<<

This is the pre-print, author accepted manuscript of the paper published in the journal Sign Language and 21:2 (2018), 257-283. https://doi.org/10.1075/sll.00020.kel. It is under copyright and the publisher should be contacted for permission to re-use or reprint the material in any form.

Agent-backgrounding in Turkish Sign Language

* Corresponding Author: Aslı Özkul. [email protected], İstanbul Bilgi University, Teaching Department, Istanbul, Turkey.

Author Names & Affiliations

Meltem Kelepir Boğaziçi University Department of Linguistics Istanbul, Turkey

Aslı Özkul İstanbul Bilgi University English Language Teaching Department Istanbul, Turkey & Boğaziçi University Department of Linguistics Istanbul, Turkey

Elvan Tamyürek Özparlak Boğaziçi University Department of Linguistics Istanbul, Turkey

List of digital files or other materials (video): - - Agent-backgrounding in Turkish Sign Language

Abstract This paper reports our observations regarding the constructions in Turkish Sign Language (TİD) in which agents are backgrounded. The data have mainly been elicited based on the questionnaire developed by Barberà and Cabredo Hofherr (see the introduction to this volume) designed to identify strategies and constructions used in sign languages for backgrounding agents. We have observed in TİD many of the agent- backgrounding strategies reported in the literature that signed (and spoken) languages employ. Use of non-specific indefinite pronominals is a major strategy, and this paper is the first study that identifies these forms in TİD. Moreover, we show that TİD, has a special sign that derives exclusive non-specific indefinite pronominals, OTHER. Finally, we argue that in TİD whereas lateral high R-locus is unambiguously associated with non-specificity, non-high (lateral and central) loci are underspecified in terms of specificity. We claim that the R-locus of indefinite pronominals observed in impersonal contexts in TİD in fact consists of two spatial features [+high] and [+lateral] which correspond to the pragmatic features non-specific and exclusive.

Keywords: impersonal agents, agent-backgrounding, indefinites, non-specificity, , high R-locus

1 Introduction

Studies which investigate how agents are backgrounded in languages have shown so far that there are four main strategies: (i) -reducing operations, (ii) null- subjects of non-finite predicates, (iii) impersonal uses of personal pronouns, and (iv) dedicated referentially deficient pronouns. The strategies in (iii) and (iv) are categorized as R(eference)-impersonals (Siewierska 2011 and Barberà & Cabredo Hofherr this volume). This paper mainly focuses on the strategy in (iv) TİD, namely, dedicated referentially deficient pronouns, but it also contains brief discussions of the other strategies. TİD is widely used by the Deaf in the Republic of Turkey and its speakers are estimated to be around sixty thousand, as stated in the results of Turkish Disability Survey (TDS 2002) (Kemaloğlu & Kemaloğlu 2012:66-67). Miles (2000) reports that a sign language was used in Ottoman palaces and courts, but it is not clear whether the TİD spoken today has descended from that language (Zeshan 2002, 2003). No claim has been made so far regarding the genetic relationship of TİD to any of the sign languages spoken in the rest of the world. Thus, for the purposes of this paper, we assume that it is not related to any of the other languages described in this volume. The data analyzed in this paper have been collected from three native and seven fluent Deaf signers of TİD who reside in Istanbul. Three of them are male and seven of them are female. Their average age at the time of this study was 38. Their education levels range from primary school to university.

1 Data elicitation was mainly conducted following the methodology of the R- impersonals questionnaire adopted by Barberà & Cabredo Hofherr from the Paris-Jena project Towards a typology of impersonal human pronouns. This questionnaire contains different contexts aimed at eliciting different types of impersonal constructions, and the informant is asked to translate the given after the context is provided in written local spoken language in the most natural way into his/her sign language (see Barberà & Cabredo Hofherr this volume). In our study, we adapted these contexts to Turkish culture: some contexts were slightly changed and some new ones were developed. All of these contexts were written in Turkish in a slide presentation file to be presented to the informants on a laptop computer. One of the co-authors of this paper is a fluent Deaf signer of TİD, and she was involved in the development of these written contexts. The elicitation task with contexts was later carried out with the Deaf researcher and eight (one male and seven female) informants, with the following procedure: the informant first read the context provided in Turkish on the slide, and when (s)he understood the context, (s)he moved to the next slide which had a sentence in Turkish and translated it to TİD. When necessary, the Deaf researcher helped the informant understand the context and the Turkish sentence. Their responses were recorded with three cameras. In addition to the elicitation tasks conducted based on the questionnaire, further elicitation and grammaticality judgment tasks were carried out to clarify certain points that came up during the analysis of the data. These later judgements are based on the utterances and judgements of the Deaf co-author of this paper and one male native and one male near-native signer who did not take part in the questionnaire tasks and two female informants who had already taken part in the questionnaire. Agent-backgrounding strategies that we have observed in TİD can be summarized as the following: a. no overt expression of the agent b. null subjects interpreted as impersonal 3rd person pronouns c. 1st and 2nd person pronouns with generic uses d. indefinite agents expressed with pronouns and/or In Sections 2, 3, 4 and 5 below, each of these strategies is discussed with reference to their frequency of use in the impersonal construction contexts mentioned above. Needless to say, in almost all of the contexts, there was not a single strategy that all the signers adopted. However, some strategies were more predominantly adopted for some of the contexts. We should also note that these preferences were possibly strongly influenced by the strategies used in the Turkish sentences that the signers were asked to translate1. For instance, overt non-specific indefinite pronominals were predominantly elicited when the sentence to be translated contained the Turkish counterpart birisi 'someone'. The findings are summarized and the theoretical and typological implications discussed in Section 6.

2 2 No overt expression of the agent

Many of the utterances elicited lacked an overt that could be identified as the . Since it has been observed that null arguments are common in sign languages, one possible approach to the structure of such sentences could be to analyze them as having null subjects (Özsoy & Branchini (in press)). Note, however, that in sign languages such as American Sign Language (ASL), where the licensing of null arguments has been studied extensively, it has been shown that they have to be anaphoric (Koulidobrova 2017:406), namely, they have to have an antecedent in the discourse (see also Bahan et al. 2000, who argue that only manual/non-manual agreement licenses null arguments in ASL). No analysis of licensing of null arguments in TİD is yet available; however, keeping in mind the possibility that anaphoricity may be a requirement in TİD as well, we divide the sentences in our data with no overt agents into two categories: (i) sentences in which the referents of the unpronounced agents can be recovered from the context, and (ii) sentences in which the referent of the unpronounced agent is less obvious. Examples of the first category are sentences elicited in generic conditionals, locative universals, and contexts that trigger corporate readings. These will be illustrated in 2.1. below. Examples of the second category are episodic sentences elicited in contexts where there is no clue as to the referent of the agent. We discuss these in 2.2. and raise the possibility that these may involve agent- reduction rather than null subjects. Future research on the licensing of null arguments in TİD will hopefully shed more light on the issue.

2.1 Null subjects

In our data, null subjects are frequently observed in generic conditionals, locative universals, and corporate readings (see Barbera & Cabredo Hoffher, this volume, for an explanation of these notions). These are illustrated in (1) below.

(1) a. Generic conditional Ø ICE-CREAM EAT SICK BE IX2 ‘If you eat ice-cream, you will get sick.’

b. Locative universal FRANCE Ø SNAIL EAT ‘In France, they eat snails.’

c. Corporate reading Ø TAKSIM MOSQUE BUILD ‘They will build a mosque in Taksim.’

These sentences were elicited as translations of active Turkish sentences. Even though there was no antecedent for the null subject in the discourse, one could argue that the referent is recoverable from the given context: in (1a), a generic conditional, the

3

nd consequent clause contains a 2 person singular pronoun, IX2, so the null subject in the antecedent clause can be understood to be co-referential with it. (1b), a generic statement, starts with a locative phrase, FRANCE, and thus, the referent of the subject can be understood to be the people living in France2. Finally, (1c) involves the construction of a new mosque, enabling a corporate reading, and the null subject can be understood to be a government institution responsible for such constructions (see also 2.5 in the introduction to this volume). Note, however, that we represent these examples in (1) as active transitive sentences with null subjects, since (i) they are translations of active transitive Turkish sentences and (ii) the referent of the subject is recoverable. However, there is no convincing evidence at the moment that excludes the possibility of a passive or any other agentless construction instead of null subjects in these examples. The sentences discussed in the next section have episodic (and not generic) readings, and in to the examples in (1) above, the contexts do not provide a clue as to the identity of the agents.

2.2 Agent-reduction

Agent-reduction is considered to be one of the agent-backgrounding strategies, perhaps the most radical one. The examples in our data that are potential candidates for agent-reduction were mostly produced in vague existential and indirect evidential contexts (see Barbera & Cabredo Hoffher, this volume, for an explanation of these notions). These display properties similar to the structures we called “passive-like”3 in earlier work (Kelepir & Özkul 2015). The examples discussed in this section, taken from Kelepir & Özkul (2015), were not elicited as translations of Turkish sentences. Rather, the consultants were either given contexts in written Turkish and asked what they would say in that context, or they were presented with a picture and asked to describe it. In the contexts given and the pictures presented, there were no clues as to the referent of the agent. The following example was elicited in a context where the signers were told that while they were looking for a place to buy in a neighborhood they see a “for sale” sign on the window of an apartment in a building, they like the building and the area, and go back there the next day to take a look at it and realize that the sign has been taken away. They were asked how they would express this situation later to a friend.

(2) bn HOUSE SELL ‘The house has been sold.’

(3) was elicited as a response to two pictures depicting a breakfast table. The first picture shows the food on the table, untouched, whereas the second picture shows the same table with food partially eaten and drinks partially drunk.

4

(3) CL.BREAD.CUT FINISH ‘The bread has been cut.’

As illustrated in (2) and (3), these constructions typically contain either the perfective marker, a mouth gesture ‘bn’, or the completive marker, FINISH (see Karabüklü 2016 for an analysis of these markers), or both. These aspectual markers contribute a resultative reading to the construction (Sze 2010). (4) provides a further example, with another perfective marker that occurs in negative contexts, puffed cheeks ‘pc’ (Karabüklü 2016) (together with the negation marker backward head tilt ‘bht’)4.

(4) bht pc CL.BREAD.CUT.neg ‘The bread has not been cut.’

Similar examples have been discussed for other sign languages. Sze (2010) shows for Hong Kong Sign Language (HKSL) that when the agent is not known and the is inanimate, signers tend to express these situations with resultative constructions containing the aspectual markers ALREADY and FINISH. Barberà & Cabredo Hofherr (2017) identify a “non-agreeing central construction” in Catalan Sign Language (LSC) whereby the agent is left unexpressed, the patient is inanimate, the verb is articulated in neutral space in front of the signer, and the verb is often followed by what they call the resultative marker ALREADY5. They argue that this construction is an intransitive non-active construction comparable to a middle verb that may not imply an agent. See Kelepir & Özkul (in preparation) for a more detailed discussion of this construction in TİD.

3 Null subjects interpreted as impersonal 3rd person plural pronouns

In this section, we illustrate cases where there is no overt noun phrase expressing the agent, but there are other clues indicating that the sentence is interpreted as having an agent. The first type of such examples contains what looks like a 3rd person plural on the verb (3.1.) and the other type involves mouthing of the Turkish verb with 3rd person plural inflection (3.2.). In both cases, the agent is interpreted as impersonal.

3.1 Plural marking of the verb

In some of the utterances triggered by anchored existential contexts, the signers produced a form immediately following the verb and expressing the plurality of the external , as in (5) and (6). This form has an arc movement and can be articulated mono-manually as in Figure 1 or bi-manually as in Figure 2.

5

(5) FEDERATION PRESIDENT3a TEN MINUTE LATE SAY IX3a DRUNK LATE Ø UNDERSTAND-3pl 'The president of the federation was ten minutes late and they/people (impersonal) understood that he was drunk.'

Figure 1. UNDERSTAND 3pl

The example in (6), in addition to the 3rd person plural marking of the verb, contains a mouthing of the Turkish verb yapıyorlardı (make-prog-3pl-past) ‘they were making’. The pronunciation of this Turkish verb is provided between slashes in the non-manual tier.

(6) (IX1 CHILD OUR HOUSE NEAR PARK EVERY EVENING BARBECUE) __/japıjorlardı/ Ø MAKE-3pl 'When I was a child, in the park near our house, they/people (impersonal) used to have barbecue every evening.'

Figure 2. MAKE 3pl

The glossing in the examples reflects an assumption that this form, 3pl, is a suffix. However, given the scarcity of affixes in sign languages, this has to be analyzed in detail in future work.

6

3.2 Mouthing of the Turkish verb inflected for 3rd person plural

In all sub-contexts of existentials (i.e. vague existentials, anchored existentials, indirect evidentials), two strategies stand out: (i) non-specific indefinite pronominals meaning ‘someone’ and (ii) in utterances with a null subject, the mouthing of the Turkish verb inflected for 3rd person plural, as exemplified in (6) above and (7) below. This example was elicited in a context where the signer enters a hotel room and sees the remnants of a cigarette. The signer mouths the Turkish verb içmişler (smoke- perfective-3pl) ‘they have smoked’.

(7) /iʧmiʃlɛr/ HERE Ø SMOKE 'Here they (impersonal) have smoked.'

4 First and second person pronouns with generic uses

As described in the introduction to this volume, some languages (spoken and sign) allow generic uses of 1st and 2nd person pronouns, while others do not. The following example from TİD clearly is generic and the subject of the main clause is a 2nd person singular pronominal (IX2).

(8) Ø ICE-CREAM EAT SICK BE IX2 ‘If you eat ice-cream, you will get sick.’

However, it is not clear whether the signers interpret IX2 as non-deictic, i.e. impersonal. After all, this could be a generic statement for the addressee. Thus, in order to make sure that IX1 or IX2 do not have deictic readings, we developed gender-specific contexts such as "When/If I am a groom, I have to pay for the bride's hairdo on the wedding day" uttered by a female signer, as in (9), or "When/If you are pregnant, you should not lift heavy objects" addressed to a male interlocutor, as in (10). These are accepted as grammatical by our informants in the given contexts, and our consultants translated these statements with conditionals. Preliminary observations on adverbial clauses in TİD (Özkul, in preparation) indicate that head thrust is a common non-manual marker in both conditional and “when”-temporal clauses, whereas brow raise occurs only in the former. Moreover, irrealis conditionals are expressed with repetitive head nod in addition to brow raise (and head thrust). Since the examples below that we have elicited in the generic conditional contexts contain only the non-manual markers of realis conditionals, namely brow raise and head thrust, but not the additional non- manual marker observed in irrealis conditions, namely repetitive head nod, we conclude that the first and second person pronominal forms in (9) and (10) function as impersonal pronouns.

7

br ______hth (9) IX1 GROOM BE, IX1 PAY HAIRDRESSER ‘If I am a groom, I pay for the hairdresser.’ (uttered by a female signer)

br ______hth (10) IX2 PREGNANT, HEAVY SOMETHING LIFT NOT NECESSARY ‘If you are pregnant, you should not lift something heavy.’ (addressed to a male signer)

5 Indefinite agents: pronouns and/or verb agreement

In this section we discuss the different types of non-specific indefinite pronominals, including a clusivity distinction, and expressing impersonal agents through verb agreement.

5.1 Non-specific indefinite pronouns

It is generally assumed that speakers use indefinite noun phrases when they presuppose that their addressee can not identify the entity they are talking about. Indefinites are usually further categorized as specific and non-specific. The term “specificity” has been used in the literature to express different notions (see Kelepir 2001; von Heusinger 2002). In one sense of the term, the referent of a specific indefinite is identifiable by the speaker (i.e. it refers to an entity the speaker has in mind), whereas that of a non-specific indefinite is not (Fodor & Sag 1982, among others). Other uses of the term “specific” include wide scope interpretation and partitive interpretation (see Enç 1991; among others). In this paper we use the term “non-specific” in the first sense explained above, namely, referring to cases in which the referent of the indefinite pronoun cannot be identified by the speaker/signer. Non-specific indefinite pronouns are commonly used in agent-backgrounding contexts (Barberà & Quer 2013). It has been reported that some sign languages may mark the distinction in specificity through the articulation of the indefinite noun phrase in different R-loci (high vs. low) and/or through the absence/presence of certain non-manual markers (Barberà 2012, 2015). No comprehensive analysis of TİD indefinites is available yet; however, we discuss here our preliminary observations regarding the non-specific indefinite pronominals that our consultants produced during the elicitation tasks. We have observed that in TİD an indefinite articulated in high R-locus is only interpreted as non-specific, whereas one articulated in central-low locus may be interpreted as specific or non-specific. The distinction is expressed with non-manual markers, i.e. non-specific indefinites are articulated with brow furrowing, lowered mouth corners and averted eye gaze, whereas these are absent on the specific indefinites (see Barberà 2016:26; Mantovan 2017; Barberà & Cormier 2017, among

8 others, for non-manual marking of indefiniteness, (non)-specificity, and uncertainty in different sign languages).6 Below we provide an example of an indefinite form we gloss as ONE^PERSON^C_PERSON7 'someone', which is interpreted as non-specific with the accompanying non-manual markers.

Figure 3. ONE PERSON C_PERSON

This noun phrase has been taken from an example elicited with the context in (11a). 'Non-sp' in (11b) subsumes all the non-manuals associated with non-specific indefinite reference described above.8

(11) a. Context: “You are at a friend’s house. You see the red flash telling you that someone is ringing the bell. The host does not seem to see it. What do you say to your friend?":

b. ______non-sp ONE^PERSON^C_PERSON DOORBELL PRESS ‘Someone is ringing the door.’

Regarding the morphology of the form, ONE^PERSON^C_PERSON 'someone' consists of the indefinite determiner ONE, and two other morphemes: PERSON and C_PERSON (Figure 3). PERSON has the narrowed-O-handshape (Kubuş 2008:158) and is articulated on the cheek with two quick movements. This sign is considered to be a lexical item, a noun expressing the meaning 'person'.9 C_PERSON, on the other hand, has the TİD C-handshape (ASL small-C; Kubuş 2008:156) and is articulated with a downward movement in the frontal plane. This sign is not listed as a lexical item in dictionaries but has been claimed to derive agentive nouns in forms such as WORK^C_PERSON 'worker' (Taşçı & Göksel 2014:12-13; Dikyuva et al. 2015:151).10 ONE^PERSON^C_PERSON 'someone', is interesting at least in two respects: (i) there are two signs that are reported to express the meaning 'person' and (ii) the existence of this form shows that C_PERSON is not only used to derive deverbal nominals. The next question is whether this form should be considered a noun phrase with an (non-specific) indefinite determiner and a head noun (perhaps a compound

9 head noun, PERSON^C_PERSON) or a morphologically complex pronoun (in the form of a compound) occupying the head position of the noun phrase.11 A discussion along these lines would take us outside the scope of this paper, but suffice it to say that all three of these morphemes "blend" into each other during articulation and cannot be separated by, for instance, an adjective, as in *ONE OLD PERSON^C_PERSON with the intended meaning 'someone old', and even though PERSON is articulated with two quick movements on the cheek when it functions as an independent lexical item meaning ‘person’, it is articulated with one quick movement when it occurs in the complex ONE^PERSON^C_PERSON, suggesting the possibility that the latter could be a morphologically complex pronoun such as someone in English12. To summarize, ONE^PERSON^C_PERSON 'someone' is a frequently used indefinite form. It may be interpreted as specific or non-specific depending on the non- manual markers accompanying it.13 In the next section, we discuss a further distinction within non-specific indefinites: clusivity of the addressee, which is expressed by means of different R-loci in the signing space.

5.2 Exclusive-inclusive distinction in indefinite pronouns

Barberà & Cobredo Hofherr (this volume) mention that some (spoken) languages have impersonal pronouns that exclude the speaker and others that include the speaker. The exclusive-inclusive distinction we report in this paper for TİD agent-backgrounding constructions, however, is not related to the exclusion/inclusion of the speaker but of the addressee (and a set of people salient in the discourse). The forms that express the exclusion of the addressee are preferred in contexts where the signer would like to convey her/his presupposition that the unknown agent is from outside the group of a set of people salient in the discourse, such as the addressee, the co-workers in the department or the lab, family members, etc. These contexts usually involve inappropriate acts such as throwing bottles into the sea, leaving trash on a desk, stealing, smoking etc., which coincidentally have constituted most of the contexts in our study. We call this "the exclusive someone reading". In these contexts a signer may prefer not to use the neutral indefinite pronominal ONE^PERSON^C_PERSON. Expression of such exclusion is achieved with at least three different strategies: (i) the use of an indefinite form, lexically specified for exclusivity, containing the morpheme OTHER, (ii) the articulation of an indefinite pronominal in lateral-high R- locus, ONElat-high and (iii) the articulation of an agent agreement morpheme in lateral- high R-locus.

5.2.1 OTHER OTHER expressing a non-specific agent can occur by itself or more frequently, together with ONE or with ONE^C_PERSON. Notice that OTHER^ONE^C_PERSON, in (12a), is articulated not only in a high R-locus but also on the lateral plane together with the

10 non-manual markers of non-specificity (see Figures 4 and 5). ONE^OTHER, in (12b), has also been observed as a possible order.

(12) a. ______non-sp OTHER^ONE^C_PERSON SEA WATER BOTTLE THROW FINISH 'Someone (exclusive to our group) threw the water bottle into the sea.’

Figure 4. OTHER ONE C_PERSON

b. ____non-sp ONE^OTHER SHOE 1STEAL3lat-high 'Someone (exclusive to our group) stole my shoe.'

OTHER contributes the exclusive reading unambiguously, since an utterance such as the ones in (12) above cannot be followed by the question “Did you do it?”. We provide the following additional stills since they show much more clearly that OTHER is articulated on the lateral plane (away from the signer's body) and the movement of OTHER ends in a high location.

Figure 5. OTHER

Note, finally, that OTHER in these contexts has a function different from the modifier other in noun phrases such as I like the other dress better. When OTHER is used in this function, the eye gaze is not averted but is directed to the addressee.

5.2.2 ONElat-high In addition to forms with OTHER, ONE can also be used in these exclusive contexts. When it is, it is articulated in the same region as OTHER, namely, in a high R-locus on

11 the ipsilateral side of the signer, hence the subscript “lat-high”, accompanied by the non-manual markers of non-specificity described above.

___non-sp 14 Figure 6. ONElat-high : exclusive, non-specific

It is crucial to note that due to the nature of the impersonal construction and the contexts developed to elicit it, in most of the cases, speakers/signers presuppose that the unknown agent is not among the addressees or in the immediate group of people salient in the discourse. However, there can also be cases where the speaker/signer is again not familiar with the referent of the agent (i.e. non-specific agent) but presupposes that the unknown agent is from among the immediate group of people in the social context. In those cases, ONE would be signed not in a lateral high R-locus but in a central-low R-locus. This would imply that the agent is “one from among us”. We call this the “in-group-partitive” reading, a partitive reading with the referent taken from the group including the addressee (and the signer) and other people salient in the discourse who would be considered to be belonging to the immediate group of people in the social context.15

___non-sp Figure 7. ONEcentral-low: in-group-partitive, non-specific

12

Recall that in the previous section, we argued that the exclusive reading of forms with OTHER and ONElat-high is supported by the fact that a follow-up question such as “Did you do it?” is not possible. Such a follow-up is, however, possible with ONEcentral-low forms. What is odd with these forms is contexts where the agent is clearly exclusive, shown in (13) below:16

(13) a. Context (exclusive): You went to the shopping mall. When you searched for your phone, you realized that your phone was gone. You went home and told your partner about it.

Would you say the following in this context?

______non-sp

# POSS-IX1 PHONE ONEcentr-low 1STEAL3central-low. 'Someone (from our group) has stolen my phone.’

A final remark is on the specific vs. non-specific distinction: In 5.1. we stated that we analyze the specificity interpretation of the indefinites in this paper with respect to the identifiability of the referent by the signer, and we argued that the indefinite forms discussed in this section are interpreted as non-specific (i.e. their referents cannot be identified by the signer) when accompanied by the bundle of non-manual markers that express non-specificity. As described in 5.1., Barberà (2015) has shown that in LSC, high R-locus is associated with non-specificity, whereas low R-locus is associated with specificity. Both of these regions are in the lateral plane, so the high vs. low contrast in the lateral plane is associated with the contrast between non-specific vs. specific interpretations, respectively. Our examples of in-group partitive indefinites so far have been non-specific, since they were elicited in impersonal contexts. However, they can be interpreted as specific when the non-manual markers of non-specificity are absent. Note that the locus of an in-group partitive indefinite is not only central but also low, hence, the subscript central-low in the examples above (see Figure 7). However, this locus does not have a higher counterpart. In other words, there is no high vs. low contrast in the central space parallel to the high vs. low contrast reported for the lateral plane for LSC. This shows that in TİD, there is only one region available for in-group partitives in the central signing space, the lower one, and that locus is underspecified for specificity. The contrast in specificity is expressed with the presence vs. absence of the non- manual markers.17

We return to these issues in Section 6.

5.3 Expressing impersonal agents through agreement

Agents and themes may be marked on the verb in sign languages (Padden 1988; Mathur & Rathmann 2012). Agreement seems to play a role in expressing

13 impersonality of the agent (see the introduction to this volume). For instance, Barberà & Quer (2013:247) argue that in LSC a specific agreement configuration is used when the structure is impersonal. L'Huillier et al. (2015:26), on the other hand, argue that in impersonal constructions in French Sign Language (LSF) agreement may lack spatial anchoring. In TİD, as well, verbs may exhibit modifications from canonical agreement patterns in impersonal constructions. We elaborate on these in the following sections.

5.3.1 Agreement in lateral high R-locus Lateral high R-locus associated with exclusive impersonality in TİD (see Section 5.2) plays a role in agreement as well. (14a) illustrates this. The agent morpheme on the (forward) double agreement verb SEND is articulated in a high R-locus, and the agent is interpreted as exclusive to the people in the social context. Note also the non-manual markers for non-specificity which were described in 5.1.

(14) a. ae lmc bf 3lat-highSEND.LETTER1 ‘Someone (no-specific, exclusive to our group) sent a letter to me.’

Figure 8. 3lat-highSEND.LETTER1

Contrast this with Figure 9 and the example in (14b) where the signer presupposes that the impersonal agent is someone inclusive to her group and the verb SEND is signed in a central low R-locus.

b. ae lmc lmc bf 3central-lowSEND.LETTER1 ‘Someone (non-specific, inclusive to our group) sent a letter to me.’

14

Figure 9. 3central-lowSEND.LETTER1

(15) provides another example. STEAL is a backward agreeing verb and when the signer presupposes that the non-specific agent is someone outside the immediate group of people, the agreement morpheme for the agent is in a lateral high R-locus (Figure 10).18

(15) ae

lmc bf POSS1 SHOE ONElat-high 1STEAL3lat-high 'Someone (exclusive to our group) stole my shoe.'

Figure 10. ONElat-high 1STEAL3lat-high

Contrast this with (16) where the agreement morpheme is in a central low R-locus (Figure 11). In this example, the impersonal agent is interpreted to be included in the signer’s group.

(16) ae lmc bf POSS1 SHOE ONE central-low 1STEAL3central-low 'Someone (non-specific, inclusive to our group) stole my shoe.’

15

Figure 11. 1STEAL3central-low

Consistent with our analysis of the central-low locus as encoding a referent from the in-group (see 3.2.), a pronominal form with OTHER, which lexically encodes exclusivity, is unacceptable in a central locus. Moreover, when OTHER expresses the agent in a sentence with an agreeing verb, the agent morpheme on the verb must be articulated in a lateral-high R-locus, as in (17a); signing it in central-low R-locus is unacceptable, as shown in (17b).

(17) a. OTHER POSS1 SHOE 1STEAL3lat-high ‘Someone (exclusive) stole my shoe.’

b. *OTHER POSS1 SHOE 1STEAL3central-low Intended: ‘Someone (exclusive) stole my shoe.’

5.3.2 Single argument agreement verbs signed in high R-locus In addition to the prototypical agreeing verbs discussed in the previous section, a number of verbs that are commonly categorized as plain verbs, such as PLAY and BREAK, also show agreement in lateral-high R-locus in TİD, as has also been observed for LSC in Barberà & Hofherr (2017). Costello (2015a) considers only body-anchored verbs as genuine plain verbs and calls verbs that are not body-anchored, but are articulated at the locus associated with an argument, as “single argument agreement verbs” (Costello 2015a:127). In our examples, these verbs show agreement with the impersonal subject/external argument. Consider Figure 12 and Figure 13. In both, the verb is signed in the same location as the pronominal subject, ONElat-high. The fact that the signer signs the subject ONElat-high in different sides in these two utterances clearly shows that the verb is not only signed in any lateral high location, but also agrees with the R-locus of the subject.

16

Figure 12. ONElat-high PLAYlat-high

Contrast the location of the verb BREAK in Figure 13a with that of BREAK in Figure 13b. Figure 13b has been taken from an utterance where the agent is definite. In this example BREAK does not agree with any of its arguments and is signed in the neutral space. 19,20

Figure 13. a. ONElat-high BREAKlat-high b. BREAKneutral

To summarize, similar to the findings in Barberà & Cabredo Hofherr (2017) for LSC, in TİD as well, verbs such as PLAY and BREAK that do not regularly display canonical agreement patterns can agree with a lateral high R-locus when the agent is impersonal.

6 Discussion of the findings and conclusion

In this study we have shown that TİD employs various strategies in agent- backgrounding that are also similar to those observed in other sign languages, discussed in the introduction to this volume. Among these are the use of null subjects, agent reduction, 3rd person plural marking of the verb, mouthing the Turkish verb with 3rd person plural inflection, generic uses of 1st and 2nd person pronouns, non-specific indefinite pronominals, and agreement in high R-locus. We focused mainly on non- specific indefinite pronominals. The table below summarizes our findings regarding the different forms of indefinite pronominals.

17

Table 1: Summary of the properties of the indefinite pronominals

lexical Clusivity R-locus non-manual marking for markers (NMMs) specificity

O NE^PERSON^C_PERSON neutral neutral none may or may not (and other related forms) have NMMs of non-specificity

O NEcentral-low neutral inclusive central low may or may not (in-group have NMMs of partitive) non-specificity

O THER^O NE^C_PERSON (and non-specific exclusive lateral high NMMs of other related forms) non-specificity

O NElat-high non-specific exclusive lateral high NMMs of non-specificity

In the previous section we discussed the role the R-loci play in both indefinite pronominals and agreement. We elaborate on the implications of this on the pragmatic use of space below. The first implication concerns the correlation between the height of R-loci and their interpretation in terms of specificity. Barberà (2015) and her related previous and subsequent work argue that high R-locus is associated with non-specificity, whereas low R-locus is associated with specificity in LSC. Note that this proposed correlation between the contrast in the height of loci and contrast in specificity is meant to apply in the lateral signing space, and no claim is made regarding a similar contrast in the central signing space. Recall from 5.2.2. and 5.3.1 that TİD signers use lower central space to express in-group partitives, and that we showed that these partitives can be specific or non-specific, namely, they can be used in contexts where the referents are identifiable by the signer as well as in those where it is not. They are disambiguated by the presence vs. absence of the non-manual markers of non-specificity. These findings show at least for TİD that lower component of the signing space does not encode specificity across both lateral and central parts but only in the lateral part. It can be analyzed either as underspecified for specificity or as restricted to encode such distinction only in the lateral signing space. This may not be so surprising if higher region in the central signing space is not employed in establishing (abstract/syntactic) R-loci. As far as we know, in TİD this region may only be used for topographic or honorific reference. Therefore, when the central space is used for in-group partitives, and the R-loci are established in the lower area, this area is underspecified for specificity. The second implication of our findings is that the lateral signing space encodes exclusion of the addressee. It has been reported that many sign languages make a clusivity distinction in 1st person plural pronouns such as WE and TWO_OF_US and the distinction is related to whether or not the addressee is included. While inclusive pronouns are signed at the center of the signer's chest, exclusive pronouns are signed on the left or the right side as in ASL and British Sign Language (BSL) (Cormier,

18

2012:233). These pronouns may exclude not only the addressee but also any referent salient in the discourse. Our findings on exclusive indefinites parallel these reports on exclusive pronouns. Exclusive indefinites (i.e. “someone other than you and people in our group”) are also signed in the lateral signing space whereas inclusive indefinites (i.e. in-group partitive, “someone that could be you or one of the people in our group”) are signed in the central space.21 Thus, we conclude that at least for TİD, “(lateral) high R-locus” encodes not only one but two pragmatic features: non-specificity and exclusivity. As we discuss in detail in Kelepir et al. (in press) a general look at the R-loci of referential pronouns shows that this distribution of spatial components is not limited to indefinites and 1st person plural pronouns but can be extended to all pronouns: whereas 2nd person pronouns (i.e. pronouns whose referent includes or is identical to the addressee) are signed in the central signing space, 3rd person pronouns are signed in the lateral signing space. Based on these facts, we hypothesize the following correlation of spatial features with pragmatic features for TİD referential loci (of pronominal signs and agreement morphemes): [+high] is unambiguously non- specific/non-referential, [-high] is underspecified for specificity/referentiality. [+lateral] is unambiguously exclusive and [-lateral] unambiguously inclusive.22 The table below represents potential combinations of the features and corresponding pronominal forms that have been attested. For clarity of presentation and to include the personal pronouns, we use the term “referential” for both “specificity of indefinites” and “referential use of pronouns”. The first line in the combination cell is for indefinites, the second line for personal pronouns. Note that to present a potential example of a non-referential personal pronoun signed in a lateral high R-locus, we have added to this table the impersonal 3rd person plural pronoun reported for LSC (Barberà 2016) but not attested in TİD. Other examples are from TİD.

Table 2: Combinations of features and attested pronominal forms [+lateral] [-lateral] exclusive inclusive non-specific, exclusive indefinite, (indefinite not observed)

[+ high] e.g. ONElat-high, OTHER non- referential impersonal 3rd pl pronoun THEY (pronominal not observed) (LSC)

(indefinite not observed yet) non-specific/specific inclusive indefinites,

[-high] e.g. ONEcentral-low rd underspecified 3 person referential pronouns for 2nd person referential pronouns referentiality

19

Consider the combination of the features [+high] and [-lateral] in the table above. No (abstract/syntactic) R-loci have been attested in this region, other than perhaps signs pointing upwards for topographic reference and honorific use (Perniss 2013). Hence, no indefinite or personal pronoun is articulated there. Consider also the combination of [-high] and [+lateral]. In our analysis, an indefinite articulated in an R-locus with these features would be exclusive but underspecified for specificity. Since there are special forms in TİD for non-specific exclusive indefinites, there could potentially be specific exclusive indefinite forms as well. This could be expressed as ‘someone (I know), excluding you and our group’. We have not observed this in our study but such a special indefinite form can potentially exist. These observations, if correct, contribute to the discussion of the correspondence between pragmatic features and components of signing space (Barberà 2015 and related work), the significance of contrast for the division of signing space (Steinbach & Onea 2016), and the theories of iconic mapping between loci and the objects they denote (Schlenker & Lamberton 2012). Finally, clusivity in languages is usually associated with personal pronouns (Filimonova 2005). Our study shows that this concept must be extended to indefinite pronouns, at least in TİD.

References

Barberà, Gemma. 2012. A unified account of specificity in Catalan Sign Language (LSC). In Rick Nouwen, Anna Chernilovskaya & Ana Aguilar-Guevara (eds.), Proceedings of Sinn und Bedeutung 16. Volume 1, 43-55. MIT working papers in linguistics. Barberà, Gemma. 2015. The Meaning of Space in Sign Language. Reference, Specificity and Structure in Catalan Sign Language Discourse. Berlin/Boston: De Gruyter Mouton and Ishara Press. Barberà, Gemma. 2016. Indefiniteness and specificity marking in Catalan Sign Language (LSC). Sign Language & Linguistics 19(1). 1-36. Barberà, Gemma & Josep Quer. 2013. Impersonal reference in Catalan Sign Language (LSC). In Laurence Meurant, Aurélie Sinte, Mieke van Herreweghe & Myriam Vermeerbergen (eds.), Sign Language Research Uses and Practices: Crossing Views on Theoretical and Applied Sign Language Linguistics, 237-258. Berlin/Boston: De Gruyter Mouton and Ishara Press. Barberà, Gemma & Patricia Cabredo Hofherr. 2017. Backgrounded agents in sign language: passives, middles or impersonals? Language 93(4). Barberà, Gemma & Patricia Cabredo Hofherr. This volume. R-Impersonals in sign languages: An introduction. Barberà, Gemma & Kearsy Cormier. 2017. Reference. In Josep Quer, Carlo Cecchetto, Roland Pfau, Caterina Donati, Markus Steinbach, Carlo Geraci & Meltem Kelepir (eds.), SignGram Blueprint: A Guide to Sign Language Grammar Writing. Berlin: De Gruyter Mouton.

20

Bahan, Benjamin, Judy Kegl, Robert Lee, Dawn MacLaughlin, & Carol Neidle. 2000. The licensing of null arguments in ASL. Linguistic Inquiry 31(1). 1–27. Cormier, Kearsy. 2012. Pronouns. In Roland Pfau, Markus Steinbach & Bencie Woll (eds.) Sign Language: An International Handbook, 227-244. Berlin: Mouton de Gruyter. Costello, Brendan. 2015a. Language and modality: Effects of the use of space in the agreement system of lengua de signos española (Spanish Sign Language). Amsterdam: Universiteit van Amsterdam dissertation. Costello, Brendan. 2015b. Getting (more) impersonal in LSE. Talk at Workshop on Impersonals and Passive in Sign Languages, U. Pompeu Fabra, Barcelona. (15- 16 June, 2015). Dikyuva, Hasan, Bahtiyar Makaroğlu & Engin Arık. 2015. Türk İşaret Dili Dilbilgisi Kitabı [Turkish Sign Language Grammar Book]. Ankara: TC Aile ve Sosyal Politikalar Bakanlığı. Enç, Mürvet. 1991. The of specificity. Linguistic Inquiry 22. 1-25. Filimonova, Elena (ed.) 2005. Clusivity: Typology and case studies of the inclusive- exclusive distinction. Amsterdam: Benjamins. Fodor, Janet Dean & Ivan Sag. 1982. Referential and quantificational indefinites. Linguistics and Philosophy 5. 355-398. Heusinger, Klaus von. 2002. Specificity and in sentence and discourse structure. Journal of Semantics 19. 245-274. Isenhath, John O. 1990. The Linguistics of American Sign Language. Jefferson, NC: MacFarland and Company. Janzen, Terry, Barbara O’dea & Barbara Shaffer. 2001. The construal of events: Passives in American Sign Language. Sign Language Studies 1(3) 281-310. Karabüklü, Serpil. 2016. Time and aspect in Turkish Sign Language (TİD): Manual and nonmanual realization of 'finish'. Istanbul: Boğaziçi University MA Thesis. Kegl, Judy. 1990. argument structure and verb-class organization in the ASL Lexicon. In Ceil Lucas (ed.), Sign language research: theoretical issues, 149- 175. Washington, DC: Gallaudet University Press. Kelepir, Meltem. 2001. Topics in Turkish Syntax: Clausal Structure and Scope. Ph.D. dissertation, Massachusetts Institute of Technology. Distributed by MIT Working Papers in Linguistics, Cambridge, MA. Kelepir, Meltem & Aslı Özkul. 2015. Passive-like constructions with inanimate themes in Turkish Sign Language. Talk at Workshop on Impersonals and Passive in Sign Languages, University of Pompeu Fabra, Barcelona. (15-16 June, 2015). Kelepir, Meltem, Aslı Özkul & Elvan Tamyürek Özparlak. In press. Clusivity distinction in indefinites in Turkish Sign Language (TİD). Proceedings of the 53rd Annual Meeting of the Chicago Linguistic Society, May 25-27, 2017, University of Chicago. Kelepir, Meltem & Aslı Özkul. In preparation. Passive-like constructions in Turkish Sign Language. Kemaloğlu, Yusuf & Pınar Yaprak Kemaloğlu. 2012. The history of sign language and deaf education in Turkey. Kulak Burun Boğaz İhtisas Dergisi 22(2). 65-76.

21

Koulidobrova, Elena. 2017. Elide me Bare. Natural Language & Linguistic Theory 35(2). 397-446. Kubuş, Okan. 2008. An analysis of Turkish Sign Language (TID) phonology and morphology. Ankara: Middle East Technical University MA Thesis. L’Huillier, Marie-Thérèse, Marie-Anne Sallandre & Brigitte Garcia. 2015. Impersonal reference to humans in LSF: a first glance. Talk presented at the workshop on Sign languages and R-impersonal pronouns, Centre national de la recherche scientifique (CNRS), Paris. (6 February, 2015). Mantovan, Lara. 2017. Determiners. In J. Quer, C. Cecchetto, R. Pfau, C. Donati, M. Steinbach, C. Geraci & M. Kelepir (eds.) SignGram Blueprint: A Guide to Sign Language Grammar Writing. Berlin: De Gruyter Mouton. Mathur, Gaurav & Christian Rathmann. 2012. Verb agreement. In Roland Pfau, Markus Steinbach & Bencie Woll (eds.), Sign language – An international handbook, 136-157. Berlin: De Gruyter Mouton. Miles, Mike. 2000. Signing in the Seraglio: Mutes, dwarfs and gestures at the Ottoman Court 1500-1700. Disability & Society 15(1). 115-134. Özkul, Aslı. In preparation. Adverbial Clauses in TİD. İstanbul: Boğaziçi University dissertation. Özsoy, Sumru & Chiara Branchini. In press. Clause Structure. In Josep Quer, Carlo Cecchetto, Roland Pfau, Caterina Donati, Markus Steinbach, Carlo Geraci & Meltem Kelepir (eds.), SignGram Blueprint: A Guide to Sign Language Grammar Writing. Berlin: De Gruyter Mouton. Padden, Carol A. 1988 [1983]. Interaction of morphology and syntax in American Sign Language (Outstanding Dissertations in Linguistics, series IV). New York: Garland Press. Perniss, Pamela. 2013. Use of sign space. In Roland Pfau, Markus Steinbach and Bencie Woll (eds.), Sign language. An international handbook. Berlin, Mouton De Gruyter. 412-431. Saeed, John I. & Lorraine Leeson. 1999. Detransitivisation in Irish Sign Language. Paper presented at the European Science Foundation Intersign Meeting on Morphosyntax (corpora and tagging). Sienna, Italy, March 1999. Published at the Intersign website: http://www.sign-lang.uni- hamburg.de/Intersign/Workshop3/ Schlenker, Philippe & Jonathan Lamberton. 2012. Formal Indices and Iconicity in ASL. Logic, Language and Meaning: 18th Amsterdam Colloquium, Springer 7218. 1- 11. Siewierska, Anna. 2011. Overlap and complementarity in reference impersonals: ‘Man'-constructions vs. third person plural-impersonals in the languages of Europe. In Anna Siewierska & Andrej Malchukov (eds.) Impersonal Constructions: A Cross-linguistic Perspective, 57-89. Amsterdam: John Benjamins. Steinbach, Markus & Edgar Onea. 2016. A DRT-Analysis of Discourse Referents and Anaphora Resolution in Sign Language. Journal of Semantics 33. 409-448.

22

Stokoe, William C., Dorothy Casterline & Carl Croneberg. 1965. A Dictionary of American Sign Language on Linguistic Principles. Washington DC: Gallaudet College Press. Sze, Felix. 2010. Is there passive in Hong Kong Sign Language? Poster presentation at TISRL 10. Purdue University. (30 September - 2 October 2010). Taşçı, Süleyman & Aslı Göksel. 2014. The morphological categorization of polymorphemic lexemes: A study based on lexicalized fingerspelled forms in TİD. Dilbilim Araştırmaları Dergisi, Special Issue in Honor of Prof. A. Sumru Özsoy, 165-180. Wilbur, Ronnie. 1987. American Sign Language: Linguistic and Applied Dimensions. Boston: College-Hill Press. Zeshan, Ulrike. 2002. Sign Language in Turkey: The story of a hidden language. Turkic Languages 6(2). 229-274. Zeshan, Ulrike. 2003. Aspects of Türk İşaret Dili (Turkish Sign Language). Sign Language and Linguistics 6(1). 43-75. Zucchi, Sandro, Carol Neidle, Carlo Geraci, Quinn Duffy & Carlo Cecchetto. 2010. Functional markers in sign languages. In Diane Brentari (ed.), Sign Languages, 197-224. Cambridge: Cambridge University Press.

Author Contact Information

Meltem Kelepir [email protected] Boğaziçi University Department of Linguistics Istanbul, Turkey

Aslı Özkul (corresponding author) [email protected] İstanbul Bilgi University English Language Teaching Department Istanbul, Turkey

Elvan Tamyürek Özparlak [email protected] Boğaziçi University Department of Linguistics Istanbul, Turkey

Acknowledgements We would like to thank those who contributed to the development of this work with their comments during the FEAST Conference (Barcelona, UPF, 4-5 May 2015) and the Workshop on Impersonals and Passive in Sign Languages (Barcelona, UPF, 15-16 June 2015). We also thank our informants Zehra Gül Yiğit, Yusuf Ermez, Yasin Yiğit,

23 Nuriye Köroğlu, Jale Erdul, Gamze Uğuz, Neveda Öner, Suna Keskin, Filiz Birinci Sezer and Cem Barutçu. We thank the two anonymous reviewers and the editors of this volume for their invaluable feedback which helped us improve this paper. The errors are, of course, all ours. Finally, we would like to note that this study has been possible thanks to the Sign-Hub project which has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement no 693349 and thanks to the project “A Signing Avatar System for Turkish to Turkish Sign Language Machine Translation” which has received funding from the Scientific and Technological Research Council of Turkey (TUBITAK) with the grant number #114E263.

Notes

1 Even though to our knowledge no systematic work has been done on Turkish agent-backgrounding, we can report that in addition to the use of passive constructions, plural marking of the verb with a null subject pronoun with impersonal interpretation and use of a singular indefinite pronoun birisi ‘someone’ are common strategies. The use of 2nd person pronouns in generic contexts such as “When you’re pregnant, you shouldn’t lift heavy objects.” is also possible. 2 As observed for other sign languages such as LSC (Barberà & Cabredo Hofherr (2017)) and LSE (Costello 2015b), in locative universals, the locative phrase such as FRANCE is usually signed utterance-initially in TİD. 3 Passivization is a controversial issue in sign language linguistics (Stokoe et al. 1965; Wilbu r 1987; Kegl 1990; Isenhath 1990; Saeed & Leeson 1999; Janzen et al. 2001), so we have refrained from calling these constructions passive, but for ease of exposition, we have called them "passive-like". 4 A special perfective marker in negative contexts was previously observed for LIS (Italian Sign Language) (Zucchi et al. 2010). 5 Barberà & Cabredo Hofherr (2017) consider verbs such as break as plain verbs based on the fact that they are not inflected for agreement when the agent is not impersonal, i.e. they are signed in the central signing space and do not involve change in direction of path movement or orientation. They observe, however, that these verbs do get inflected for agreement when their agent is impersonal. The verbs in those cases agree with a high R-locus. This contrasts with the categorization of verbs proposed in Costello (2015a) which considers only those verbs which are body-anchored as plain verbs. Costello (2015a) argues that if a verb is inflected for agreement at all, it should not be considered a plain verb, and considers cases where an otherwise uninflected verb displays agreement with a high R-locus a case of a single argument agreement (see 5.3.2). 6 These non-manual markers are obligatory for non-specific reading. However, note also that it is possible that they, especially lowered mouth corners, may have a modal function, i.e. expressing uncertainty, rather than non-specificity. The clarification of this issue requires further research. 7 A circumflex accent is used to mark compounds in this article. 8 The abbreviations of non-manual markers are: ‘non-sp’: bundle of non-manual markers expressing non-specificity, ‘ae’: averted eye gaze, ‘lmc’: lowered mouth corners, ‘bf’: brow furrowing, ‘br’: brow raising, ‘bht’: single backward head tilt, and ‘pc’: puffed cheeks. 9 Dikyuva et al. (2015) gloss the sign with narrowed-O-handshape as ADAM 'man' and the one with the C-handshape as KIŞI 'person'. Since in our data this form is gender-neutral and our consultant translates it as KIŞI 'person', we gloss it as PERSON. Moreover, Dikyuva et al. (2015) gloss the sign with C- handshape as KIŞI 'person' but we gloss it as C_PERSON to differentiate it from the other sign. 10 It is worth noting that the starting point in the downward movement of C_PERSON is lower when it is used in forms such as WORK^C_PERSON 'worker' but it seems to be higher when it is used in the non- specific pronoun ONE^PERSON^C_PERSON 'someone'. However, it is not clear at the moment whether this difference is due to phonological assimilation to the previous sign or to pragmatic significance.

24 11 Dikyuva et al. (2015) analyze WORK^C_PERSON 'worker' as a compound and C_PERSON as a lexical root meaning 'person', but also add that the latter might be in the process of grammaticalization into an agentive suffix. 12 An anonymous reviewer asks whether OLD ONE^PERSON^C_PERSON is possible. Both this and ONE^PERSON^C_PERSON OLD are possible. 13 The following combinations are also possible: PERSON^C_PERSON, ONE^PERSON and ONE^C_PERSON, the last one used very infrequently. 14 We gloss this pronominal as ONElat-high and not as ONEup as in Barberà (2016), since articulation of these signs in the lateral vs. central space is crucial for our analysis. 15 We thank the editors for the suggestion of this term. 16 For a more detailed discussion of clusivity distinction in indefinites in TİD, see Kelepir et al. (in press). 17 An anonymous reviewer raises a number of questions regarding the indefinite forms described in this section. The first question is whether there is any meaning difference between different combinations of ONE and OTHER and also between different combinations of ONE, PERSON and C_PERSON. We have not found any difference. The second question is whether signing of these verbs in the lateral signing space could be due to contrast in the context. The contexts the informants received did not involve contrast. The context for Figure 12, for instance, says “You enter a room and there is nobody there but you see that the materials for the game OKEY are spread on the table. What would you say?” 18 The shared non-manual markers throughout the examples (14-16) are averted eye gaze, lowered mouth corners and brow furrowing. Averted eye gaze mainly accompanies the agent agreement morpheme of the agreement verb or the indefinite pronoun. Lowered mouth corners may occur with the indefinite pronoun or it can spread over the verb phrase. The spreading domain of brow furrowing is usually larger than the former two non-manual markers. It usually spreads over the verb phrase or the entire clause. 19 An anonymous reviewer raises the question whether all these different indefinite forms can be used interchangeably, and more specifically, whether any of these forms would be unacceptable in unambiguously specific contexts, such as “I have to meet with some professor. It's that guy over there.” The identification of subtle differences between these indefinite forms and ways of expressing specificity requires its own research project, so we leave this to future studies. 20 Here we identify the verb’s signing space as “neutral” as opposed to “central” to differentiate between the marked use of the central space for discourse referents, namely, for inclusion of the addressee from the use of the same area as neutral signing space in the absence of any referential loci. 21 Note that signers may vary in their use and interpretation of exclusivity. We hypothesize that when an exclusive indefinite pronominal is used, its restriction minimally excludes the addressee, but it can also exclude a set of people salient in the discourse (‘people in our group’). Thus, how restricted this restriction set is may depend on the signer and the context, a fact which should be taken into consideration when designing elicitation tasks for this topic. 22 This implies that all R-loci in the central space have to include the addressee as a potential referent and all R-loci referring to non-2nd person referents have to be considered to be in the lateral space. If there are cases in which a non-2nd person referent is located in the central space, then this generalization has to be modified as “[-lateral] is underspecified for clusivity”. In that case, “central” and “low” would be neutral components of signing space, whereas “lateral” and “high” would be marked.

25