1 Cindy van Boven (10761500) rMA thesis, July 2019 Supervisor: dr. Roland Pfau Second examiner: prof. dr. Kees Hengeveld
The marking of imperatives in Sign Language of the Netherlands
Universiteit van Amsterdam Graduate School of Humanities Research master’s in Linguistics
2 Abstract
This study provides an overview of the syntactic, morphological, and prosodic marking of imperatives in Sign Language of the Netherlands (NGT). The imperative form (which includes six subtypes: orders, invitations, suggestions/advice, permissions, instructions, and recommendations) has been found to exhibit specific morphological and syntactic features in many spoken languages, while it has barely been researched for sign languages. In order to investigate imperatives in NGT, the present study combines the analysis of corpus data with data elicitation. The results show that the imperative exists as a grammaticalized sentence type in the language, and that it is characterized by the presence of at least one of three features: (i) manual markers (particles), (ii) non-manual (prosodic) markers, and (iii) omission of the subject. Two other cross-linguistically common imperative markers – reduced verbal morphology and word order – were not found in NGT imperatives. Moreover, a specific and absolute distinction between imperative subtypes is not made in NGT imperative marking, however, the language broadly distinguishes orders from orders by means of non-manuals. Comparison of our results to previous research shows that NGT imperatives exhibit formal properties that are already known to single out imperatives in both spoken languages and other sign languages. Yet, importantly, cross-linguistic, and specifically intra-modal variation in the marking of imperatives can also be identified.
3 Table of contents:
Abstract ... 2
1. Introduction ... 5
2. Imperatives in spoken and sign languages ... 6
2.1 Defining imperatives ... 6
2.2 Imperative markers in spoken languages ... 7
2.2.1 A marked word order ... 7
2.2.2 The imperative subject ... 7
2.2.3 Reduced verbal morphology ... 8
2.2.4 Embedding and negation ... 9
2.2.5 Intonation ... 9
2.2.6 Particles ... 10
2.3 Imperative markers in sign languages ... 10
2.3.1 Manual markers ... 11
2.3.2 Non-manual markers ... 13
2.3.3 Other characteristics ... 14
2.3.4 Preliminary observations on imperatives in NGT ... 15
2.4. Research questions and predictions ... 16
3. Methodology ... 17
3.1 Corpus search ... 17
3.1.1 The Corpus NGT ... 17
3.1.2 Corpus data collection ... 17
3.2 Data elicitation ... 17 3.2.1 Participants ... 18 3.2.2 Stimuli ... 18 3.2.3 Procedure ... 18 3.3 Data analysis ... 19 4. Results ... 20 4.1 Data overview ... 20 4.2 Imperative markers in NGT ... 21
4.2.1 Manual markers of NGT imperatives ... 22
4.2.2 Non-manual markers of NGT imperatives... 25
4.2.3 Subject of NGT imperatives ... 31
4
4.2.5 Verbal morphology in NGT imperatives ... 33
4.3 Co-occurrence of NGT imperative markers ... 35
4.4 Obligatoriness of NGT imperative marking ... 36
4.5 Differences between imperative types in NGT ... 36
5. Discussion ... 38
5.1 Comparison of corpus and elicited data ... 38
5.2 A typological perspective on NGT imperatives ... 39
5.2.1 Optional null subject and overt noun subjects ... 39
5.2.2 Non-obligatory particles for all imperative types ... 40
5.2.3 Non-manuals to distinguish orders from non-orders ... 41
5.2.4 Verb-initial constructions ... 43
5.2.5 No reduced verbal morphology... 44
5.2.6 Role shift and embedding ... 45
5.2.7 Intra-modal comparison ... 46
5.3 Alternative: semi-modal constructions with BETTER ... 47
6. Conclusion ... 49
References ... 50
Appendix A. Glossing conventions sign language examples ... 53
Appendix B. Overview of searches in the Corpus NGT ... 54
Appendix C. Overview of participants data elicitation ... 55
Appendix D. Stimuli data elicitation in randomized order and specified for subtype ... 56
5 1. Introduction
Many investigated languages have a designated imperative clause type, i.e., a grammatical form specialized to elicit a certain behavior from an addressee. The imperative form can be used for several imperative subtypes, including (at least) orders, invitations, suggestions/advice, permissions, instructions, and recommendations (Portner 2007; Quer et al. 2017). Cross-linguistically, this sentence type has been found to exhibit specific morphological and syntactic features that set it apart from other sentence types (e.g., Sadock & Zwicky 1985; Aikhenvald 2010). For instance, imperatives are often found to exhibit a marked word order, null subjects, reduced verbal morphology, specific intonation, and particles, and they are often resistant to embedding and negation. Imperatives are relatively well-studied for spoken languages, and it has even been stated that they are a universal feature of languages (Van der Wurff 2007).
Yet, this sentence type has barely been studied for sign languages, while sign languages are by now widely recognized as fully-fledged, natural languages. Some studies have started to investigate imperatives in sign languages – the most elaborate study, Donati et al. (2017), concluded that the imperative also exists as a grammaticalized sentence type in Italian, French, and Catalan Sign Language, as it displays the same properties as imperatives in spoken languages. Still, this was found for only three sign languages, while there are, of course, many more sign languages all over the world. Therefore, this study focuses on the description of imperatives in another sign language: Sign Language of the Netherlands (Nederlandse Gebarentaal, NGT).
The main question in this study is how (the different subtypes of) imperatives are syntactically, morphologically, and/or prosodically marked in NGT. In order to answer this question, it was investigated (i) which imperative markers occur in NGT, (ii) to what extent imperative markers co-occur, (iii) whether the marking of imperatives is obligatory, and (iv) whether the different subtypes are marked differently. Analysis of both corpus data and elicited data shows that the imperative exists as a sentence type in NGT, marked by several particles, specific non-manuals, and a null subject. While none of these features always occur, an imperative is obligatorily marked by at least one feature; still, all three types of markers can also occur simultaneously. Additionally, particles are always combined with other imperative markers. Moreover, NGT broadly distinguishes between orders and non-orders by means of non-manuals, however, the distinction is not absolute and a more specific distinction between subtypes is not made. The results confirm that imperatives do exist in sign languages, just as in spoken languages, and that languages in the visual-spatial modality should be included in typological and comparative studies.
Section 2 provides a more detailed definition of imperatives, and describes previous research into this sentence type in both spoken and sign languages. This results in our research questions and predictions. Section 3 describes the methodology used in this study, i.e., how the corpus search was conducted and how data were elicited, and how both data sets were analyzed. Section 4 describes the results from both data sets, and Section 5 will discuss these results further by comparing the corpus and elicited data, by providing a typological perspective on NGT imperatives, and by briefly discussing an alternative to the imperative, the semi-modal construction. Finally, Section 6 will draw a conclusion.
6 2. Imperatives in spoken and sign languages
2.1 Defining imperatives
An imperative can be defined as “a grammatical form that is specialized to elicit a behavior from the addressee” (Donati et al. 2017: 117; see also von Fintel & Iatridou 2017). Imperatives are one of the four well-recognized sentence types (along with declaratives, exclamatives, and questions). An example of an imperative is given in (1).
(1) Read this book!
Note that our definition does not equate imperatives to orders or commands, as imperative sentences are not the only way to express commands (e.g., Aikhenvald 2010; Donati et al. 2017; von Fintel & Iatridou 2017). For instance, they can also be expressed through yes/no-questions (2a) as well as deontic modal constructions (2b). It is important to note that imperatives can be distinguished from the latter, because modal constructions are propositional and can therefore be true or false, whereas imperatives cannot; compare (2b) to (1) (Portner 2007; Donati et al. 2017).
(2) a. Could you pass me the salt? b. You should read this book!
Thus, defining imperatives as ‘orders’ would be too broad. At the same time, it would also be too narrow, as many languages use the same imperative form for functions other than orders: a variety of (directive) meanings can be displayed by this sentence type (Portner 2007; Aikhenvald 2010), and the imperative can be divided into (at least) six subtypes. Apart from orders, these imperative types include invitations, suggestions/advice, permissions, instructions, and recommendations (Portner 2007; Quer et al. 2017). Example (1), as mentioned above, is an order, which expresses the will of the speaker for someone to do something. Invitations warmly encourage someone to do something, and are expressions of politeness as in (3a). In suggestions/advice, the speakers advises the addressee on what to do in order to improve the addressee’s situation (3b). Permissions express an authorization, as they are usually a reply to a request (3c). Moreover, a speaker may provide instructions on how to carry out a specific action (including directions; 3d). Finally, recommendations express the speaker’s desire or concern for a future situation (3e) (Quer et al. 2017: 306-308). (3) a. Have a piece of cake!
b. Buy healthy food for your kids! c. May I take your pen? – Yes, take it! d. Take the first street on the left.
e. Don’t go too fast with your motorbike! [Quer et al. 2017: 306-308] These imperative subtypes can be disambiguated by means of context, but in some languages the different subtypes have also acquired a formal realization in the imperative system, for instance by means of differentiating particles (see Section 2.2.6) (Aikhenvald 2010)1.
Cross-linguistically, imperatives have been found to exhibit morphological and syntactic features that characterize this sentence type (see, e.g., Aikhenvald 2010). Sadock & Zwicky (1985) is one of the first studies that investigates cross-linguistic imperative data,
1 Thus, according to Aikhenvald (2010: 202; 421), languages can even have more than one imperative form,
7 investigating imperatives in 23 languages. They point out that imperatives and the other clause types are mutually exclusive – instances like (4), i.e., interrogative and imperative at the same time, are not found (Sadock & Zwicky 1985; van der Wurff 2007: 19).
(4) * Give it to me?! [van der Wurff 2007: 19]
Thus, many investigated languages have a designated imperative clause type, and the next sections will describe some of its morphological and syntactic features in spoken languages. 2.2 Imperative markers in spoken languages
2.2.1 A marked word order
In some languages, imperatives have a marked word order. Consider the French example (5a) as compared to (5b) – in imperatives, the clitic follows the verb in French, whereas it precedes the verb in declaratives. Clitic pronouns occurring in a different order in imperative clauses is attested widely in Romance languages (see e.g. Aikhenvald 2010 for Rumanian and Catalan; Donati et al. 2017 for Italian). In many of these languages, including French, this type of enclisis is not found with other verb forms, but only in imperatives, where it is obligatory (Rooryck 1992; van der Wurff 2007: 80).
(5) a. Fais-le! do.IMP-it
‘Do it!’ [French; van der Wurff 2007: 80]
b. Tu le fais you it do
‘You do it.’ [French]
To give another example, (6) shows the marked word order of Dutch imperatives. In spite of the fact that Dutch is a verb-second language (6a), the imperative verb must be clause-initial (6b) – the language does not allow (overt) topics in imperative clauses (6c) (van der Wurff 2007: 76).
(6) a. Jij drinkt de wijn. you drink.2S the wine ‘You drink the wine.’
b. Drink (jij) die wijn later maar op! drink.IMP (you) that wine later ADV up ‘Finish that wine later!’
c. * Die wijn drink later maar op! that wine drink later ADV up
‘That wine, finish later!’ [Dutch; adapted from van der Wurff 2007: 76] 2.2.2 The imperative subject
Mauck et al. (2005) examine the syntactic and semantic properties of imperative subjects across a variety of languages. They indicate that imperative subjects differ from declarative and interrogative subjects, and thus mark the sentence type. In English, null subjects appear in imperatives as in (7a), although this language normally does not allow for null subjects in
8 declarative sentences (Mauck et al. 2005: 134). In addition to null subjects as well as overt second person subjects, English also allows for quantificational subjects in imperatives (7b), which “quantify over a set of addressees” (Mauck et al. 2005: 136).
(7) a. ø eat your dinner!
b. Everyone raise your hand! [Mauck et al. 2005: 134-136] Thus, in English addressee and subject coincide. However, in other languages, this is not necessarily the case. For instance, Bhojpuri (a modern Indic language) allows for referential third person subjects in imperatives. In example (8), the subject of the imperative is not addressed, but the addressee is asked to ensure the subject does what is asked (Mauck et al. 2005: 138). Sentences like the one in example (8) employ the same verbal paradigm as sentences that are uncontroversially classified as imperatives, and therefore Mauck et al. (2005) argue that they are imperatives.
(8) Lajke (aapan) tini baje aaveN children-NOM (self) 3 o’clock come-IMP.3P
‘(Your) children come at 3 o’clock!’ [Bhojpuri; Mauck et al. 2005: 138] Mauck et al. (2005: 143) mention that “a special role for the addressee seems definitional of the notion of imperative”, whether the addressee is the subject (7ab) or not (8). However, they also note that certain languages exhibit clauses that resemble imperatives, yet the subject is associated with the speaker instead of, or in addition to, the addressee. For instance, in Classical Sanskrit, first person singular imperatives are interpreted as promises (9a), whereas first person plural imperatives are exhortatives (9b) (Mauck et al. 2005: 144).
(9) a. AavyuSam jaagrtaat aham daybreak watch-IMP.1S I ‘I will watch until daybreak.’
b. Carama vasudhaam idam
wander-IMP.1P land-NOM this-NOM
‘Let’s wander this land.’ [Classical Sanskrit; Mauck et al. 2005: 144] Mauck et al. (2005) note, however, that ‘imperative’ may not be the best label for this type. The authors argue that imperatives and these first-person forms may be members of the same clause type (which they call ‘jussive’). Here we will focus only on imperatives, i.e., first-person forms will not be included in the present study.
2.2.3 Reduced verbal morphology
Cross-linguistically, the verbal morphology of the imperative is often limited to a minimal number of forms (Zhang 1990; Mauck 2005; Aikhenvald 2010). In his survey, Mauck (2005) shows that very few languages have a complete paradigm in imperatives. Most languages have special verbal forms for second person imperatives only. Special forms for second person and first person plural imperatives, inclusive of the addressee, also exist. Only a few languages have second and third person imperative forms, and some have imperatives that inflect for all persons (Evenki, Yidin, Dyirbal, Sinaugoro, Hixkaryana, Caddo, Classical Sanskrit, and Bhojpuri) (Mauck 2005, as cited in Zanuttini 2008).
When the imperative morphological paradigm has only one form, the null subject is interpreted as second person (Zanuttini 2008) – this is evident from the English example (7a).
9 Example (10) shows that in Italian, where verbs usually display rich verbal inflection, the imperative consists of the stem, a root plus a thematic vowel, without any person marking (Donati et al. 2017: 119-120). (10) a. Io mangio I eat.PRS.1SG ‘I eat.’ b. Tu mangi you eat.PRS.2SG ‘You eat.’ c. Mangia (tu)! eat.IMP.2SG you
‘(You) eat!’ [Italian; Donati et al. 2017: 119-120] 2.2.4 Embedding and negation
A syntactic property of imperatives described in many studies, is that they typically cannot be embedded (e.g., Sadock & Zwicky 1985). Many languages use the subjunctive, the infinitive, or a deontic modal construction for embedding instead; compare (11a) to (11b) (Donati et al. 2017: 122).
(11) a. * He says that go and get him.
b. He says that you should go and get him. [Donati et al. 2017: 122] However, note that some languages do allow for embedded imperatives, such as Korean. An example is given in (12) (Portner 2007: 357).
(12) Inho-ka Sooni-ekey cip-ey ka-la-ko malha-ess-ta. Inho-NOM Sooni-to home-to go-IMP-COMP say-PAST-DEC
‘Inho said to Sooni to go home.’ [Korean; Portner 2007: 357] Just as embedded imperatives, syntactically negated imperatives also do not exist in all languages, and negation of imperatives often differs from negation of declarative sentences. Sadock & Zwicky (1985) report that out of the 23 investigated languages, 75% form the negative imperative either by using a special negator, or by means of a verb form that differs from the verb form in affirmative imperatives. Other languages have replacement options, for instance in Spanish, where negative imperatives are ungrammatical, but grammatical counterparts with an infinitive or subjunctive exist (Rivero 1994). The present study will not consider negative imperatives, but focuses on affirmative imperatives.
2.2.5 Intonation
Imperatives are also expressed by means of prosodic cues. For instance, in Lote, an Oceanic language spoken in Papua New Guinea, the only way to differentiate a second-person imperative from a second-person statement, is by means of intonation. Thus, in this language, imperative-specific intonation patterns are the only imperative markers (Aikhenvald 2010). Moreover, for Spanish, Robles-Puente (2011) reports that higher F0 values are found in imperatives as compared to declaratives. However, he also emphasizes that there is not one unique and categorical Spanish imperative intonation, but that there are various phonetic modifications resulting in several F0 configurations, that express imperativity, and clearly
10 differ from declarative intonation. Furthermore, Hellbernd & Sammler (2016) find that prosody alone can be used to mark imperatives (more specifically, suggestions and warnings) for both existing German words and non-words. They conclude that these prosodic cues can be produced and understood independently of semantic meaning, contextual information, and emotional coloring of the utterance.
Note that the exact intonation contour may differ between imperative subtypes, which is, for instance, the case for English (e.g., a “low fall” indicates an order, while “high fall-low rise” is more likely used for a request; see Aikhenvald (2010: 90)). Moreover, not all languages have a specific imperative intonation contour: languages with imperative-specific verbal marking may not exhibit any particular intonation to distinguish imperatives; this is the case in, for instance, Urarina, an isolate from Peru (Aikhenvald 2010: 90).
2.2.6 Particles
A language can use certain particles to distinguish between the different imperative types introduced in Section 2.1. For instance, in Badiotto, an Italian dialect, imperative sentences are ungrammatical without one of the four particles ma, mo, pa, pö (this does not apply to negative imperatives, where these particles are optional). Crucially, different particles are used for different imperative types: e.g., mo signals that the sentence is an order (13a), whereas ma is used for advice and permission (13b) (Poletto & Zanuttini 2003: 4-5).
(13) a. Mànge-l mo! eat-it mo ‘Eat it!’
b. Mànge-l ma che spo crësceste eat-it ma that then grow
‘Eat it and you’ll grow.’ [Badiotto; Paletto & Zanuttini 2003: 4-5] Thus, the overview above makes clear that imperatives are relatively well-studied in a number of spoken languages. Van der Wurff (2007: 57) even states that “the [affirmative] imperative clause type seems to be a universal feature of languages”. However, while imperatives are indeed well-studied for spoken languages, they have barely been studied for sign languages. Therefore, we should be cautious in considering the imperative a language universal. Some studies have started to investigate imperatives in sign languages. These studies will be discussed in the next section.
2.3 Imperative markers in sign languages
Sign languages, like spoken languages, also distinguish between the sentence types declarative, interrogative, imperative, and exclamative (for an overview, see Cecchetto 2012). While previous studies have investigated sign language declaratives (e.g., Volterra et al. 1984) and interrogatives (e.g., Aarons et al. 1992; Zeshan 2004; Aboh & Pfau 2011; Branchini et al. 2013), to date there is very little research on sign language imperatives, and no research on exclamatives.
This section will focus on the few available studies on sign language imperatives. Given the fact that sign languages are articulated in the visual-spatial modality, imperative morphology can be expressed simultaneously rather than concatenatively, for instance by non-manual markers (cf. intonation). Moreover, separate manual signs can function as imperative markers (cf. particles) (Quer et al. 2017). The next subsection will discuss these manual markers, after which we will turn to non-manual marking. Then other syntactic and morphological characteristics found in sign language imperatives are discussed. Finally, we
11 discuss some preliminary observations on NGT imperatives. For the glossing conventions used in the sign language examples, see Appendix A.
2.3.1 Manual markers
Donati et al. (2017) investigated the marking of imperatives in three sign languages: Italian Sign Language (LIS), French Sign Language (LSF), and Catalan Sign Language (LSC). They presented deaf native signers from each language (2 for LIS, 3 for LSF, and 2 for LSC) with type-specific elicitation contexts, that elicited the different imperative types. Moreover, after being presented with the elicitation contexts, the informants had a discussion and presented the outcome of that discussion to the researchers. Informants were then asked for grammaticality judgments.
For all three sign languages, Donati et al. (2017) found that specific signs are used to mark imperative sentences. First of all, LIS has an imperative sign glossed as B-INDEX (Figure 1a; (14)), which appears with all different imperative types, while its movement varies across uses. B-INDEX is directional and thus agrees with the arguments of the sentence, and it appears
in postverbal position. Furthermore, when the addressee is a plural entity, the signer can start the imperative with an attention getter (Figure 1b). When the imperative implies a movement from the addressee, a different marker is used in LIS, namely MOVIMP (Figure 1c; (15)). This sign also occurs in postverbal position, and cannot co-occur with B-INDEX.
Figure 1a. B-INDEX in LIS Figure 1b. Attention getter in LIS Figure 1c. MOVIMP in LIS
(Donati et al. 2017: 129-131)
(14) JUMP B-INDEX2
‘Jump!’ [LIS; Donati et al. 2017: 130]
(15) PEN TAKE MOVIMP
‘Go and take a pen!’ [LIS; Donati et al. 2017: 130]
LSF also has the MOVIMP sign, however, in LSF the handshape of this sign differs across imperative types. 1-MOVIMP (16) has a 1-handshape, as in LIS (Figure 1c), and is used for
commands. B-MOVIMP (Figure 2; (17)), on the other hand, has a B-handshape, and is used for
all other imperative functions. As in LIS, both forms require some sort of displacement of the addressee. MOVIMP in LSF can appear in different syntactic positions (e.g., in initial and in final position). The two different MOVIMP types cannot co-occur, and also cannot be doubled.
As in LIS, MOVIMP is directional. Note that the B-INDEX sign does not occur in LSF imperatives, although it does exist in the language (Donati et al. 2017).
12 Figure 2. B-MOVIMP in LSF (Donati et al. 2017: 136-137)
(16) BAR2 INDEX21-MOVIMP3 BITE
‘Go and bite the bar.’ [LSF; Donati et al. 2017: 137] (17) INDEX2 BAR BITE 2B-MOVIMP3
‘Please, go and bite the bar.’ [LSF; Donati et al. 2017: 137] Finally, in LSC, several particles can be used to mark the imperative. They usually appear in postverbal position, and do not require a displacement of the addressee, unlike the MOVIMP
signs in LSF and LIS. The most common particle has a 1-handshape, and is used for direct commands. When it is articulated with a B-handshape, it is used for polite requests. Again, these particles are directional. Donati et al. (2017) do not provide any examples for these two particles. They do, however, illustrate two other particles: VA (Figure 3a) and VINGA (Figure 3b) (both meaning ‘come on!’). Finally, for body-anchored verbs (that cannot modify their movement for agreement), additional auxiliary elements are used to convey an imperative, i.e., an agreement auxiliary (18a), a pronoun-pointing sign (18b), or a pronoun with person form (18c) (Donati et al. 2017).
Figure 3a. VA in LSC Figure 3b. VINGA in LSC
(Donati et al. 2017: 145) (18) a. LISTEN/SEE AUX1
b. LISTEN/SEE INDEX1 c. LISTEN/SEE PERSON1
‘Listen to me!’ [LSC; Donati et al. 2017: 146]
Zeshan (2003)also describes particles that mark a sentence as an imperative in Indo-Pakistani Sign Language (IPSL). She distinguishes KARO (Figure 4a), which marks a “neutral positive imperative, distant force”, JA:O (Figure 4b), which marks a “nonpolite positive imperative,
13 immediate force”, and NAKARO, which marks negative imperatives (Zeshan 2003: 164-165). These particles always occur in clause-final position (i.e., after the predicate). Again, these particles cannot be combined.
Figure 4a. KARO in IPSL (Zeshan 2003: 164-165)
Figure 4b. JA:O in IPSL
2.3.2 Non-manual markers
Donati et al. (2017) not only found manual imperative markers in the three investigated sign languages; they also found non-manual markers, i.e., linguistic elements expressed on the face and body rather than on the hands. In LIS, there are specific non-manual markers for each imperative type, that obligatorily spread over the sentence. For instance, orders are characterized by furrowed brows (fb) and tensed eyes (19a), while recommendations are marked by raised brows (br) and wide eyes (19b). Note that these non-manual markers can occur without the manual markers, i.e., they can mark imperatives on their own.
(19) a. fb
KNEEL-DOWN
‘Kneel down!’
b. br
INDEX2 MOPED SLOW
‘Drive your moped slowly!’ [LIS; Donati et al. 2017: 132] As for LSF, Donati et al. (2017) note that the most natural way to mark an imperative is by means of raised eyebrows, without a manual sign. The raised eyebrows may co-occur with other non-manual markers, which add the specific type of imperative (e.g., a head nod to signal a request). Normally the eyebrow raise spreads over the sentence, but when it occurs in an imperative with a manual imperative marker, it does not always spread over that manual marker. Finally, LSC imperatives can also be expressed non-manually. For instance, furrowed brows accompany commands, while head nod and protruding lips accompany permissions.
Another study into non-manual markers accompanying imperatives was conducted by Brentari et al. (2018). They investigated how strongly and how consistently the different imperative types are encoded non-manually in American Sign Language (ASL). They also tested the comprehension of these non-manuals by ASL signers, as compared to German Sign Language signers and American and German non-signers, who had no prior exposure to ASL. Their results show that in ASL, commands (marked by e.g. wide eyes and shorter sign durations) are the most easily distinguished from the other imperative types, while what they call “addressee goal imperatives” (such as advice and permission) were not found to have
14 unique prosodic patterns. As for comprehension, the imperative meanings are accessible to the groups without ASL exposure, yet, as the cues have been further conventionalized in the language, the ASL signers performed better overall.
Thus, these studies show that imperatives are marked non-manually in sign languages, yet the extent to which these markers are grammaticalized for the different imperative types in different sign languages, differs per sign language.
2.3.3 Other characteristics
As became clear in Section 2.2, apart from particles and intonation, there are many more syntactic and morphological characteristics that have been identified for spoken language imperatives. Donati et al. (2017) find that some of these characteristics also hold for imperatives in the three investigated sign languages. For instance, in LIS, the imperative resists ‘normal’ negation – the movement of the manual negator differs from that in declarative sentences, and is marked by different non-manual markers than negative declarative sentences (raised brows instead of furrowed brows). Moreover, the manual imperative markers are incompatible with manual or non-manual negation in LIS. In LSC, too, the negative forms that typically occur in negative imperatives differ from those used in negative declaratives.
As for the imperative subject, null subjects seem to be preferred in LIS imperatives (note, however, that they are also allowed in declaratives). LIS imperatives also allow the overt expression the subject, as long as the subject referent includes the addressee. However, certain imperative types in LIS (invitations and suggestions) never allow the overt expression of the subject. In LSC, the interpretation of quantificational subjects is restricted to the set of addressees. However, in exhortatives, quantificational subjects do not necessarily include the addressee, as shown in (20).
(20) WHO^INDEX3PL2HELP1
‘Someone help me!’ [LSC; Donati et al. 2017: 148]
Moreover, in LSC, imperatives display a marked word order. The basic word order of LSC is SOV, but imperative sentences are VO. As for embedding, LSF provides some interesting examples. The manual imperative marker MOVIMP can be found under role shift when
embedded2 under attitude predicates, such as ‘say’ and ‘request’ (21a). Interestingly, MOVIMP
can also be embedded without role shift (21b). Thus, in LSF imperatives do not resist embedding.
(21) a. PIERRE SAY INDEX2 MOVIMPIX BITE
‘Pierre said bite him.’
b. PIERRE WANT INDEX21-MOVIMP BITE
‘Pierre wants you to bite him.’ [LSF; Donati et al. 2017: 144] Donati et al. (2017) conclude from their investigation that imperatives also exist as a grammaticalized sentence type in sign languages. However, this is based on only three sign languages, and thorough research into imperatives in other sign languages is needed. Therefore, the current study will describe the properties of imperatives in Sign Language of the Netherlands (Nederlandse Gebarentaal, NGT). Before we turn to the research questions, some preliminary observations regarding imperatives in NGT will be discussed.
15 2.3.4 Preliminary observations on imperatives in NGT
Maier et al. (2013) conducted a study that provides some first insights in what imperatives look like in NGT. They presented four Deaf native NGT signers with a written list (in Dutch) of typical examples of imperatives, preceded by a description of the context, “ranging from very polite requests to commands proper” (Maier et al. 2013: 367). Based on the signed translations, the researchers conclude that imperatives in NGT can be marked in different ways, shown in Table 1 (Maier et al. 2013: 368). Both manual and non-manual markers, as well as subject omission, were found in NGT, just as Donati et al. (2017) found for LIS, LSF, and LSC. For the NGT particles, Maier et al. (2013) only provide examples of COME-ON and
REQUEST, shown here in Figure 5ab.
Table 1. Findings from Maier et al. (2013: 368) regarding marking of NGT imperatives
Figure 5a. COME-ON in NGT (Maier et al. 2013: 370-371)
Figure 5b. REQUEST in NGT
The authors note that the same imperative morphology is used for both commands directed towards addressees, and for commands directed towards non-addressees. Maier et al. (2013), however, do not really provide a detailed analysis of the conditions under which the imperative markers in Table 1 occur. They only note that “none of them is obligatory for marking an imperative, but we hypothesize that at least one, and probably more, should be present in order for the utterance to count as an imperative” (Maier et al. 2013: 368). Thus, it remains unclear when exactly these markers (co-)occur. Maier et al. (2013) also do not distinguish between different imperative types in their analysis, and thus do not analyze whether different types of imperatives are marked differently.Moreover, it is unclear whether they also took into account other relevant imperative properties, such as word order.
Finally, a methodological issue of this study is that the signers signed NGT translations of stimuli in written Dutch. This is problematic, as Dutch may have had an influence on NGT, i.e., maybe the way the signers expressed the imperatives was influenced by the fact that they were translating from Dutch, a language that, for instance, also allows for null subjects in imperatives (Bennis 2007). Thus, further research into NGT imperatives is necessary, using only signed and/or visual stimuli.
16 2.4. Research questions and predictions
The aim of this study is to describe how the different subtypes of imperatives (orders, invitations, suggestions/advice, permissions, instructions, recommendations) are marked in NGT. The main question is therefore:
RQ. How are (the different subtypes of) imperatives syntactically, morphologically, and/or prosodically marked in NGT?
In order to provide this description, four sub-questions are asked: RQ.1. Which markers occur in NGT imperatives?
RQ.2. Do the different markers co-occur? If yes, how? RQ.3. Is marking of imperatives obligatory?
RQ.4. Are the different types of imperatives marked differently?
Note that ‘marker’ is used here in a broad sense, and includes not only manual and non-manual markers, but also all other syntactic and morphological properties that have been found to mark imperatives in both spoken and sign languages, as discussed in the previous sections.
The imperative has been shown to be a grammaticalized sentence type in many spoken languages (Section 2.2), as well as in some sign languages (Section 2.3), with specific morphological and syntactic properties in both modalities. Some of these properties have been identified in preliminary observations in NGT (although we should interpret these with caution). Therefore, it is predicted that the imperative exists as a grammaticalized sentence type in NGT, exhibiting at least some of the (syntactic and morphological) properties identified in previous research. Moreover, NGT imperatives might also be marked in ways that have not been identified before, i.e., we should not exclude the possibility that language - or modality-specific imperative markers occur.
17 3. Methodology
This study combines the analysis of corpus data with data elicitation, in order to provide a complete picture of imperative marking in NGT. First, the Corpus NGT (Crasborn et al. 2008; Crasborn & Zwitserlood 2008) was searched for imperatives. Corpus data provide us with naturalistic, (semi-)spontaneous signing, and therefore this type of data comes close to what imperatives look like in actual language use, and it might reveal relevant variation. Section 3.1 provides more details on the corpus search. However, as there is no guarantee that the corpus search yields all the imperative types we are interested in, the corpus data were supplemented by elicited data. Data elicitation is less spontaneous and naturalistic, however, it guarantees that all imperative types are included in our data set. Section 3.2 discusses the data elicitation further. The corpus data and elicited data were analyzed in the same way; data analysis is discussed in Section 3.3.
3.1 Corpus search 3.1.1 The Corpus NGT
The Corpus NGT (Crasborn et al. 2008; Crasborn & Zwitserlood 2008) consists of recordings of 92 deaf native NGT signers (age 17-84 years), and contains over 70 hours of dialogue. Participants performed several tasks such as retelling video clips and signing about an event, but were also encouraged to sign about anything they like. Part of the video data has been transcribed using the annotation tool ELAN (Crasborn & Sloetjes 2008) by NGT signers that are fluent in the language (usually native signers), guided by the Corpus NGT Annotation Conventions (Crasborn et al. 2015). The core tiers that are annotated in Dutch are the gloss and translation tiers. Furthermore, for some clips non-manual behaviors have been annotated as well.
3.1.2 Corpus data collection
The annotated part of the Corpus NGT was searched for imperatives. Since sentence type is not annotated, it is not possible to search for ‘imperative’ in the corpus. Therefore, we searched for specific constructions on the translation tier in order to yield imperative sentences. Since imperatives are often followed by an exclamation mark in their written form, the first search was an exclamation mark on the translation tier. Moreover, a search on the translation tier for several discourse (modal) particles that commonly appear in Dutch imperative sentences was conducted (i.e., maar, eens, even, nou, toch). Finally, we searched for some verbs that commonly appear in imperatives on the translation tier (i.e., laat ‘let’, doe ‘do’, kom ‘come’, ga ‘go’, kijk ‘look’, loop ‘walk’, gooi ‘throw’, pak ‘take’). In total, these searches yielded 6.470 hits. For an overview of the conducted corpus searches, see Appendix B. Instances where the translation included an imperative with a non-literal meaning, such as bekijk het maar (comparable to English ‘go fish’), were excluded. Moreover, instances where the translation included a deontic modal construction, but the signed utterance did not, were included. Following this procedure, 60 imperatives remained for analysis.
3.2 Data elicitation
Data were elicited in collaboration with Spruijt (2019), i.e., one elicitation task was designed that elicited both affirmative imperatives (the topic of this research) and negative imperatives (the topic of Spruijt (2019)).
18 3.2.1 Participants
In total, seven participants (5 female, 2 male, mean age 49 years) were included in this study. All participants are d/Deaf native NGT signers. All participants were presented with a short background questionnaire (in written Dutch and in NGT; see Appendix E). Most (5) participants associate themselves with the sign region Amsterdam, while one participant specified Groningen and one participant Rotterdam. Three participants come from a hearing family, while the other four have at least one deaf relative. For an overview of all participants, see Appendix C.
3.2.2 Stimuli
The signers were presented with a total of 30 video-recorded signed situations, in which the response is expected to be an imperative. 12 of these situations aimed at eliciting negative imperatives (Spruijt 2019), and are not considered here. The other 18 situations aimed at eliciting affirmative imperatives. The six imperative subtypes introduced in Section 2.1 (orders, invitations, suggestions/advice, permissions, instructions, recommendations) were equally represented: three elicitation contexts per subtype were presented for affirmative imperatives, and two contexts per subtype for negative imperatives. A deaf native signer was consulted after we designed the stimuli, and based on her feedback, we changed one item.
Each stimulus consisted of two parts: first a situation was described in NGT, and then the signer was asked to sign his/her response to an imaginary interlocutor. Note that each context was designed such that it (supposedly) elicits a specific imperative subtype: compare the stimulus for an order (22a) to the stimulus for an invitation (22b). Importantly, stimuli were only presented in NGT, in order to avoid the influence of Dutch.
(22) a. PT2 PERSON3A SIGN3A++. OTHER PERSON3B PT3B THROUGH++ INTERRUPT++
SIGN++3B. REQUEST PT2 PT3B SAY3B WAIT PT3B
‘You’re signing with someone, but someone else keeps interrupting you. Please tell him/her to wait.’
b. PT2 MOTHER PT3A3AVISIT2. REQUEST PT2 ASK3A PT3A SIT.
‘Your mother is visiting you. Please ask her to take a seat.’
The 30 stimuli were presented in randomized order, which was identical for all participants. The full list of stimuli (English translations), in randomized order and specified for subtype, can be found in Appendix D.
3.2.3 Procedure
After having answered the background questionnaire (Appendix E), participants were presented with a clip that showed the instructions for the task in NGT. This clip first explains that imperatives are a sentence type, and briefly discusses the definition of imperatives. This served to ensure that participants knew what type of sentence we were looking for – we wanted them to sign imperatives, and made this explicit. Then instructions for the task were given in the same clip: participants were told that they would be presented with 30 situations, and they were asked to sign an imperative to their (imaginary) interlocutor, that followed from that situation.
The stimuli were presented in one long video on a computer in front of the participant. Participants were asked to pause the video before responding, both in the instructions and after each presented situation, so as to not be hindered by time pressure. Signers were asked to answer in one or two sentences. Finally, they were told that if they had any remarks regarding
19 the presented situation, they were encouraged to sign these following their response and before continuing the video.
Four participants were tested in a recording studio; the other three were recorded in their homes. Either one or two researchers were present for each recording session, and answered questions that participants had regarding either the procedure or the stimuli. Participants’ answers were video-recorded with their informed consent.
3.3 Data analysis
The imperatives extracted from the corpus and the elicited imperatives were analyzed in the same way. They were first divided into the six different subtypes. As it was not yet clear which formal features these different categories employ (if any), we assigned each imperative to a subtype based on its semantics, following the definitions of the subtypes provided in Section 2.1, taking context into account (as context plays an important role in disambiguating imperative types; see Aikhenvald (2010)). For instance, while (23a) was analyzed as an invitation, as it warmly encourages the addressee to come and watch the movie, (23b) was analyzed as a recommendation, as it expresses that the speaker thinks it is beneficial for the hearer to go watch the movie in the future.
(23) a. VERY FUN MOVIE SEE. COME COZY SEE
‘I’m watching a very good movie. Come and watch it (cozily)!’ [p03] b. PT1 SEE MOVIE PT3A GOOD++. PT2 REQUEST PT2SEE PT3A MOVIE GOOD
‘I saw a very good movie. (Please), you go see that movie too, it’s good!’ [p04] Some elicited sentences were excluded from analysis, as they did not fit with any of the subtypes and could not be analyzed as imperatives (for instance, deontic modal constructions and yes/no-questions). After the imperatives were divided into the subtypes, their formal features were analyzed. They were annotated in ELAN for the possible imperative markers discussed in Section 2.2, i.e.:
• Particles/Manual imperative markers (incl. syntactic position and directionality) • Non-manual markers (incl. spreading domain)
• Subject (null/overt subjects; inclusion of addressee) • Word order (marked compared to declarative sentences?) • Reduced verbal morphology (yes or no)
• Other markers (possibly not identified in previous literature)
Based on the data annotation, it was identified which of these markers occur (RQ.1), which markers co-occur and which do not (RQ.2), whether there are instances in which no marker occurs (RQ.3), and whether certain markers occur in certain imperative types, but not in others (RQ.4).
20 4. Results
This section will present results from both the corpus analysis and the data elicitation. Each subsection will start from the corpus data, and then describe to what extent the elicited data confirm the observations from the corpus, and show where they supplement the corpus findings.
4.1 Data overview
A total of 60 imperative sentences from the Corpus NGT were analyzed. These included 5 out of the 6 imperative types introduced in Section 2.1. Due to the nature of the data, only one permission and no instructions were found in the corpus. For an overview, see Table 2.
Table 2. Imperatives extracted from the Corpus NGT
Imperative type Number
Orders 23 Invitations 21 Suggestions/advice 9 Permissions 1 Instructions 0 Recommendations 6 Total: 60
Moreover, a total of 96 imperatives were elicited. These included all imperative types; an overview is given in Table 3.
Table 3. Elicited imperatives
Imperative type Number
Orders 35 Invitations 22 Suggestions/advice 9 Permissions 6 Instructions 18 Recommendations 6 Total: 96
Note that the elicited results are, again, skewed, given that (i) some items elicited an imperative type that was different from the target type (e.g., an order was signed while a permission was the target answer), and (ii) some items did not elicit an imperative at all, while others elicited more than one. For instance, one item that did not work as expected, is given in (24a). This item was meant to elicit permissions, such as “yes, go ahead” or “yes, do it”. Yet, this is not what our participants signed – a prototypical answer is exemplified in (24b). Such answers imply that the participant commands the employee to clean (well), rather than providing the employee with an authorization. Therefore, answers like (24b) were analyzed as orders and not as permissions (based on semantics and context, as was explained in Section 3.3)3.
3 The test also included a stimulus that is similar to (24a) in terms of content, but which actually aimed at
eliciting an order rather than a permission. The English translation of that stimulus is: ‘You are the boss, and he/she is your employee. His/her task is to clean the office, but he/she is doing nothing. Please tell him/her to go clean the office now’. Thus, here the signer was not asked to provide authorization for cleaning, but to actually
21 (24) a. PT2 BOSS. EMPLOYEE PERSON3A3AASK2 PT3B OFFICE CLEAN PT3A. PT2 THINK GOOD
IDEA. REQUEST PT2 TELL3A CLEAN PT3B
‘You’re the boss, and an employee asks you whether he/she should clean the office. You think that’s a good idea. Please tell him/her to clean it.’
b. SEE MESS SPACE. REQUEST GOOD CLEAN
‘I see that the space/office is a mess. (Please) clean it well.’ [p05] The next sections will focus on which imperative markers occur in the data sets (4.2), which markers co-occur (4.3), whether the marking of imperatives is obligatory (4.4), and whether there are differences between the imperative types (4.5). A more general comparison of the two data sets will be provided in Section 5.1.
4.2 Imperative markers in NGT
Table 4 shows how often each marker occurs in the corpus data, Table 5 shows this for the elicited data. The following subsections will discuss each type of marker in more detail. Table 4. Occurrence of imperative markers in 60 NGT imperatives (corpus data)
Marker Number of occurrences
(out of 60 imperatives) Manual markers - PALMS-UP - HEE - PALMS-FORWARD 174 imperatives (28%) - 13 - 6 - 2 Non-manual markers (leaning towards
addressee, furrowed or raised brows, head nod)
45 imperatives (75%)
Null subject 47 imperatives (80%)
Deviation from basic word order Unclear5
Reduced verbal morphology 1 imperative (1.7%)
command the employee to go clean. Yet, as it turns out, both items eventually elicited orders, as explained above.
4 17 sentences were marked manually; 4 sentences contained more than one marker, i.e., one sentence was
marked by both PALMS-UP and HEE (26b), one sentence contained two instances of HEE (26c), one instance contained PU twice (25c), and one sentence contained two instances of PALMS-FORWARD (27). Therefore the total adds up to 21 instances.
5 NGT basic word order is difficult to identify. The basic word order has been claimed to be SOV, however, it is
possible to diverge from the basic word order, for instance when subject and object are irreversible (Pfau & Bos 2016), or in the case of topicalization (Crasborn et al. 2009). Thus, we cannot clearly define what constitutes a ‘deviation from the basic word order’. The only order that would be clearly deviant would be VSO. Yet, not one ‘deviant’ word order was found to systematically occur in the analyzed imperatives.
22 Table 5. Occurrence of imperative markers in 96NGT imperatives (elicited data)
Marker Number of occurrences
(out of 96 imperatives) Manual markers - REQUEST - HEE - ADVICE - PALMS-UP - COME-ON - PALMS-FORWARD 546 imperatives (56.3%) - 46 - 11 - 2 - 3 - 3 - 1 Non-manual markers (leaning towards
addressee, furrowed or raised brows, head nod)
90 imperatives (94%)
Null subject 60 imperatives (62.5%)
Deviation from basic word order Unclear (see footnote 5) Reduced verbal morphology 1 imperative (1%) 4.2.1 Manual markers of NGT imperatives
First we discuss the corpus data, the majority of which does not include a manual marker: in 43 out of 60 instances (72%), there is no particle that marks the sentence as an imperative. In the other 17 instances, one (or more) of three manual markers appear. 12 sentences start or end with a PALMS-UP (PU) sign (Figure 6); in total PU appears 13 times7. This sign spatially
agrees with the locus associated with the subject of the sentence in 3 instances; when it agrees with the subject, it appears in initial position, as in (25a). In the other 10 instances, PU does not agree with the subject – in the majority (8 instances), it appears in final position, as in (25b). There is one sentence where PU is non-directional and in initial position, and in one
instance, PU appears in both initial and final position, and is non-directional (25c).
Figure 6. The PU sign in the Corpus NGT
[CNGT2045, s081, 00:02:44.600-00:02:44.800]
6 54 sentences were marked manually; 12 sentences contained more than one marker, i.e., REQUEST and HEE in 6
instances (28b), ADVICE and REQUEST in 2 instances (29), HEE and COME-ON in one instance (30a), and REQUEST and PU in 3 instances. Therefore the total adds up to 66 instances.
7 PU appears 13 times in 12 sentences – in one sentence, it occurs twice (25c); in one sentence, it co-occurs with
23 (25) a. PU2 COME ‘Come (here)!’ [CNGT2326, s091, 00:00:08.360-00:00:09.760] b. COME++ PU ‘Come/join!’ [CNGT0534, s026, 00:01:44.240-00:01:54.960] c. PU INSIDE PU ‘Come inside!’ [CNGT1836, s075, 00:00:56.880-00:00:57.715] The second marker, which occurs 6 times in 5 imperative sentences, is the attention caller
HEE8 (palm down, waving movement executed at wrist; see Figure 7). This marker can also be directed towards the subject (4 instances). It generally occurs in initial position (26ab), but in one instance, it occurs in both initial and final position (26c). Finally, in one instance, it occurs simultaneously with the rest of the sentence (26d), also shown in Figure 7.
Figure 7. HEE (signer’s left hand) in the Corpus NGT
[CNGT0529, s026, 00:01:13.120-00:01:13.920] (26) a. HEE2 DIRECTLY
‘You focus on me.’ [CNGT0294, s017, 00:06:11.432-00:06:13.412] b. HEE EXAMPLE PU
‘Give an example.’ [CNGT0539, s025, 00:00:51.560-00:00:52.365] c. […] HEE2 WAIT-A-MINUTE HEE2
‘[…] wait a minute!’ [CNGT0094, s002, 00:00:03.560-00:00:05.470] d. left hand: HEE2
right hand: A-BIT PT2 BACK
‘Go back a bit.’ [CNGT0529, s026, 00:01:12.960-00:01:13.760] Note that in (26b) HEE and PU combine; this is the only sentence in the corpus data set that
combines the two. Finally, in one instance, a PALMS-FORWARD (PF) sign is used to mark the imperative, as in (27). PF appears in both initial and final position, and the latter agrees with the subject, i.e. the palms are directed towards the addressee.
8 The prosodic structure of the analyzed examples from both data sets does not suggest that HEE can be analyzed
as a clause on its own, as imperative non-manual markers (see Section 4.2.2) spread over HEE, and it is clearly part of the imperative clause – therefore it is not translated as a discourse particle ‘hey’, which would be an alternative analysis.
24 (27) PF SELF PF2
‘You decide that yourself.’ [CNGT0518, s025, 00:01:15.350-00:01:16.350] Now we turn to the elicited data. The majority of the elicited imperatives (54 out of 96, 56.3%) is marked manually. PU occurs, but only in 3 instances. All 3 instances are sentence-final, and only one spatially agrees with the locus associated with the subject of the sentence. As in the corpus data, HEE also occurs in the elicited data, in 11 instances. In all instances, it agrees with the subject, and in all but one instance, it occurs in initial position; in one sentence HEE appears as the third sign. No combinations of PU and HEE occur in the elicited
data. Again, one instance of PF occurs, in initial position, agreeing with the subject.
Moreover, it is striking that one marker that does not appear at all in the corpus data, occurs often in the elicited data (in 46 instances), namely REQUEST. This marker is almost
always accompanied by the non-manual pouted lips, as shown in Figure 8. One participant (p07) forms an exception – he is the only signer that does not consistently pout his lips with
REQUEST.
Figure 8. REQUEST in elicited NGT data
REQUEST does not agree with the subject, as it is body-anchored. Usually (in 34 instances) it occurs in sentence-initial position, as in (28a). In 9 instances, it is the second sign of the sentence – in 6cases this is because it is preceded by another marker as in (28b). In all cases but one, it is preceded by HEE – in the other case it is preceded by ADVICE (see below; (29)). Finally, in three sentences, it is the third sign; in these instances, again either HEE or ADVICE
occurs in initial position. Thus, REQUEST combines with other markers, most often HEE, and
when they combine, HEE occurs first.
(28) a. REQUEST CLEAN-UP PT3A
‘(Please) clean that up.’ [p02]
b. HEE2 REQUEST FAST OPEN-DOOR
‘Open the door, fast!’ [p05]
Moreover, one signer uses the noun ADVICE as a marker in 2 sentences – she first signs
ADVICE before introducing the imperative. She does not seem to use it as a verb (‘I advise
you’), but rather to announce that her advice will follow, as in (29). In both instances, ADVICE
25 (29) ADVICE REQUEST DOG TAKE
‘Take a dog!’ [p04]
Finally, another marker that does not occur in the corpus data, but three times in the elicited data (by three different signers) is COME-ON, shown in Figure 9 and (30a). COME-ON occurs in
preverbal position, and is directional towards the addressee. Note that in these instances, the verb COME was used without its literal meaning, i.e., the interlocutor was not actually asked to come towards the speaker. Rather, the data suggest that COME-ON was used to mark that the
(passive) addressee is expected to go and do something. This becomes clear when we compare (30a) to (30b) – in (30a), the sister was already standing next to the door (and thus ‘come’ is not literal), while in (30b) the addressee is literally expected to come towards the speaker.
(30) a. HEE2 SISTER COME-ON2 DO PT3A OPEN-DOOR
‘Sister, open the door!’ [p03]
b. HEE2 COME SIT
‘Come (here) and sit.’ [p05]
Figure 9. COME-ON in NGT
4.2.2 Non-manual markers of NGT imperatives
The corpus data make clear that some non-manuals mark imperative sentences. An overview of the non-manuals found in the corpus data is provided in Table 6. Here we distinguish between two broad types: the position of the head and/or the body, and facial expression.
26 Table 6. Non-manuals found in 60 NGT imperatives (Corpus data)
Non-manual marker Number of occurrences
(out of 60 imperatives) Position of the head/body
- Towards the addressee - Alone
- Combined with head nod - Neutral - Head nod - Other 24 (40%) - 19 - 5 19 (31.7%) 9 (15%) 8 (13.3%) Facial expression - Neutral - Furrowed brows - Alone
- Combined with squint - Combined with mouth down - Combined with pout
- Brow raise + eyes wide - Other - Not visible 23 (38.3%) 20 (33.3%) - 12 - 4 - 3 - 1 7 (11.7%) 9 (15%) 1 (1.7%)
In almost half of the corpus imperatives, the signer leans towards the addressee. However, this non-manual marker is clearly not obligatory, as in 31.7% of the imperatives the position of the head/body is neutral. Moreover, a head nod in the direction of the addressee also occurs regularly (sometimes combined with leaning; in total 23.3%).
As for facial expression, furrowed brows commonly appear (33.3%) in the corpus data, and can combine with other non-manuals. Yet, again, this is not obligatory, as in 38.3% of the analyzed imperatives, the facial expression remains neutral. On the other hand, 11.7% of the imperatives in the corpus is marked by raised brows and wide eyes. The category ‘other’ in Table 6 includes non-manuals used to mark, for instance, topicalization or role shift (e.g. leaning towards one side). Thus, in the remainder of this study, we will consider leaning forward to the addressee (Figure 10a), head nod, furrowed brows (Figure 10b), and brow raise/eyes wide (Figure 10c) to be non-manual imperative markers (NMM in Figure captions).
Figure 10a. NMM leaning forward [CNGT0121, s007,
00:00:36.600-00:00:37.755 ]
Figure 10b. NMM furrowed brows [CNGT0210, s011,
27 Figure 10c. NMM brow raise/eyes wide
[CNGT0060, s005,
00:02:02.280-00:02:02.360]
The imperative non-manual markers on the head/body (first row in Table 6) can also combine with those on the face (second row in Table 6). In total, 45 out of 60 (75%) imperatives in the corpus occur with one or more non-manual imperative markers on the face, the body, or both – 15 imperatives occur without any imperative non-manual markers. Table 7 shows that imperative non-manual markers on the face and on the body can occur on their own (i.e., single markers), and which imperative non-manual markers on the face and body combine (i.e., combinations). Examples of the three most frequently occurring non-manual markers in the corpus are given in (31) (forward = fw in gloss; furrowed brows = fb in gloss).
Table 7. Non-manual imperative markers in 60 NGT imperatives (corpus data)
Non-manual markers Number of occurrences
(out of 60 imperatives)
Single markers
Leaning towards the addressee 10
Furrowed brows (possibly co-occurring with other non-manual markers that do not mark the imperative)
8
Head nod 6
Brow raise/eyes wide 4
Total: 28 (46.7%)
Combinations
Furrowed brows + leaning towards the addressee 9
Furrowed brows + head nod 2
Leaning towards the addressee + head nod 2
Brow raise/eyes wide + head nod 1
Brow raise/eyes wide + head nod + leaning towards the
addressee 2
Furrowed brows + leaning towards the addressee + head nod 1
Total: 17 (28.3%)
No imperative non-manual markers 15 (25%)
(31) a. head/body fw
STOP PT2 WRITE
28 b. head/body fw fb PU2 COME ‘Come (here)!’ [CNGT2326, s091, 00:00:08.360-00:00:09.760] c. fb OPEN++
‘Open the door!’ [CNGT0050, s006, 00:00:48.880-00:00:49.600] These imperative non-manual markers usually spread over the entire sentence, as exemplified in (31). Only the head nod does not spread over the entire sentence, and always occurs at the beginning of the imperative, i.e., accompanying the first sign, as in (32) (head nod = hn in gloss). Although in principle a head nod can spread, in this data set it appears to be a punctual marker.
(32) hn
PU PT2 GO
‘Go there!’ [CNGT2045, s081, 00:02:44.600-00:02:45.720] The elicited data largely confirm the corpus findings, as shown in Table 8.
Table 8. Non-manuals found in 96 NGT imperatives (elicited data)
Non-manual marker Number of occurrences
(out of 96 imperatives) Position of the head/body
- Towards the addressee - Alone
- Combined with head nod - Neutral - Head nod - Other 69 (71.8%) - 50 - 19 3 (3.1%) 17 (17.7%) 7 (7.3%) Facial expression - Neutral - Furrowed brows - Alone
- Combined with squint - Combined with pout - Brow raise + eyes wide
- Only pout during REQUEST; rest neutral - Other 5 (5.2%) 44 (45.8%) - 9 - 30 - 5 20 (20.8%) 14 (14.6%) 13 (13.5%)
Again, leaning towards the addressee, sometimes combined with a head nod, is common and even occurs in the majority (71.8%) of the elicited imperatives. However, whereas in the corpus data the signers really leaned forward, as shown in Figure 10a, this was not always the case in the elicited data – often the leaning was a bit more subtle, with the body merely turned to where the addressee was localized, without really leaning forward, as in Figure 11. Possibly, this is because in the corpus data there was an interlocutor present, while this is not the case for the elicited data (where the imaginary addressee was localized in space).
29 Figure 11. NMM turned to the addressee in data elicitation
Again, a head nod towards the addressee also occurred regularly on its own (in 17.7%of the elicited imperatives). As for facial expression, the elicited data again largely confirm what was found in the corpus: furrowed brows most commonly appear (45.8%), followed by brow raise and eyes wide (20.8%). The marker REQUEST is almost always combined with pouted lips, with the exception of one participant, as was already mentioned in the previous section. Note, however, that in the elicited data, the position of the head/body and the facial expression are neutral in a very small percentage of the data (3.1% and 5.2%, respectively), while the corpus data more often occurred without any non-manual markers (likely because of explicit instruction; for further discussion see Section 5.1).
The elicited imperative non-manual markers on the head/body (first row of Table 8) also combine with the elicited imperative non-manual markers on the face (second row of Table 8), as shown in Table 9. This, again, makes clear that a large percentage of the elicited imperatives is marked non-manually: only 6 instances occur without any imperative non-manual markers on the face, body, or both. Table 9 shows that in the elicited data, imperative non-manual markers on the face or on the body occur on their own (single markers), however, they more often occur in combination. The combination of furrowed brows and leaning towards the addressee (sometimes also combined with a head nod) occurs most often, and is exemplified in Figure 12. Moreover, brow raise/eyes wide (br/ew in gloss) combined with leaning to the addressee (33a) and pouting during REQUEST combined with leaning to the
addressee (33b) also occur rather often. Both combinations sometimes occur together with head nods.
30 Table 9. Non-manual imperative markers in 96 NGT imperatives (elicited data)
Non-manual markers Number of occurrences
(out of 96 imperatives)
Single markers
Head nod 8
Leaning towards the addressee 3
Brow raise/eyes wide 2
Furrowed brows (possibly co-occurring with other non-manual markers that do not mark the imperative)
1
Only pout during REQUEST 1
Total: 15 (15.6%)
Combinations
Furrowed brows + leaning towards the addressee 29 Brow raise/eyes wide + leaning towards addressee 9 Only pout during REQUEST + leaning to addressee 9
Furrowed brows + head nod 5
Brow raise/eyes wide + head nod 3
Only pout during REQUEST + leaning to addressee + head nod 3
Only pout during REQUEST + head nod 1
Leaning towards addressee + head nod 1
Furrowed brows + leaning towards the addressee + head nod 9 Brow raise/eyes wide + leaning towards addressee + head nod 6
Total: 75 (78.1%)
No imperative non-manual markers 6 (6.3%)
Figure 12. NMM body towards addressee combined with furrowed brows (33) a. head/body fw
br, ew
COME
31 b. head/body fw
pout
REQUEST PT3A CLEAN
‘(Please) clean it.’ [p04]
Again, the non-manual markers usually spread over the entire imperative, with the exception of the pouting lips that only occur during REQUEST. Moreover, unlike the corpus data, spreading of head nod does occur in the elicited data, i.e., in some instances the imperative is accompanied by a repeated head nod, as shown in (34).
(34) repeated hn fb
2VISIT1 NOW 2VISIT1
‘Visit me (now).’ [p05]
4.2.3 Subject of NGT imperatives
All imperatives in the corpus data address a subject referent that includes the addressee, either singular or plural. Thus, no imperatives were found in the corpus that address a third (or first) person, excluding the addressee. Yet, two types of subject noun phrases can be distinguished in the corpus: those that include the present addressee, i.e., the interlocutor of the signer (35a), and those that refer to an addressee that is not present (35bc). In this latter case, the signer is often role shifting (35b; 30 instances), i.e., the signer assumes the role of another person or character, and addresses an imaginary addressee that is currently not present. Another option is that the signer is quoting him-/herself from the past (35c; 12 instances). Thus, in the majority of the corpus data, i.e., in 42 out of 60 imperatives (70%), the subject NP refers to an addressee that is not present, which is probably due to the nature of the data: since signers are telling and re-telling stories, they are more likely to role shift and less likely to directly ask their interlocutor to do something. Role shifting will be discussed in more detail in Section 5.2.6.
(35) a. DO THEN QUESTION
‘Go to the next question.’ [CNGT0537, s026, 00:00:15.800-00:00:16.605]
b. body fw
[PT3A YELL] OPEN++
‘[She yelled at the driver] open the door!’
[CNGT0050, s006, 00:00:48.120-00:00:49.605] c. WAIT-A-MINUTE
‘[Until I yelled] stop!’ [CNGT0518, s025, 00:01:39.040-00:01:39.840] Moreover, in the majority of the corpus data (47 out of 60 sentences, i.e. 78.3%), the subject was omitted. A null subject thus is a common (but not dedicated) imperative marker in NGT, as is clear from most of the previous examples. Yet, omitting the subject is not obligatory, since in 13 cases (21.7%), the subject was overt, as in (36).
(36) STOP PT2 WRITE
