assumes hs is itself a marker of negation, then the movement of hs from dependent or embedded environments to the matrix could be treated as an instance of negative transport commonly understood (Horn 1978). Consider the following English example: I don’t think she’ll come, rather than I think she won’t come. In this example, the negator appears in the matrix clause, rather than the complement clause. However, potential corpus equivalents do not seem to behave as might be expected, i.e. being linguistically constrained.

In Auslan we see not so much the ‘movement’ of the negator hs from the embedded or complement clause to the matrix clause, rather we see its spread, i.e. hs is still retained in the latter, even if it also appears in the former. This should, in principle, leave the complement clause still negated while the matrix is also negated (I don’t think she won’t come). However, the meaning does not support this:

This type of spreading of hs seems to be subject to the kinds of pragmatic and semantic factors we have already discussed, rather than parallel negative transport. Hence the translation in (47). The reference is to neonatal screening for deafness, i.e. the hs is either part of the denial of a presupposition (namely, that neo-natal screening for deafness is good and supported by deaf people too), or is simply regressive spreading from the complement clause and does not change the polarity of the matrix, or it is in fact overdetermined, i.e. both factors operate.

The same phenomenon is attested when what is supposedly moved is both a hs and a neg-negator:

Unlike negative transport, there still appears to remain overt negation in the complement clause. Interestingly this is not manifested in the meaning because the two negatives do not cancel each other out (≠ ‘I think that it’s a good idea.’). It appears more like a case of double negation vis-à-vis the neg-negator, and spreading vis-à-vis the hs, as well as indexing a response frame (hence the second, better, translation). As we have seen, in Auslan multiple negations are possible, without producing a positive reading, see (18).

Nonetheless, there are some attested examples that look much closer to negative transport:

The manual signs in this example create a pattern that closely parallels English negative transport, but the nonmanuals do not (if we assume they are, in fact, part of the grammatical negative marking of the clauses). With respect to the manual signs, we should not be surprised to find English-like constructions appearing in Auslan due to language contact. As for (49), it is either not the same as the English structure because there remains a negative marking in the complement clause (the hs) and this does not occur in English negative transport; or it is very much the same as the English because the hs is not, in fact, part of grammatical negation. After all, English speakers can be easily imagined saying I don’t think it’ll work while headshaking throughout (cf Padden 1988; van Gijn 2004).

Once again it appears hard to support the notion that hs is a formal marker of grammatical negation in Auslan. The following example thus appears to be a case of spreading rather than negative transport.

And the following appears to be a case of spreading and/or response frame indexing rather than negative transport.

In sum, not only do hs spreading patterns in negated clauses not appear to be linguistically constrained, but the participation of hs in negative transport also appears open to question.

5.7 Summary

The results show that Auslan users negate clauses using only a headshake extremely rarely; instead, signers almost always negate clauses using neg-related manual signs which, in just over half of cases, are also likely to be accompanied by a headshake. Many of these headshakes add pragmatic negative appraisal about the co-occurring clause, or negate an unstated or implied proposition or assumption. In these clauses the headshake does not, strictly speaking, also formally mark clause negation additional to that provided by the neg-related manual sign. The pragmatic role of headshaking is reinforced by the fact that they are much more likely to co-occur during manually negated clauses that are part of a response or a response frame. The remaining cases in which headshaking does not add negative appraisal appear to be semi-regular non-obligatory pairings of a nonmanual sign (the headshake) with a manual sign (the neg-related element). Headshakes also frequently appear in non-negated clauses: so frequently, in fact, that this is overwhelmingly the most common construction in which headshakes are found. Headshakes in manually negated clauses thus appear to be like the majority of headshakes in non-negated clauses in the corpus which, in turn, are like headshaking in the face-to-face language of comparable spoken language using communities (cf Kendon 2002; Harrison 2009, 2010; Andrén 2014). Finally, position and spreading behaviour of headshakes during manually negated clauses appears related to pragmatic and semantic factors rather than language-specific syntactic constraints. ^[14] This interpretation of the spreading data accords with observations on the meta-comment role that headshake plays in the language, the rarity of hs-only clause negation, and the variability of the position of neg-related signs within clauses, complex clauses and clause complexes.