Modal clauses in Sign
Language of the Netherlands
Grammatical structure and grammaticalization
Master thesis General Linguistics
Track Sign Linguistics
By: Doatske de Haan
Student number: 10208690
Supervisor: Roland Pfau
Second reader: Vadim Kimmelman
Date: 08/12/2015
1
0. Abstract
Work on spoken languages, among which Sæbø (2011), has shown that there are five types of modal clauses (a sub type of adverbial clauses), namely conditionals, result clauses, causal clauses, purpose clauses and concessive clauses. Their structure in sign language has not been studied much, and there is no research at all on modals in Sign Language of the Netherlands (NGT). The research that does exist in this area focusses mostly on conditionals (e.g. Liddell, 1986 for American Sign Language; Dachkovsky, 2008 for Israeli Sign Language) with Wilbur (2015) being a noteworthy exception, who focusses on other types of adverbial clauses also, and on clause order. Together they show that non-manuals are important markers of clause type, at least for conditionals, and that modal clauses often appear in fixed order.
This work investigates modal clauses in Sign Language of the Netherlands (NGT), trying to answer the question what the phonological and grammatical structure of the different types are, and its possible grammaticalization paths. Furthermore, I compare the NGT patterns with those identified in previous research on other sign languages.
Using naturalistic corpus data from the Corpus NGT (Crasborn et al., 2008), I found 99 examples of modal clauses. The data show that in NGT there are four types of modals, namely conditionals, result-, causal- and purpose clauses. All types can be introduced by an overt conjunction, but this does not seem to be obligatory. Only conditionals are also marked non-manually with (a.o.) a brow raise. This discrepancy is interpreted as being the result of the preferred clause order: as in other sign languages, clause order in NGT is fixed: conditionals prefer to appear before the main clause that they modify, whereas the other types follow the main clause.
Furthermore, I show that result clauses have appeared in NGT very recently, because they are signed in two ways in the corpus; as two main clauses and as one complex clause. I argue that the former is a step before real the appearance of result clauses and that the latter are the first result clauses. This is supported by the fact that the lexical item out of which the conjunction grammaticalized, is still present in the corpus too, and looks phonologically similar to the conjunction. Nevertheless, only future research that compares my data with a future version of NGT can falsify or confirm this hypothesis.
2
Table of content
1. Introduction page 3
2. Adverbial clauses in spoken and signed languages page 4
2.1 Types of adverbial- and modal clauses page 5
2.2 Modal clauses in sign languages page 7
2.3 Grammaticalization of adverbial clauses and conjunctions page 11
2.4 Summary page 13
3. Methodology page 14
3.1 Participants page 14
3.2 Data annotation and classification page 14
3.3 Quantitative overview page 16
4. Modal clauses in NGT page 17
4.1 Conditionals page 17
4.1.1 Manual markers of conditionals in NGT page 18 4.1.2 Prosodic marking of conditionals in NGT page 20 4.1.3 Clause order of conditionals in NGT page 24 4.1.4 Grammaticalization of conditionals in NGT page 24
4.2 Result clauses page 26
4.2.1 Manual markers of result clauses in NGT page 26 4.2.2 Prosodic marking of result clauses in NGT page 28 4.2.3 Clause order of result clauses in NGT page 30 4.2.4 Grammaticalization of result clauses in NGT page 31
4.3 Causal clauses page 34
4.3.1 Manual markers of causal clauses in NGT page 35 4.3.2 Prosodic marking of causal clauses in NGT page 36 4.3.3 Clause order of causal clauses in NGT page 39 4.3.4 Grammaticalization of causal clauses in NGT page 39
4.4 Purpose clauses page 41
4.4.1 Manual markers of purpose clauses in NGT page 41 4.4.2 Prosodic marking of purpose clauses in NGT page 42 4.4.3 Clause order of purpose clauses in NGT page 43 4.4.4 Grammaticalization of purpose clauses in NGT page 44
5. Discussion page 44
5.1 Manual markers of modal clauses in NGT page 45 5.2 Prosodic marking of modal clauses in NGT page 45
5.3 Clause order of modal clauses in NGT page 46
5.4 Grammaticalization of modal clauses in NGT page 47
6. Conclusion page 48
3
1. Introduction
To date, sign languages have received comparably little attention within the field of linguistics, although it has been widely accepted that they are natural languages. Sign languages, just as spoken languages, have rich lexicons and complex grammatical systems, with for example rich morphology, and complex clauses. Nevertheless, to date we know very little about the properties of these grammatical systems in different sign languages.
This research tries to fill a small part of that gap, and zooms in on Sign Language of the Netherlands (Nederlandse Gebarentaal: henceforth NGT). In this research I will look at modal clauses, a sub type of adverbial clauses, which are subordinated by a main clause and modify the meaning of the sentence as a whole. Previous research by Sæbø (2011) on spoken languages has shown that modal clauses come in five different types, according to their semantic function: conditional clauses, result clauses, causal clauses, purpose clauses and concessive clauses.
Based on naturalistic corpus data, I attempt to answer the following research question: What
is the structure of modal clauses in NGT? More specifically, I investigate what the manual- and
non-manual properties of modal clauses in NGT are and what the clause order is and secondly, I attempt to show how, and out of which source, the different types of modal clauses grammaticalized.
In section 2, I zoom in on previous research on phonological and grammatical properties of adverbial clauses. I start with a description of adverbial clauses in spoken languages, because these have been more frequently studied, as opposed to adverbial clauses in sign languages. Secondly, I zoom in on research on sign languages. I also address the grammaticalization of adverbial clauses, in spoken languages as well as in sign languages. In section 3, the methodology of the present study is covered. Section 4 gives an overview of the results: I describe the manual- and prosodic marking and the clause order of all types of modal clauses that I find in the corpus, as well as their possible grammaticalization paths. In section 5, the results of this study will be discussed: I compare the different modal clause types in NGT with each other, and I compare the NGT patterns with those identified in previous research on other sign languages. Section 6 concludes.
4
2. Adverbial clauses in spoken and signed languages
In this section, previous research on adverbial clauses, in particular modal clauses, in spoken languages as well as in sign languages, is addressed. Adverbial clauses are different from complement clauses, because adverbial clauses can modify any sentence whereas complement clauses must be selected by the verb of the main clause as a complement. Examples of complement clauses from Dutch and English (see the translation) are given in (1):
(1) Ik dacht [dat je zou vertrekk-en].
I think.IMPF [that you should leave-INF] ‘I thought [that you were leaving]’
In both cases, the complement (between brackets) is selected as object by the verb denken (‘’think), and the left boundary of the complement is marked by the complementizer dat (‘that’). This is different from adverbial clauses, which do not have to be selected by the verb, that is, they are adjuncts. Section 2.1 gives an overview of different types of adverbial clauses and describes their function. Here, I use Sæbø (2011)’s classification. Nevertheless, this work is on spoken languages only. Therefore, I will zoom in on previous research on modal clauses in sign languages in section 2.2. Although there is not much work on modal clauses in sign languages, Liddell (1986) and Dachkovsky (2008) investigated conditionals, which are a subtype of modal adverbial clauses. The properties of conditional clauses they describe might be important for this research, too. The only study that goes beyond conditionals is Wilbur (2015). She takes Sæbø (2011)’s classification as a starting point, and describes the properties of adverbial clauses in ASL. Thirdly, section 2.3 deals with the grammaticalization of adverbial (modal) clauses and conjunctions, in spoken languages as well as in sign languages. Firstly, Heine & Kuteva (2002) describe the characteristics of grammaticalization in general, for signed as well as for spoken languages. Ohori (2011) describes the grammaticalization of subordination, and Pfau and Steinbach (2011) and Fischer and Lillo-Martin (1991) give concrete examples of the grammaticalization of conjunctions in sign languages. Finally, section 2.4 summarizes.
5
2.1 Types of adverbial- and modal clauses
According to Sæbø (2011), there are three types of adverbial clauses: temporal clauses, modal clauses, and instrumental and free adjunct clauses. I will first briefly describe temporal clauses and instrumental and free adjunct clauses, since they are not the focus of this study. Subsequently, modal clauses, which will be central to our discussion, will be discussed in more detail.
The first type of adverbial clauses, as described by Sæbø (2011), consists of temporal clauses. Temporal clauses help to situate states or events in time. They do this indirectly, by use of other states or events: the temporal clause consists of an event and through the adverb that links the temporal clause and the main clause, the event in the main clause is situated. This is shown in example (2) for Dutch, the subordinated clause placed between brackets:
(2) We zijn verhuisd [toen ik negen was].
We be.PL move.PART [when.CONJ I nine be.IMPF] ‘We moved away when I was nine’
The temporal clause between brackets situates the main clause temporally, where the temporal relation is indicated by toen. There are four subtypes of temporal clauses, namely existential when- and while-clauses, universal when-clauses, since- and until-clauses, and before- and after-clauses, but since temporal clauses are beyond the scope of this research, I will not go into detail.
The second type of adverbial clauses consists of instrumental and free adjunct clauses. These clauses can carry a wide variety of meanings, which are underspecified and depend on the context in which the clause appears. There are three subtypes of instrumental and free adjunct clauses, namely instrumental clauses, participial clauses and absolute clauses. Sentence (3) is an example of an instrumental clause, between brackets, which presents one action as an instrument of another (example translated from Sæbø (2011: 1439; his example (59)):
(3) Rosa Parks stond op [door te blijven zitten].
Rosa Parks stand.IMPF up [by.CONJ remain.INF sit.INF] ‘Rosa Parks stood up by remaining seated’
The last, and for this study most important, type of adverbial clauses which are described by Sæbø (2011) are modal clauses. Modal adverbials relate the subordinate clause to the main clause through an accessibility relation between possible worlds. There are five different types of modal clauses, but the basic meaning of modal clauses is evident in the first type, namely the conditional clauses. In
6 conditional clauses, the event that is expressed in the main clause is not presupposed, as in example (4):
(4) Ik zal afstuderen [als ik goed doorwerk]
I will graduate.INF [if.CONJ I good continue.working] ‘I will graduate if I continue working’
The subordinate clause if I continue working gives the condition that must be met for the main clause
I will graduate to be true, and this holds for all conditionals. All other types of modal clauses are
semantically related to conditional clauses.
The second type of modal clauses is result clauses. Result clauses convey a causal relation, for example (5a) and (5b):
(5) a. De portier opent de deur [zodat de gast naar binnen kan]
The porter opens the door [so.that.CONJ the guest to inside can] ‘The porter opened the door so that the guest could enter’
b. De film is zo leuk, [dat ik ook het vervolg wil zien]
The movie is so nice, [that.CONJ I also the sequel want see.INF] ‘The movie is so nice, that I also want to watch the sequel’
As becomes clear from examples (5a) and (5b), result clauses are subordinate clauses that state the result of the main clause.
The third type of modal clauses, that is, causal clauses, is semantically related to result clauses. Causal clauses can be used to answer why-question, as is shown in example (6):
(6) a. Hij is verdrietig [omdat zijn hond is overleden]
He is sad [because.CONJ his dog is die.PTCP] ‘He is sad because his dog has died’
In the main clause, the effect or consequence of the subordinate clause is given, which in turn specifies the cause or reason for the proposition in the main clause.
The fourth type of modal clauses is purpose clauses. Purpose clauses can also answer why-questions, but in a slightly different way than causal clauses. In purpose clauses, the subordinate clause gives the purpose of the proposition in the main clause. Consider example (7):
7 (7) a. Ik neem de trap [om energie te besparen]
I take the stairs [in.order.to.CONJ energy to save.INF] ‘I take the stairs in order to save energy’
The last type of modal clauses is concessive clauses. Concessive clauses, as in (8), give an idea that puts forward the opposite of what is described in the main clause:
(8) a. Hij fietst naar school [ondanks dat het regent]
He cycles to school [although.CONJ that.CONJ it rains] ‘He cycles to school, although it is raining’
The concessive clause in (8) specifies a condition for the main clause not to be true; it is raining suggests that he wouldn’t cycle; but the opposite turns out to be the case in the main clause.
Sæbø’s (2011) classification provides a detailed description of the different types of adverbial clauses, but nevertheless his work does not deal with typologically relevant aspects such as the presence of a conjunction (it is not in all languages obligatory to introduce a modal clause by an overt conjunction), the prosody of adverbial clauses, and the order of the main and adverbial clause. The next section zooms in on research that deals with the properties of adverbial clauses in sign languages, including an overview of these typologically relevant aspects.
2.2 Modal clauses in sign languages
Now that we are familiar with the subcategorization of adverbial clauses, and especially modal clauses, as proposed by Sæbø (2011), I describe what we know about their realization in sign languages. While there is some research on the structure of complement clauses in various sign languages (e.g. Tang & Lau, 2012), adverbial clauses have to date only received little attention in the sign language literature. The research that exists on modal clauses focusses mostly on conditionals, which are only a sub type of modals, while the other types of modal clauses in sign languages have barely been investigated. At least in ASL, conditionals are not necessarily signed with an overt conjunction for the conditional to be a grammatical sentence. This is because conditionals are accompanied by clear non-manuals that distinguish them from other clause types. These non-manual properties are what the first two studies that I describe focus on.
One of the first researches on conditionals was the one by Liddell (1986), who investigated syntactic preferences of adverbial clauses, focusing on non-manual properties. Liddell (1986) argues that ASL conditionals display a clear syntactic structure, that is, conditionals are always signed with the
8 same non-manual characteristics in ASL. This is contrary to what had been claimed in previous research on ASL conditionals which identified diverse non-manuals, for example Baker & Padden (1978), Coulter (1978, 1979), and Baker & Cokeley (1980); although they all find that a brow raise during the conditional clause is important. Liddell sets apart why the results in previous research vary in this respect, and argues that ASL conditionals do always go together with the same non-manual markers1.
He states that conditionals in ASL are characterized by raised eyebrows (‘br’), a rotated head position (‘rot-r’) throughout the conditional clause, and the head thrust (‘ht’) during the final sign of the conditional clause. Two of Liddell’s examples are given in (9ab) (adapted from Liddell 1986: 252):2
ht Rot-r br
(9) a. BORN GIRL NAME S-U-N-N-Y
‘If it is a girl, her name will be Sunny’ ht
Rot-r br
b. TOMORROW RAIN PICNIC CANCEL
‘If it rains tomorrow no picnic’
Liddell (1986) thus makes clear that this type of modal clauses in ASL is accompanied by typical non-manual behavior and argues that this marks the syntactic structure of the clause, that is, that facial expressions fulfil a grammatical function, and that is why they are always present.
Dachkovsky (2008) also explores conditional clauses, but she distinguishes two different types of conditional clauses, namely neutral- and counterfactual conditionals, and investigates how the difference between these two clause types is expressed in Israeli Sign Language (ISL). She thus offers a more fine-grained analysis of conditionals than Liddell. In spoken languages, counterfactuals are marked by a form that is not peculiar only to counterfactuals: English, for example, makes use of the past tense to show counter factuality: compare sentence (10a), which is a neutral conditional, and sentence (10b), which gives a counterfactual:
1 Note that, as you can see in the examples in (9), the conjunction is not necessarily expressed in ASL.
2 Signs are glossed in small caps, in English. INDEX2/3refers to a pointing sign that is used to refer to the second
or third person respectively (pronominalisation). S-I-G-N refers to finger spelled signs. SIGN-SIGN means that two words were needed to gloss one sign. Reduplication is expressed with ++. Non-manual markers are indicated by a line on top of the gloss, that marks the scope of the non-manual marker. The different meanings of the different non-manual markers are given in the text. See Pfau et al. (2012) for extensive glossing conventions.
9 (10) a. If I have a headache, I don’t go to school
b. If I had had a headache, I wouldn’t have gone to school
As opposed to English, ISL does not have overt tense markers, but Dachkovsky (2008) shows that ISL counterfactuals, too, are marked by an existing form that is also used in other grammatical domains, namely a squint. This is shown in example (11), where we provide one of the counterfactuals of Dachkovsky (2008: 74)3:
br, squint
(11) IF HE STOP SMOKE, HE LIVE
‘If he had quit smoking, he would be alive’
Dachkovsky (2008) shows that the squint distinguishes the counterfactual from the neutral conditional, since the other non-manuals that are present (e.g. the brow raise) accompany both the neutral and the counterfactual. Furthermore, she shows that the squint alone is not distinctive, that is, a squint is used to mark other clause types as well, such as topics and relative clauses. It nevertheless always has the same function, namely to show that the given information is not automatically available for the interlocutor. Therefore, according to Dachkovsky (2008), non-manuals are comparable to tones in spoken languages: the meaning on its own is general, but the meaning is obtained in a given syntactic context. For sign languages, this implies that the combination of non-manuals leads to the interpretation of the sentence.
Taken together, Liddell (1986) and Dachkovsky (2008) seem to agree on the fact that manuals are important for conditionals, but they differ in their analysis: Liddell (1986) argues that non-manuals are syntactic markers, and Dachkovsky (2008) analysis them as intonation markers.
Very recently, other types of modal clauses have received more attention in a study by Wilbur (2015), who investigates the preference for clause order in more types of complex sentences with adverbial clauses in ASL, taking Sæbø’s (2011) subcategorization of adverbial subordinate clauses as a starting point. As far as non-manuals are concerned, Wilbur (2015) confirms that they play an important role in signing complex clauses. In addition, however, she addresses clause order and shows that not only discourse factors such as information structure play a role in clause order, but that sign languages do have syntactic restrictions on word order, also in adverbial clauses. In ASL, for instance,
10 adverbial clauses have a strong preference to appear before the clause they modify4: the adverbial
clause precedes the main clause, and sentences in which the main clause precedes the adverbial clause are judged ‘Englishy’ by native ASL-speakers. Consider two of the examples of conditional clauses5
from Wilbur (2015: 9), here given as (12a) and (12b):
br
(12) a. IF ANYONE COMES, INFORM1 TAP
‘If anybody comes, let me know/ Let me know if anybody comes’ br
b. IF PETER TTY-CALL MARY, JOHN LEAVE WILL
‘If peter calls Mary, John will leave/ John will leave if Peter calls Mary’
Although the participants were given the English stimuli in both orders (main clause first versus modal clause first), the participants translated them mostly with the adverbial clause preceding the main clause, as is shown in (12a) and (12b). Wilbur argues that this cannot be analyzed as case of focus, because there are counterexamples which always appear after the main clause, such as adverbial clauses introduced by the conjunction UNDERSTAND’ (see Section 2.3). Secondly, ASL permits complement subordinate clauses to appear after the main verb that embeds them, for example in wh-clefts. What Wilbur (2015) thus shows is that adverbial clauses in ASL have their own syntactic preferences and that word order is just as important as non-manuals. Nevertheless, although Wilbur (2015)’s paper is on ‘adverbial clauses’, she does go into detail about other types of modal clauses than conditionals. As a consequence, we are still insecure whether or not word order preferences and non-manual behavior is as important for other types of modal clauses as it is for conditionals.
Taken together, although research on adverbial clauses in sign languages is limited to only three studies by Liddell (1986), Dachkovsky (2008) and Wilbur (2015), these studies clearly illustrate that that it can be a combination of various non-manuals (either analyzed as syntactic- or intonation markers) that adds the intended meaning to the signed structure, at least for conditionals. Furthermore, they show that sign languages have clear syntactic preferences, at least when it comes to conditionals. Not only do they have word order preferences, but also non-manuals are important to look at.
4 Which is often, but not necessarily, the sentence initial position.
5 Unfortunately, although her paper is on adverbial clauses in general, Wilbur (2015) does not give examples of
modal clauses other than conditionals. She does address UNDERSTAND’-clauses (Fischer & Lillo-Martin 1990) but does not specify the modal clause type. Based on the meaning of UNDERSTAND’(see page 12 for examples), I suppose UNDERSTAND’-clauses are conditionals, too.
11
2.3 Grammaticalization of adverbial clauses and conjunctions
As becomes clear from the previous section, adverbial clauses in sign languages are complex structures that are characterized by properties such as word order preference and accompanied by specific non-manuals. Cross-linguistically, complex structures tend to arise from simple structures, through a process of grammaticalization. Grammaticalization is a type of language change that is, in general, characterized by phonological erosion, decategorialization and desemanticization (Heine & Kuteva, 2002). Phonological erosion means that phonological material is lost in the grammaticalization process. In spoken language, this phonological material can for example be syllables, while in sign language, phonological erosion can reduce the movement of the sign, or turn a more complex into a less complex handshape. Decategorialization means that the word class changes from content word to function word, for example from noun to conjunction. The last characteristic of grammaticalization, desemanticization, means that the meaning (semantics) of the lexical item is ‘bleached’, which means that the meaning becomes more neutral. Therefore, the grammaticalized lexical item can be used in more contexts.
In sign languages too, grammaticalization often goes hand in hand with phonological change, characteristic non-manual behavior and the loss of syntactic privileges of occurrence. This process of grammaticalization takes a long time, and therefore different lexical items may be present at different stages of development. Ohori (2011) describes the process of the grammaticalization of subordination and states that the origin of subordinate constructions, for example modal clauses with conjunctions, can often be traced back through the semantics of the relation between the adverbial clause and the main clause. Case markers are a frequent origin of conjunctions, and so are lexical items that denote the semantic relation between the adverbial clause and the main clause.
A good example of a lexical item that grammaticalized into a conjunction because of its overlapping semantics is the ASL conjunction UNDERSTAND’.Fischer and Lillo-Martin (1990) describe the change from the verb TO-UNDERSTAND (13a)6to the conjunction UNDERSTAND’,which means something
like ‘provided that’ (13b)7. The verb
UNDERSTAND and the conjunction UNDERSTAND’differ from each other
in that the conjunction is accompanied by a non-manual marker (brow raise) and in the fact that the conjunction UNDERSTAND’ cannot take modals or subjects, whereas the verb can (i.e. decategorialization). In contrast to other types of ASL adverbial clauses (see above), adverbial clauses with UNDERSTAND’have a preference to follow the main clause, as is shown in sentence (13b):
12
hn
(13) a. ME UNDERSTAND MEAN
‘I understand what it means’
br
b. ME GO-TO STORE NOW NIGHT, UNDERSTAND’ YOU WATCH MY CHILDREN, OK?
‘I’ll go to the store tonight provided that you babysit, okay?’
Fischer and Lillo-Martin (1990) set out that UNDERSTAND and UNDERSTAND’ differ from each other phonologically as well as syntactically, thus describing the same steps as Heine & Kuteva (2002).
Pfau and Steinbach (2011) also investigated the grammaticalization of a lexical item, namely
REASON in German Sign Language (Deutsche Gebärdensprache; DGS). The noun REASON developed into a complementizer that introduces causal modal clauses, as shown in (14a) (Pfau & Steinbach, 2011: 686):
(14) b. INDEX1 SAD REASON POSS1 DOG DIE
‘I am sad because my dog died’
This grammaticalization process probably started with a structural reanalysis of multiple clauses (simple structure), such as (14b)(Pfau & Steinbach, 2011: 686):
(14) b. I am sad. The reason is: my dog died
This sentence has the same semantics as the complex clause that was to become the result of the grammaticalization process, the sentence we saw in (14a).
Pfau and Steinbach (2011) show that there are clear differences between the sentence (14a) and sentences in which REASON is not used as conjunction, for example the sentence in (14b). First, in sentence (14a) there is no prosodic break between REASON and the following sign, whereas in (14b) there is. Secondly, when REASON is used as grammatical item, the sign is phonologically reduced: the inherent repetition that is present in the lexical sign REASON is omitted so that only one movement is
left in the grammatical item REASON.This implies a grammaticalization process from the noun REASON
to the conjunction REASON.
Taken together, grammaticalization is characterized by phonological erosion, decategorization and semantic bleaching, in spoken languages as well as in sign languages. To illustrate this, Fischer and Lillo-Martin (1990) and Pfau and Steinbach (2011) give concrete examples of grammaticalization of conjunctions in ASL and DGS.
13
2.4Summary
Previous research by Sæbø (2011) shows that there is, among adverbial clauses, a subtype of modal clauses. Modal clauses are clauses in which the subordinate clause is related to the main clause through an accessibility relation between possible worlds. According to Sæbø (2011), there are five different types of modal clauses, namely conditional clauses, result clauses, causal clauses, purpose clauses and concessive clauses. Liddell (1986) and Dachkovsky (2008) show that in sign languages non-manual markers are important, although they both stick to a different analysis of non-non-manuals: Liddell (1986) for example shows that in ASL, conditionals are characterized by a brow raise, a rotated head position throughout the conditional clause, and a head thrust during the final sign of the conditional clause. He interprets this as syntactic marking, whereas Dachkovsky (2008) analyses non-manuals as intonation markers. Wilbur (2015) investigated syntactic preferences of modal clauses in ASL, and adds to previous research that in ASL, most modal clauses have a clear order preference, namely to appear before the main clause. In this research, I shall thus look at both word order preferences in NGT modal clauses, and at the non-manual markers that accompany these clauses.
Previous research on grammaticalization shows that in spoken languages as well as in sign languages, lexical items can grammaticalize to grammatical items. In this process, the syntactic context in which the word/sign can appear broadens, the word/sign may undergo phonological erosion, and the semantics become more neutral. In this study, I will also look at the grammaticalization of the conjunctions that I find in NGT, and describe the grammaticalization path (if discoverable).
Taken together, in this study I will firstly describe the manual- and non-manual characteristics of modal clauses in NGT. Thereafter, I will describe the clause order of modal clauses and relate my results to previous research sketched in section 2.2. Lastly, I will look at the possible grammaticalization paths that are at play in modal clauses, and link my findings to the previous research that has been described in section 2.3. But before discussing the results of this study, I will provide details about the methodology in section 3.
14
3. Methodology
This study is based on naturalistic corpus data which are collected in the Corpus NGT (Crasborn et al., 2008). I use a total of 35 video clips from this corpus, that all contain discussions in pairs about deaf culture and the use of sign language. This added up to a total of about one and a half hour of material for analysis8.
3.1 Participants
In the 35 clips that I analyze, a total of 22 signers participate. All signers are native signers of NGT from the Groningen region, which is located in the North-East of the Netherlands. Fourteen signers are female and eight signers are male. The majority of the signers is between 21 and 30 years old, namely ten out of the 22. Four signers are between 31 and 40, six between 41 and 50, and two are younger than 20.
3.2 Data annotation and classification
The 35 clips are analyzed in ELAN, the EUDICO Linguistic Annotator, which is a computer tool that is designed for the annotation of video clips (Sloetjes & Wittenburg 2008)9. The manual articulations of
both hands of both signers have previously been annotated by the Corpus NGT team, and most of the clips already contained a free translation of the signed texts in Dutch. Figure 1 shows a screenshot of ELAN, to give the reader an impression of what the video-linked annotation looks like:
8 The same 35 clips were used as in Oomen & Pfau (2015), who analysed the clips in terms of negation. 9 URL: https://tla.mpi.nl/tools/tlatools/elan/
15 Figure 1: Screenshot of a clip with accompanying annotations in ELAN. The clip is annotated for the manual activities of the right and the left hand of the signer on the left, the right and the left hand of the signer on the right. There are also two tiers for translations; one for signer on the left, and one for the signer on the right.
For the identification of modal clauses in these clips, I make use of the two strategies described in Oomen & Pfau (2015): Firstly, I use the search function in ELAN to search for conjunctions, such as
omdat (‘because’), als (‘if’) in the translation tiers, because this is the fastest way to identify the use of
conjunctions. Secondly, I watched all clips to check whether I missed conjunctions. I do this because the signs were often translated very freely, in a way that is more common in conversational Dutch, and I feared that conjunctions present in the signing might have been adapted or lost in the translation process. A second reason to watch all clips is to identify adverbial clauses that do not contain an overt conjunction, as we already know is the case for ASL conditionals. This might also be the case for NGT.
I glossed all examples of modal clauses that I come across in a separate word document. This allowed me to add detailed information about the prosodic and the syntactic structure of the sentence. Second, I grouped the examples according to type of modal clause. This results in a four-way classification of examples, as shown in table 1. The first group consists of conditionals or if-clauses,
16 containing the NGT conjunctions IF or WHAT-IF. The second type consists of result clauses, which include
the NGT sign RESULT.The third group of examples are the causals, which contain the NGT sign BECAUSE. The last group of examples contains the purpose clauses which are marked by either IN-ORDER-TO or
PURPOSE.Note that all types of modal clauses that are described by Saebo (2011) and in section 2.1 are attested in our NGT data, except for the concessive clauses.I did not find any instances of concessive clauses in the 35 clips that I analyzed.
3.3 Quantitative overview
The 35 clips contain a total of 93 modal clauses. The distribution of these modal clauses over the five groups is given in table 1:
Table 1: Overview of number of clauses per clause type
Conditionals 42 examples
Result clauses 21 examples
Causal clauses 26 examples
Purpose clauses 4 examples
Concessive clauses 0 examples
Total: 93 examples
As becomes clear from table 1, most instances of modal clauses in the 35 clips are conditionals, almost half of all examples found were of this type. I zoom in on conditionals in section 4.1. In section 4.2, the result clauses are addressed, and section 4.3 deals with causal clauses. I only found four examples of purpose clauses, which is not enough to draw firm conclusions concerning their structural properties. Nevertheless, I will discuss the general pattern of purpose clauses in NGT in section 4.4.
17
4. Modal clauses in NGT
In the 35 clips of the corpus, I find four different types of modal clauses. These four different types have different manual- and non-manual properties as well as different word order preferences. Therefore, I will describe all clause types in different sections, with separate subsections for manual markers and prosodic marking for each type. I also include a subsection on clause order. The last subsections of each section deal with grammaticalization of the modal clause itself, and of the conjunctions that introduce the respective modal clause. For some modal clauses the source out of which the conjunction grammaticalized is clear. This allows me to compare the conjunction with the lexical item from which the conjunction grammaticalized.
4.1 Conditionals
The first type of modal clauses that I describe are conditional clauses. Conditionals appear very frequently in the video clips, namely 42 times. Recall from section 2.1 that conditionals convey the basic meaning of all modal clauses: conditionals give the condition that must be met for the main clause, that ordinates the conditional clause, to be true.
In the corpus, I find three different ways of marking a conditional, shown in table 2: signers make use of two different conjunctions, glossed as WHAT-IF and IF,but the conjunction may also be
non-overt, which implies that the conditional is only marked non-manually.
Table 2: Instances of conditionals
Marking of conditional Number of examples
IF WHAT-IF no overt conjunction 16 15 11 Total 42
18
4.1.1 Manual markers of conditionals in NGT
There are two different signs, namely a sign glossed as WHAT-IF (STEL),of which an example is shown in (15a),and a sign glossed as IF (ALS),of which an example is shown in (15b). During the conditional, the brows are either lowered (‘bl’), or raised (‘br’). ‘blf’ indicates a body lean forward, ‘blb’ a body lean backward. A slash (/) is used to mark a prosodic break:
bl, blf
(15) a. [WHAT-IF INDEX1 NOT-KNOW SIGN] / INDEX1 ASK-AROUND
(CNGT0068:01:25)
‘If I don’t know a sign, then I ask around’
br, blf bl, blb
b. IF PARTY INDEX3A / GO3A EVERYBODY GO3A / KNOW GOOD3A
(CNGT0064:00:28)
‘If there is a party, then you go there, because everyone goes, and you know it is a good one’
Both have the same function, namely to introduce a conditional. Both signs are about equally frequent in my corpus, since I found sixteen examples of IF and fifteen examples of WHAT-IF.
WHAT-IF is signed against the chin, usually with a flat, spread 5-hand, but with the thumb usually on the palm in front of the index finger. The hand makes a short movement, in which the top-two joints of the index-finger touch the chin, with a short tap movement, and the palm of the hand facing sideways. According to the Dutch Sign Center (Nederlands Gebarencentrum), this sign usually co-occurs with the mouth gesture ‘ffff’; however, what I find is that WHAT-IF is accompanied by the
mouthing stel (‘assume’). An example is shown in figure 2:
19 The handshape of the sign seems to be a bit flexible: in most of the examples, the handshape is an opened 5-hand, but one signer signs WHAT-IF with an angled hand against her chin instead of an opened, flat hand; and there are also two signers who sign WHAT-IF with the index finger only.
The IF sign, the second conjunction that can introduce conditionals in NGT, is signed with the
index finger extended. An example is shown in figure 3. The biggest difference between WHAT-IF and IF
is the handshape: IF is signed with an index finger pointing upwards. The palm orientation is contra, just like in the WHAT-IF sign. WHAT-IF is always signed against the chin, with one small tapping movement, but the sign IF is more often signed high in neutral space, in front of the mouth, and there is no movement that accompanies this sign. However, some signers combine the handshape of IF with a
tapping movement against the chin, that is, the movement that usually accompanies WHAT-IF.The mouthing that co-occurs with IF is als (‘if’). It turned out that signers mixed up properties of both IF and
WHAT-IF: they for example used the spread 5-hand handshape of the WHAT-IF sign together with the mouthing als – or the other way around. To be able to give exact numbers of IF and WHAT-IF in the corpus, I chose the handshape to be decisive.
The third way of signing a conditional is without the use of a manual conjunction, these conditional clauses are only marked non-manually. I found this eleven times in the data, and this strategy will be discussed in the next subsection. However, before addressing non-manual marking, I briefly describe another manual sign that appears in conditionals, but in the main clause.
In spoken Dutch (or English), the main clause can be introduced by the adverb dan ‘then’, but this is not obligatory. Compare sentence (16a) and (16b)10:
(16) a. Als ik goed doorwerk, dan zal ik afstuderen
If.CONJ I good continue^working then.ADV will I graduate ‘If I continue working, then I will graduate’
b. Als ik goed doorwerk, zal ik afstuderen
If.CONJ I good continue^working will I graduate ‘If I continue working, I will graduate’
(16a) and (16b) are semantically the same, the only difference is the adverb dan ‘then’ which is not obligatory, nor adds any semantics. In my NGT-corpus, I also came across the adverb THEN, this is shown in example (17):
10 Note that it is only possible to add the adverb dan ‘then’ when the conditional proceeds the main clause:
20
rb,blf blb
(17) [WHAT-IF INDEX1 SISTER DEAF GIVE-BIRTH] THEN TOO HAVE HEARING PARENTS.
(CNGT0431:00:43)
'If my sister gave birth to a deaf baby, then he would have hearing parents too’
The adverb THEN is signed with the index-hand in neutral space, the palm facing the body and the finger
pointing aside. There is no movement that accompanies this sign, at least not in this example11.
Nevertheless, I only find one example of the adverb THEN in my corpus.
4.1.2 Prosodic marking of conditionals in NGT
In NGT, conditional clauses are prosodically all signed in the same way, with a clear prosodic structure that distinguishes conditionals from other clause types. Here we consider two prosodic strategies, namely non-manual markers and pauses (i.e. prosodic breaks).
The first thing that stands out in conditionals is the position of the eyebrows. During the articulation of the conditional clause, most signers raise their eyebrows; but some signers lower their eyebrows. What is clear, however, is that the eyebrows are not in neutral position during the conditional clause. (18a) gives an example of a conditional clause during which the eyebrows were raised, and (15a), here repeated as (18b) gives an example of a conditional during which the eyebrows were lowered12. I only found one signer that did not alter – at least not visibly – the level of his
eyebrows while signing the conditional clause. The conditional clauses are placed between brackets, so that it becomes clear during which part of the sentence the level of the eyebrows is adapted:
br, blf
(18) a. [IF INDEX1 BROTHER HEARING CHILD WANTS REMOVE] / INDEX 1 SAY LOOK INDEX1
(CNGT0060: 01:57)
‘If my brother were hearing and he would take away a deaf child, then I would say: look at me.’
11 On the website of the Dutch Sign Centre (www.gebarencentrum.nl), THEN is signed with a horizontal arc
movement, away from the body.
12 Note that (18a) gives an example of a counterfactual and (18b) of a neutral conditional. Nevertheless, the
rest of my data do not support the theory that in counterfactuals brows are raised, and in neutrals the brows are lowered.
21 bl, blf
b. [WHAT-IF INDEX1 NOT-KNOW SIGN] / INDEX1 ASK-AROUND
(CNGT0068:01:25)
‘If I don’t know a sign, then I ask around’
Except for the level of the eyebrows, namely lowered versus raised, there is no difference between (18a) and (18b), in terms of prosody. What becomes clear from these examples is that it does not matter whether the eyebrows are either raised or lowered. What matters is that the eyebrows are not in neutral position when a signer signs a conditional clause.
The second prosodic property that marks conditional clauses are body leans. Signers lean their body, head and shoulders slightly forward during the conditional. Some signers only move their head or head and shoulders forward, but there is always a forward lean visible during the conditional clause. One signer leans backward during the conditional, this example is provided in (19).
hs
br, blb
(19) [WHAT-IF INDEX1 SEE INDEX2 PT] / IGNORE / INDEX1 KEEP-ON OWN SIGN
(CNGT0068:01:48)
‘If I see a sign that I do not like, then I ignore that, then I just keep on using my own signs’
The examples make clear that it does not matter whether the body lean is forward or backward; important is that body movement is present. There is only one signer out of 22 who does not adapt the position of his body. Therefore, the body lean seems obligatory when signing conditionals.
Eleven of the conditional clauses that I found in the corpus were marked only non-manually. Two of the examples are shown in (20a) and (20b):
rb, blf
(20) a. [INDEX3 LIKE INDEX3 BEAUTIFUL]/ INDEX1 ACCEPT
(CNGT0068: 00:18)
‘If I like it, if I find a sign beautiful then I might copy it’
hs, rb, blf hs
b. [INDEX3/ FIT INDEX1 TASTE] SAY INDEX1 TASTE / OWN
(CNGT0068:00:24)
22 For these two examples, although they do not contain a manual conjunction, it is very clear that these are conditionals, since the prosody matches the prosody of other, manually marked conditionals. Therefore, both examples in (20) can be considered real conditionals, although the conjunction is not overt.
Both prosodic properties that accompany the conditional clause, namely the raise of the eyebrows and the forward body lean, are not visible in the main clause and therefore, the main clause and the conditional clause are clearly prosodically separated from each other. During the main clause, the eyebrows and body return to the neutral position. In some other examples, the eyebrows are lowered when the main clause is signed (with raised eyebrows during the conditional), which makes the contrast between the conditional and the main clause even stronger. An example of a main clause during which the eyebrows are lowered, is shown in (21):
br, blf bl
(21) [IF TOGETHER] / CHANGE OF-COURSE
(CNGT0367:01:09)
‘If they are equalled, it will of course change’
Emphasized contrast between the conditional and the main clause can also be realized by the position of the body. In that case, the signer leans forward during the conditional clause, and backward during the main clause. An example in which this happens is shown in (17), repeated here as (22):
rb, blf blb
(22) [WHAT-IF INDEX1 SISTER DEAF GIVE-BIRTH] THEN TOO HAVE HEARING PARENTS.
(CNGT0431:00:43)
'If my sister gives birth to a deaf baby, then he will have hearing parents too’
In some examples the complex structure of the sentence is emphasized by a brief pause between the conditional and the main clause. This is called a prosodic break. This means that the signers signs the conditional clause fluently, pauses briefly, and then continues to sign the main clause fluently. This is shown in example (23), in which the prosodic break or pause is marked by a /:
23 rb, blf
(23) [IF SIGN LANGUAGE ] / DUTCH GRAMMAR MIX-UP .
(CNGT0429:00:19)
‘If you only use signs, then you mix the Dutch grammar’
Still, contrary to the change in eyebrow position and body position, the prosodic break does not seem to be obligatory, since about a third of the examples consist of sentences without prosodic break. An example of a conditional without prosodic break between the conditional and the main clause is shown in (24):
rb, blf
(24) [WHAT-IF SELF NOTHING KNOW] CAN ACCEPT
(CNGT0068:02:25)
‘If you don’t know one yourself, you can accept it’
Nevertheless, as in previous examples, the difference between conditional and main clause is marked by the forward body lean and the raised eyebrows during the conditional, and the body and eyebrows in neutral position during the main clause. Therefore, the difference between the two clauses is clearly visible for the interlocutor.
There are thus two non-manuals that accompany the conditional clause, namely a brow raise and a forward body lean. Besides that, there is in some cases a prosodic break between the conditional clause and the main clause. Recall that Liddell (1986) argues that in ASL non-manuals mark the syntactic structure of the conditionals, whereas Dachkovsky (2008) analyses non-manuals as intonation markers in ISL. In both ASL and NGT a brow raise during the conditional is present, but in ASL the head position is rotated through-out the conditional clause, and the final sign is marked by a head thrust, whereas in NGT the whole body leans forward (or backward, in some cases), and there is no head thrust present at the final sign. This shows that, although NGT prefers other non-manuals in conditional clauses, NGT has clear non-manual preferences in this respect, just like ASL and ISL.
24
4.1.3 Clause order of conditionals in NGT
As becomes clear from all previous examples (15 till 24), the conditional clause always precedes the main clause. In fact, this pattern is without exception in the 42 examples extracted from the part of the corpus that I used for this study.It therefore seems like conditionals have a fixed clause order in NGT, namely the conditional clause is followed by the main clause. Recall from Wilbur (2015) that this is also the case in ASL. A similarity between previous research and my research is the syntactic preference for a certain clause order in NGT conditionals, and also the order in which the conditional and main clause appear is similar.
Taken together, although there are several lexical items that can serve the function of conjunction of a conditional clause in NGT, the grammatical and prosodic structure always shows that we are obviously dealing with a conditional. Eyebrow position (mostly raised) and body lean (mostly forward) clearly belong to the prosodic structure of conditionals, as these are found in almost all of the examples. The prosodic break between the conditional and the main clause makes the distinction between the two clauses even more pronounced, although this break does not seem to be obligatory. Lastly, in NGT the conditional clause has a clear preference to precede the main clause. This makes it plausible that conditional clauses are grammaticalized, since they have explicit syntactic preferences, also described for ASL by Wilbur (2015), and distinctive non-manual behavior, which is also described for ASL by Liddell (1986) and for ISL by Dachkovsky (2008). The next section zooms in on this grammaticalization process.
4.1.4 Grammaticalization of conditionals in NGT
The previous section shows that conditionals in NGT are real grammatical structures, but little is known about a possible grammaticalization process.
The first conjunction that is used to begin a conditional is WHAT-IF, in Dutch glossed as STEL.This conjunction sometimes even co-occurs with the mouthing stel (‘assume’). In spoken Dutch there is a verb with the same phonological form, namely stellen ‘to state’, as well as a grammaticalized version of this verb stel ‘assume’, which can be used in the way als ‘if’ can be used in spoken Dutch: in all cases,
stel can be replaced by als without change of meaning or grammaticality. Compare sentence (25a), in
which the verb is used, and (25b) in which stel ‘assume’ is used:
(25) a. Je kunt stellen dat het allemaal zijn schuld is
You can state.INF that it all his fault is ‘You can state that it is all his fault’
25 b. Stel dat het zijn fout is, dan heeft hij een probleem
Assume that.COMP it his fault is, then.ADV has he a problem ‘Assume that it is his fault, then he has a problem’
Nevertheless, there are differences between sentences with stel (‘assume’) and sentences with als (‘if’): as opposed to conditionals with als, clauses with stel always appear sentence-initially. Secondly, the use of dan in the following clause is obligatory; without the use of dan, the sentence would be ungrammatical. This makes sentences as the one in (25b) different from real conditionals in spoken Dutch.
Nevertheless, I did not find such a difference between IF and WHAT-IF in my corpus: the element THEN is not used obligatory with WHAT-IF13. Therefore the verb stellen ‘to state’ would be an appropriate candidate for the starting point of the grammaticalization process. Nevertheless, in NGT a verb with the meaning of the Dutch stellen ‘to state’ does not exist, and the sign WHAT-IF cannot be used in the way the Dutch stellen ‘to state’ can be used. So the following sentence in (26) would be ungrammatical in NGT through the use of WHAT-IF:14
(26) a. *INDEX1 WHAT-IF: INDEX1 POOR(NEGATION) RICH(NEGATION) *Let me state it this way: I am neither poor, nor rich.
Thus NGT does not have the content word that would be the most logical source for grammaticalization, as is the case in spoken Dutch. Therefore, I do not know how WHAT-IF
grammaticalized to function word, but the most logical solution seems to be that WHAT-IF is borrowed from the spoken Dutch form stel ‘assume’.
For the second conjunction that is used for conditionals, IF, evidence concerning its
grammaticalization is also scarce. The fact that there is no obvious source for the conditionals, of course does not necessarily mean that there is no source or grammaticalization path. It might even be too long ago for the traces to be visible, since grammaticalization takes a long time, and does not have an end point. The corpus data suggest that the next step on the grammaticalization path is taking place right now. Remember that next to the two lexical items that serve as conjunctions for conditionals, I also found eleven examples in which the conjunction is not overt. One of the examples (20b) is repeated here as (27):
13 Recall that I only find one example of THEN in the corpus.
14 Source: the online NGT Van Dale dictionary at www.gebarencentrum.nl. This sentence is included in the
26
(27) hs, rb, blf hs
[INDEX3/ FIT INDEX1 TASTE] SAY INDEX1 TASTE / OWN
(CNGT0068:00:24)
‘If it does not fit my taste, say it is not my taste, then I make up my own sign’
In these examples, a conditional clause is signed, without overt conjunction, but with the same prosodic and grammatical structure and non-manuals as all other conditionals. The only thing that sets the sentence in (27) apart from the other examples, is that the conjunction is not overt. It seems to become optional to sign a conjunction. This can be considered the next step of the grammaticalization process, since phonological erosion is one of the characteristics of grammaticalization.
4.2 Result clauses
The second type of modal clauses that I will address, is result clauses. In section 2.1, we saw that result clauses state the result of the main clause. In the corpus, I find 21 examples of result clauses, which are all signed using the same conjunction, glossed as RESULT. I describe the manual markers of this conjunction in section 4.2.1. The following section, 4.2.2, deals with the non-manual markers of result clauses in NGT, and 4.2.3 deals with clause order. As the gloss ‘RESULT’might lead to guess, RESULT has grammaticalized out of a lexical item – a noun – that is still present in NGT. I come across this lexical item six times in my corpus. In section 4.2.4, I compare the conjunction with the verb noun and address the grammaticalization of RESULT.
4.2.1 Manual markers of result clauses in NGT
In the NGT corpus, there is one conjunction that can introduce a result clause, glossed as RESULT (see below in footnote 12 for motivation of the gloss). An example is given in (28):
(28) MEANS TOGETHER WORKING [RESULT INFLOW++]
(CNGT0534:02:09)
‘That means working together as a group, so that there is an enormous inflow’
RESULT is signed with two hands, with the same handshape, namely a fist with an extended thumb, which points upwards. The palm of the right hand is oriented towards the left side, and the palm of the left hand is oriented towards the right side. The hands are placed in neutral space behind each other; the non-dominant hand is close to the body and the dominant hand is placed in front of it,
27 touching the non-dominant hand, little further away from the body. This is made more concrete in figure 4, in which a representative example is given:
Figure 4: Signer signing RESULT
The movement is an arc movement from the wrist, with both hands, and the mouthing that goes together with RESULT is gevolg (‘result’)15.
I find considerable variation with regards to the phonological form of this sign. Firstly, in some examples the location is adapted. RESULT is usually signed in neutral space, in front of the chest (as
shown in figure 4); but in two examples (both from the same signer) RESULT is signedbefore the ipsilateral shoulder. Both times, this goes together with an adaptation of the movement, which I will describe below. In two other examples, the hands are placed next to each other, with both palms and the bended fingers touching, whereas the hands are usually placed behind each other. Furthermore, I find some handshape adaptations in the corpus. The handshape is usually present in both hands, but I also came across two examples in which the non-dominant hand is a fist, without extended thumb. There is even one example in which the signer uses a different handshape, namely the 1-hand. In this case, the index finger is extended instead of the thumb, and the index finger makes the movement the thumb usually makes. This example is given in (29):
(29) NOW SELF EASY MAKE-APPOINTMENT MAYBE / GROUP INDEX3A [RESULT(IX) GROUP INDEX3B DECREASE] SELF PITY
(CNGT0064: 00:36)
‘Now you can easily date each other, as a group, so that the group (deaf community) decreases, I pity that.’
15I thought about glossing RESULT as FOLLOW’,after Fischer & Lillo-Martin (1990), but I decided against it because
of the mouthing that accompanies the sign (gevolg ‘result’) and the fact that there is a not only a verb but also a noun out of which RESULT grammaticalized: see section 4.2.4.
28 In this example, the 1-hand is used to point from location 3a to location 3b. The use of the 1-hand in this example emphasizes the intention of the speaker.
Next to adaptations of location and handshape, there are also signers who change the movement pattern. In eight out of 21 examples, the forward turning movement is not present in both hands. In six cases only the dominant hand, which is placed in front of the non-dominant hand, moves; in two examples, the movement is totally absent. An overview is given in table 3:
Table 3: Movement patterns in RESULT
Movement pattern Number of examples
Movement present in both hands 13 Movement present in one hand, absent in one 6
Movement absent in both hands 2
Total 21
I also find three examples in which the movement is not forward, but upwards. In these examples, the signer does not start with the thumb pointing upwards, but with the thumb pointing to the chest: the starting position of the sign is thus slightly different. From there, the thumbs move upward, into the position in which RESULT usually starts. In two out of three examples, this movement is present in both hands, and in the third example, this movement is present only in the dominant hand. In two out of three times, this adaptation of movement goes together with a change of location: these examples are signed in front of the shoulder instead of in neutral space.
These phonological variants notwithstanding, it is clear that there is only one conjunction in NGT that can introduce a result clause, namely RESULT.It seems to be obligatory to sign this conjunction, since I do not find examples in which RESULT is non-overt, but it is not always present in its full form.
Often the movement is reduced, and sometimes even the handshape is not present in both hands, or the location is slightly changed.
4.2.2 Prosodic marking of result clauses in NGT
The previous section shows that in all 21 examples of result clauses that I come across in the corpus, the conjunction RESULT is present. (30) gives two examples of result clauses in NGT:
29 (30) a. TECHNOLOGY GROW-UP^WITH [RESULT 1 INDIVIDUALISTIC /2 CULTURE AMERICA INFLUENCE (….)]
(CNGT0259: 01:39)
‘They grow up with technology so that they are firstly individualistic and secondly there is influence of American culture.. ‘
b. BASIS ‘WATER’ CONNECTION-INDEX3A CONNECTION-INDEX2 CONNECTION-INDEX3B BECAUSE HAVE hs
MORE [RESULT LOOK MUCH BUT NOT]
(CNGT0861:00:33)
‘This sign for ‘water’ is the basis, and there are relations (with other signs for ‘water’), because they have more signs so that it looks more but that is not the case’
These two examples give a good impression of what most result clauses look like in NGT. The two examples give result clauses in which there is no prosodic break between the main clause and the result clause, and this is a good clue to analyzing them as being embedded: signers usually make a short prosodic break in between two main clauses but this does not happen in these result clauses; they appear as one unit. Therefore, I want to argue that at least these result clauses, which appear in a fixed order and without prosodic breaks before (or after) the conjunction, are embedded clauses.
Nevertheless, in ten out of 21 examples, there is a clear pause before RESULT.Examples (31a) and (31b) give two examples in which this is the case:
hs (31) a. HERE GUYOT SVO REALLY-WANT BILINGUALISM /[RESULT ONLY SIGNING NOT^ALLOWED]
(CNGT0429:00:09)
‘Here at Guyot, at SVO, they aim at bilingualism. Therefore, signing only is not allowed’
b. BUT STILL STAYS NORMAL EDUCATION INDEX3A LEVEL SPECIAL EDUCATION INDEX3B LEVEL LOWER /
hs
[RESULT PEDAGOGICAL ATTENTION-3A MUST ATTENTION-3B]
(CNGT0258: 00:27)
‘But still they (the children) stay in special education and go to normal education, the quality and the level of education are not good enough. Therefore, the attention goes to the quality of normal education, while they should pay attention to special education’
30 In these examples, firstly the main clause is signed, then the signer pauses, and after the pause he continues signing the rest, starting with RESULT. The prosodic break is always placed before the conjunction RESULT and never after. I also find three examples out of 21 in which there is an extended pause after RESULT,and in all three examples this pause can be interpreted as a filled pause, comparable to ‘ehm’ in spoken languages, since the fingers wiggle during this break. One of the examples in which there is a pause after RESULT is given in (32), in which the extended pause is indicated with // (compared
to /, which indicates a normal prosodic break):
hs
(32) HEARING NOT-PRESENT / INTERESTING BECAUSE DEAF WORK DEAF^WORLD FOR-EXAMPLE GUYOT AMSTERDAM SCHOOL EDUCATION // RESULT // SIGNS HAVE CLUB^HOUSE INDEX3A FEELING LESS
(CNGT0259 03:34)
‘Hearing people are not present there. That is interesting, because the deaf who work in the deaf world, for example here at Guyot, in Amsterdam in a school or in education. Therefore the need to go to the club house decreases’
This example shows that the signer made a pause to think about his intentions: the hands of the signer are in the position of the upcoming sign, the signer hesitates, and then continues signing. In example (32), this happens not only after the RESULT sign but also before the sign.
It seems like there are two types of result clauses; one with, and one without prosodic break between the main clause and the result clause. The former can clearly be analyzed as an embedded clause, nevertheless the status of the latter is less clear: the conjunction is present16, which usually
introduces a subordinate clause; but also a prosodic break, which usually separates two main clauses. Unfortunately, I did not find any other non-manual properties, such as brow raises or body leans, to accompany result clauses in NGT. The only other property that result clauses seem to have is a fixed clause order: this is addressed in the next section.
4.2.3 Clause order of result clauses in NGT
Although there do not seem to be fixed non-manuals that accompany a result clause, there is a preference for clause order. The main clause and the result clause always appear in a fixed order, namely the result clause tends to follow the main clause. Wilbur (2015) also shows that this fixed clause order is a property of adverbial clauses in ASL, although in ASL the main clause follows the adverbial
31 clause, and for result clauses in NGT this seems to be the other way around. Nevertheless, Wilbur does not address result clauses: the only type of modal clauses she addresses is conditionals.
Also the clauses which have a pause before the conjunction, stick to this fixed clause order: they always appear after the clause of which they state the result. Nevertheless, this, together with the presence of the conjunction, is the only clue to analyzing these clauses as real result clauses, since the main- and the RESULT-clause are not signed as one unit due to the prosodic break in between them.
Since these clauses are not really embedded, but also not really separable due to their fixed clause order and the appearance of a conjunction, they seem to lag behind the result clauses without prosodic break, from a grammaticalization point of view17. I will discuss this in the next section, after a
description of the verb RESULT and the noun RESULT(n).
4.2.3 Grammaticalization of result clauses in NGT
For RESULT, the source out of which the conjunction grammaticalized is obvious. NGT has a noun, for distinctness glossed as RESULT(N1) (ge-volg ‘result/consequence’). This noun has about the same
phonological form as TO-FOLLOW and the conjunction RESULT,but can also appear with a slightly different phonological form, glossed as RESULT(N2). In the corpus, I came across six clear examples, two of them shown in (33):
(33) a. INDEX1 THINK ALWAYS CAN LOOK-AT RESULT(N1)++ ‘I think you can always look at the results’
br b. THUS RESULT(N2) PALM-UP?
‘Thus what is the result?’
There are two ways of signing RESULT(N).The first way is to sign RESULT(N1) just like the conjunction
RESULT, with both hands that have the same handshape, a fist with an extended thumb pointing upwards. The palms of the hands are oriented towards each other’s side, placed in neutral space, with the non-dominant hand in front of the dominant hand. The accompanying movement is an arc movement from the wrist, with both hands. Four out of six examples in my corpus are of this kind. The second way to sign RESULT(N2)is with both hands in an index-hand, with the index fingers pointing side wards towards each other. Again, the hands are placed in neutral space, with the non-dominant hand
17 This is the reason why I chose to translate the result clauses without prosodic break with ‘so that’, an English
