1 Introduction

This paper discusses agent defocusing in two-participant clauses in Finnish Sign Language (FinSL). The aim is to find out what kind of strategies are used in FinSL when no specific, referential agent is mentioned, or the identity of the agent is blurred so that it is not (fully) referential. Well known agent defocusing constructions cross-linguistically are passive constructions (e.g. The house was built last year) (Shibatani 1985; Siewierska 2013) and different R-impersonals (reference impersonals), in which the agent is expressed with a form (e.g. a personal pronoun) that is non-referential (e.g. They eat late in Spain) (Siewierska 2011).

Agent defocusing constructions are widely known and studied in spoken languages, but there has been only little research on agent defocusing in sign languages. Interest in the topic has begun to increase in recent years, and several constructions defocusing the agent have been found in different sign languages. Constructions that defocus the agent by not overtly expressing it have been found at least in American Sign Language (ASL) (Kegl 1990; Janzen, O’Dea & Shaffer 2001; Rankin 2013), Irish Sign language (Leeson 2001), French Sign Language (Guitteny 2006) and Catalan Sign Language (Barberà & Cabredo Hofherr 2017). These include constructions where the event is told from a neutral perspective, the agent’s perspective or a patient’s perspective (see especially Rankin 2013). Also, constructions that contain an overt A argument1 used non-referentially, which can be analyzed as R-impersonals, have been found e.g. in Catalan Sign Language (Barberà & Quer 2013) and in American Sign language (Rankin 2013). In addition, work on argument focusing and defocusing using different handshape types has been carried out in several sign languages (e.g. Benedicto & Brentari 2004; Börstell 2017).2

This paper aims to describe different strategies that are used for defocusing the agent in FinSL by examining extracts from informational text. Theoretically, this work is situated in a cognitive-functional framework. It takes as a starting point the function of agent defocusing and aims to describe how this function is fulfilled in the target language. The theoretical framework of the present study is the Basic Linguistic Theory (Dixon 2010), which has been commonly used for the description of FinSL (e.g. Jantunen 2013). This paper accepts the view according to which some meaningful units of the language are not morphological, i.e., conventional and categorical, but rather gestural, i.e., unconventional and gradient (see e.g. Liddell 2003; Jantunen 2017a). For example, the spatial modification of verbs,3 which is used for referent tracking and indicating semantic roles in the clause, is seen as a form of gestural pointing instead of morphological agreement (see de Beuzeville, Johnston & Schembri 2009; Jantunen 2010; Fenlon, Schembri & Cormier 2018; Schembri, Cormier & Fenlon 2018). The issue is discussed in Section 2.2.1.

The study is based on a detailed analysis of video data which consists of 11.5 minutes of videotaped informational texts in FinSL (the main data). In addition, some supporting evidence is sought in a larger data set. The main data has been annotated carefully with the annotation tool ELAN (Crasborn & Sloetjes 2008)4 on several layers including signs and non-manual features, classified in Excel, and analyzed in detail. Native signers have been consulted during the analysis.

The agent is understood as the volitional instigator of an action in a two-participant clause. The agent is defined broadly, covering the more agent-like participant of the clause, including the experiencer arguments of verbs expressing cognition or perception such as see or look (Dixon 1979; Næss 2007; see also Dowty 1991). Similarly, the term patient is used to refer to the more patient-like participant of the clause, covering recipient and theme. A two-participant clause is defined according to the semantics of the verbal predicate, as a clause containing semantically at least two separate participants: an agent and a patient (Næss 2007). The participants do not need to be overtly expressed. As the study focuses on two-participant clauses, intransitive clauses such as Peter runs, in which the only argument is the agent, are excluded from the study. One reason for restricting the study to two-participant clauses is that one aim was to find out whether there is a passive construction in FinSL; if a language does have a passive construction, it can be formed from transitive clauses, at least.

I start by defining the concept of agent defocusing and describing agent defocusing strategies in spoken languages (2.1). This is followed by an overview of agent defocusing strategies in sign languages (2.2), including constructions relevant for agent defocusing in FinSL and sign language specific phenomena relating to referentiality and agent defocusing (2.2.1). I then outline the data and the methodology used in the present study (3), before presenting the results (4) and discussing them (5). Finally, I draw some conclusions (6).

2 Agent defocusing

2.1 Agent defocusing and its manifestation in spoken languages

The phenomenon labeled here as agent defocusing, adopted from Shibatani (1985), has been treated by different authors using different terms, such as agent demotion (Siewierska 2013), subject demotion (Comrie 1977), agent backgrounding (Barberà & Cabredo Hofherr 2017), impersonal construction (e.g. Siewierska 2008; 2011; Siewierska & Papastathi 2011, Barberà & Quer 2013), open reference (Helasvuo 2008), or functional passive (Leeson 2001), all of which have slightly different but overlapping definitions (see Shibatani 1985; Siewierska 2008). Shibatani (1985) has defined agent defocusing as including phenomena such as not mentioning the agent, mentioning the agent in a non-prominent syntactic slot, using plural forms to blur the identity of the agent, and referring to the agent indirectly using an oblique case. He argues that agent defocusing is the primary function of a passive construction. Some other researchers, on the other hand, have argued that the function of promoting the patient is more important. The reasons for defocusing the agent, according to Shibatani (1985: 831), are at least that the agent is irrelevant, unknown or obvious.

This study concentrates on semantic agent defocusing. In her article on impersonal constructions, Siewierska (2008) has emphasized that a distinction should be drawn between semantic defocusing of the agent and structural subject demotion. By defocusing, she means “diminishing the prominence or salience from what is assumed to be the norm” (Siewierska 2008: 121). According to Siewierska (2008: 121), the defocusing may involve “a) non-elaboration or under-elaboration of the instigator, b) the demotion of the instigator from its prototypical subject and topic function, or c) both demotion and under-elaboration.” Shibatani (1985: 832) notes also that in accusative-type languages, in active clauses an agent “is preferably chosen as a locus of viewpoint”. In the passive, on the other hand, focusing on an agent is avoided. Applying these definitions to FinSL, agent-defocusing constructions are compared to constructions that are typically used for expressing a fully referential agent, even though the specific terms subject and topic are not used.

Agent defocusing is defined as a strategy in which the identity of the agent is blurred by not overtly expressing it or by expressing it with a form that is not (fully) referential. The term non-referential means that there is no specific referent which can be understood to be the agent. Referentiality is, however, not a binary characteristic. Non-referential is in this paper understood to cover instances where the agent is not fully referential (see Siewierska 2008: 116). These include cases which, in order to blur the identity of an agent, use generic nouns or pronouns such as on in French or they in English, for example (Shibatani 1985; Siewierska 2008: 116–122).

Agent defocusing can be manifested in several ways, which include at least passive constructions (e.g. Shibatani 1985; Siewierska 2013), R-impersonals (Siewierska 2011) and nominalizations (e.g. the circling of the camp) (Siewierska 2008). Example (1) presents a prototypical passive clause in English.

(1) English
  The house was built last year.

Pragmatically, the patient the house is in focus, the clause serves to specify when the house was built, and mentioning the agent is unimportant or avoided. It is a prototypical passive clause (Shibatani 1985): the predicate contains semantically an agent and a patient, but the valence of the verb is reduced so that the clause is structurally intransitive. The patient is expressed as the subject, the verb has a morphological passive marking, and the agent is not overtly expressed. In English the agent could be mentioned in a by-phrase, but many languages do not allow expression of the agent in a passive clause (Shibatani 1985: 830–831).

In some languages the agent can be simply omitted without the use of a passive construction (Shibatani 1985: 836; Laitinen 2006; Siewierska 2008; 2013). In (2), from Finnish, the clause is structurally identical with a clause that has a referential third person singular subject, but the subject is omitted.

    1. (2)
    1. Finnish (Laitinen 2006: 212)
    1. Huomenna
    2. tomorrow
    1. saa
    2. get.3SG
    1. kahvi-a.
    2. coffee-PTV
    1. ‘One can get coffee tomorrow.’5

According to Siewierska (2011), reference impersonals or R-impersonals are constructions that look like regular personal constructions but contain a subject that is human and non-referential. The non-referential human subject can be expressed pronominally as in (3), lexically (usually using the word ‘person’ or ‘people’) or by a whole construction, such as a reflexive construction in Romance languages. Some languages have a special impersonal pronoun, such as on in French and man in Swedish (Siewierska 2011).

(3) English
  In France, they eat snails.

Personal pronouns, such as the third person plural as in (3), are commonly used non-referentially, some person forms being more widely used than others (Siewierska 2011; 2004: 210–213). In Finnish, the first person singular form is used non-referentially (Helasvuo 2008), as is shown in (4). While the third person plural form is widely used cross-linguistically, the first person singular does not seem to be common (Siewierska 2004: 210–213).

    1. (4)
    1. Finnish
    1. Loma-a
    2. holiday-PTV
    1. suunnitel-le-ssa
    2. plan-INF-INESS
    1. kannattaa
    2. be.worth.3SG
    1. miettiä
    2. think.INF
    1. esimerkiksi,
    2. for.example
    1. kaipaa-n-ko
    2. miss-1SG-Q
    1. toiminta-a
    2. activity-PTV
    1. vai
    2. or
    1. lepo-a.
    2. rest-PTV
    1. ’When planning a holiday, it’s worth thinking for example whether one needs activities or rest.’

In some languages, the person form may be marked only on the verb, the subject being omitted (Siewierska 2004: 213). This is the case in (4), from Finnish.

Different person forms have a slightly different meaning. The third person plural, for example, excludes the speaker and the addressee (Siewierska 2011: 62; Siewierska & Papastathi 2011: 577), while the second person and first person singular potentially include the speaker and the addressee (Helasvuo 2008; Siewierska & Papastathi 2011).

A third person pronominal subject in a spoken language can be referential or non-referential, depending on whether it has an antecedent. According to Siewierska & Papastathi (2011: 578), the third person R-impersonals as in (3) are formally constructions that have a pronominal subject that lack an overt antecedent in the discourse. Compare Examples (3) and (5). The clause in (3) is about French people in general. The pronoun they has no antecedent and it is thus non-referential. By contrast, in (5) the same clause is used after mentioning a referent, our children, which is the antecedent of the pronoun they, making it anaphoric and referential.

(5) English
  At home our children are very picky but in France they eat snails.

Following the same logic, a null (omission) can be referential (anaphoric) or non-referential, depending on whether it has an antecedent (Huang 2000: 2; Siewierska 2011: 58). If a null has an antecedent, as in (6), it is an anaphoric null. In contrast, if the null has no overtly mentioned antecedent, as in (2), the null is non-referential, i.e., the agent is defocused.

    1. (6)
    1. Finnish
    1. Huomenna
    2. tomorrow
    1. Pekka
    2. Pekka
    1. tulee
    2. come.3SG
    1. ja
    2. and
    1. saa
    2. get.3SG
    1. kahvi-a.
    2. coffee-PTV
    1. ‘Tomorrow Pekka is coming and will get coffee.’

A complication arises with the first and second person pronouns. These pronouns are deictic, i.e., they typically refer to the speaker and addressee, respectively, in a certain situation without having an antecedent (Siewierska 2004: 7). Helasvuo (2008) has argued for Finnish that the addressee has to use the context to interpret whether or not a first person singular pronoun is being used non-referentially.

The agent can be defocused to different extents, depending on prominence and specificity (Shibatani 1985; Langacker 2006). Agentless passive constructions (see Example 1) as well as omission of the agent (Example 2) defocus it to the full extent. If the agent is mentioned e.g. in a by-phrase in English, it is referential, therefore semantically not defocused, but it is syntactically demoted. Further, R-impersonals (see Examples 3 and 4) overtly mention an agent in its normal subject (A argument) position, which makes it structurally prominent. However, the person forms of R-impersonals do not refer to a specific referent, so the agent is semantically defocused.

2.2 Sign languages and agent defocusing

2.2.1 Features relevant for agent defocusing in Finnish Sign Language

A referential agent can be expressed in FinSL with a full noun phrase (NP) (Rissanen 1985: 126–127; Jantunen 2008; 2010), pronominal pointing sign (Rissanen 1998: 105) or a null (Ala-Sippola 2012; Jantunen 2013; 2018). In a prototypical transitive clause, the agent is usually expressed at the beginning of the clause, the most typical sign order being AVP6 (e.g. BOY EAT APPLE7 ‘A boy eats an apple’), but at least the orders APV and PAV are possible as well (Jantunen 2008; 2017b). It is also very common to omit the agent in FinSL, which makes prototypical transitive clauses uncommon in FinSL (Ala-Sippola 2012; Jantunen 2013). In clauses where the agent is omitted, the order of the V and P can be either PV or VP (Ala-Sippola 2012; Jantunen 2013).

A special property of sign languages is their use of space. Pronominal pointing signs that are functional equivalents of pronouns, as well as indicating verbs that can be modified spatially, can be directed deictically to referents that are present in a situation (e.g. Liddell 2003; de Beuzeville et al. 2009; Jantunen 2010; Cormier, Schembri & Woll 2013; Johnston 2013). This means that contrary to the typical use of third person pronouns in spoken languages (Siewierska 2004: 7), in order to be referential, a pronominal non-first person pointing sign does not necessarily need an antecedent (see e.g. Huang 2000) naming the referent with a full NP. Instead, the pointing sign is typically directed towards a physically present referent. However, these pronominal pointing signs are also used anaphorically. In anaphoric use, they point towards locations in the signing space that have been associated with previously mentioned referents (Rissanen 1985: 18–36; 1998: 10; Johnston & Schembri 2007: 271–272; Cormier et al. 2013; Johnston 2013).

Like pointing signs, so too indicating verbs can be used deictically or anaphorically (Jantunen 2010; see also de Beuzeville et al. 2009). An indicating verb in FinSL can be modified in the signing space so that it shows clearly which referent is agent and which is patient, usually with a movement starting from the location of the agent and ending at the location of the patient. An indicating verb can be directed towards referents that are present in the situation (deictic use). Alternatively, it can be directed towards certain locations that have been associated with previously mentioned referents (anaphoric use). When an indicating verb shows the semantic roles by modification, any overt expression of the referent is redundant and it is often omitted (Wulf, Dudis, Bailey & Lucas 2002; Jantunen 2013). Studies on agent defocusing in sign languages have shown that these indicating verbs can be used and modified in the way described, even when they do not refer to a specific referent, and some researchers (Barberà & Quer 2013; Rankin 2013) have also mentioned the non-referential use of pointing signs.

In this paper, indicating verbs as well as pronominal pointing signs in sign languages are seen as blends of morphemic and gestural elements: this follows many linguists working within the cognitive-functional framework (Liddell 2003; de Beuzeville et al. 2009; Jantunen 2010; Cormier et al. 2013; Johnston 2013; Fenlon et al. 2018; Schembri et al. 2018). Language is understood broadly, meaning that gestural elements are not seen as non-linguistic but as an inherent part of the language (see e.g. Jantunen 2017a; Schembri et al. 2018). Traditionally, the spatial modification of verbs has been seen as a morphological agreement and these verbs are widely known as agreement verbs (see Mathur & Rathmann 2012 for an overview). However, this has been questioned, first by Liddell (1995; see also 2003), who proposed that the modification of these verbs should be analyzed as a pointing gesture incorporated into the form of the sign, and his view has been supported by several studies based on corpora (de Beuzeville et al. 2009; Fenlon et al. 2018). Arguments for analyzing the spatial modification of verbs and pointings as gestural are, e.g., that contrary to the rules of grammatical agreement, it is impossible to categorize and list the morphemes, since the pointing and the verb can be directed anywhere in the space; that the modification is motivated iconically, not by grammatical class; and that the modification of the verbs is not obligatory (Liddell 2003; de Beuzeville et al. 2009; Fenlon et al. 2018; Schembri et al. 2018).

In addition to the previously mentioned indicating verbs, FinSL has two other main verb classes. Plain verbs are purely morphematic and cannot indicate semantic roles such as agent and patient in their form (Jantunen 2010). Some of the plain verbs, locatable plain verbs, can be located meaningfully in space (Ala-Sippola 2012).8 Depicting verbs (cf. depicting verbs in e.g. Liddell 2003) include a morphematic handshape that can refer to the agent or the patient and can depict iconically for example the movement or handling of an entity (Jantunen 2010). There were no depicting verbs in the agent-defocusing constructions in the data, and therefore they are not fully described here.

In addition to the hands, sign languages make use of non-manual articulators (including the head, face, eye gaze, arms, and torso), and these have also been connected to agent defocusing. The signer can represent the situation from a certain participant’s perspective, indicate a referent or show more or less the participant’s actions or feelings in the situation being described using many or just one of the non-manual articulators. Constructed action or CA has been defined as a phenomenon in which the signer uses her hands, face and other parts of the body to enact the actions, emotions, thoughts or sayings of a certain participant in a situation, and it has been suggested that it has different levels depending on how many articulators are involved and whether lexical signs are used at the same time (Cormier, Smith & Sevcikova-Sehyr 2015).

Many different terms have been used to describe the use of these non-manual articulators (for a detailed discussion, see Cormier et al. 2015). Eye gaze and orientation of the body or head have also been analyzed as non-manual agreement, similar to the “agreement” of verbs (e.g. Neidle, Kegl, MacLaughlin, Bahan & Lee 2000). In recent years, the term constructed action (CA) has been increasingly used among functionally oriented linguists, and CA is seen as a gestural phenomenon, a form of enacting, not a morphological one (for FinSL see Jantunen 2017a). It has been argued that CA is an optional discourse strategy in FinSL: some signers use it more than others (Jantunen 2017a; see also Ferrara & Johnston 2014). Also, Jantunen (2017a) has shown that CA is used more frequently with depicting verbs than with plain and indicating verbs.

Cormier et al. (2015: 20–22) suggest dividing CA into three prototypes: overt, reduced and subtle. With overt CA, the signer fully assumes the character role, and all the articulators of the signer represent the actions and emotions of the character. Elements of explicit narration, such as lexical signs or entity classifiers, are not used at all. With reduced CA, the signer also assumes the character role using many CA articulators, such as face, head, torso and eye gaze, but some narration is also involved. With subtle CA, the signer makes active use of only one or a few CA articulators, while primarily narrating with lexical signs. This study is based on the assumption that the body lean, head orientation and eye gaze used with lexical signs are elements of CA.

2.2.2 Previous research on agent defocusing in sign languages

One construction defocusing the agent in sign languages has been most frequently described and seems to be very similar across sign languages. In this construction, the agent is omitted and the event is expressed from the perspective of the patient using the phenomenon identifiable as CA (CA:patient).9 It has been identified at least in American Sign Language (Kegl 1990; Janzen et al. 2001; Rankin 2013), Irish Sign Language (Leeson 2001), French Sign Language (Guitteny 2006) and Catalan Sign Language (Barberà & Cabredo Hofherr 2017).

In this construction, the movement of the transitive, indicating verb is directed towards the body of the signer, and the signer represents the patient through CA. Dudis (2004) says that the body is partitioned: the hands produce a verb denoting the action of the agent, while the body physically represents the patient. According to Janzen et al. (2001), the most important characteristic is that the eye gaze represents the patient’s perspective, typically being directed towards the location associated with the agent. In addition, there may be other non-manual features that represent the perspective of the patient, such as leaning the body towards the location of the patient and using facial expressions to represent the emotions of the patient. Even though the movement of the verb indicates a location associated with an agent, no referent is associated with that location. Thus, even though an agent is semantically involved in the event described, the identity of the agent is not specified, i.e., it is non-referential. Example (7) is from ASL.

(7) American Sign Language (Janzen et al. 2001: 292)
  MY NAME M-J B-I-E-N-V-E-N-U / NAMEDa➔1 MJ (sign name)
  ‘My name is MJ Bienvenu. The name I have been given is “MJ”.’10

In (7), the situation is viewed from the patient’s perspective, the signer herself, and the patient is in focus. The hands in the verb NAMEDa➔1 are oriented so that the fingertips point towards the signer, and the movement is towards the signer. The agent is not overtly expressed. The movement of the verb begins at a location “a” in the signing space, but there is no referent associated with that location. This means that the agent is non-referential.

This kind of a construction has been labeled as passive by some researchers (Kegl 1990; Janzen et al. 2001; Guitteny 2006). The passive construction is said to contrast with the active in that in an active construction, the event is told completely from the agent’s perspective, whereas in the passive the perspective is that of the patient (see Janzen et al. 2001). In the active, the movement of the verb is away from the body of the signer towards a location in the signing space with which the patient is associated, and the non-manual features represent the agent’s perspective. However, as Barberà & Cabredo Hofherr (2017) point out, since the movement of the verb is not reduced in the so-called passive construction in ASL (in contrast to the analysis of Kegl 1990), the verb is not syntactically or morphologically detransitivized.

Barberà & Cabredo Hofherr (2017) argue that the High-locus construction in Catalan Sign Language, which shares many similarities with the constructions labeled as passive in other sign languages, should not be analyzed as passive but instead as a transitive construction that has a nonspecific subject comparable to agent defocusing constructions with an impersonal third person plural subject (Siewierska & Papastathi 2011). They propose that the omitted argument is a null pro subject which is licensed by agreement. According to Barberà & Cabredo Hofherr (2017), the movement of an indicating verb in a High-locus construction begins in a high location in the signing space, which has not been associated with any referent previously, and ends at the body of the signer with animate patients, or in the neutral signing space with inanimate patients. For plain verbs, the verb is articulated completely in a high location. They argue that the high location, which is used specifically for non-specific interpretation in Catalan Sign Language, contrasts with the low location associated with referential referents, and the construction cannot be analyzed simply as an omission of the agent (Barberà & Cabredo Hofherr 2017: 784–790). However, the use of CA representing the patient is not crucial but only an additional phenomenon used with animate patients.

In a study on defocusing the agent in ASL, Rankin (2013) found several constructions in which the agent is defocused on different levels. She analyzed the interplay of defocusing the agent on the semantic and structural levels. Rankin (2013: 63–64) noted that even if the “passive construction” described by Janzen et al. (2001) does exists in ASL, it is very rare in her elicited data. Rankin found only one occurrence corresponding to the “prototypical passive” (Janzen et al. 2001) in ASL, and in this utterance the patient was the signer herself (Rankin 2013: 64). Janzen et al. (2001: 301) proposed that the use of CA would be needed in order to use a passive construction for third person referents. However, Rankin shows that when the patient is not the first person referent, other strategies are used to defocus the agent.

The most common strategy in Rankin’s data was simply omission of the agent. When indicating verbs were used, the event was typically told from a neutral perspective, in which the signer does not identify herself with either the agent or the patient using CA (Rankin 2013: 32; cf. Janzen et al. 2001: 289). No special passive morphology is used, and the patient is not expressed in subject position (Rankin 2013: 45–55). Patients occur in the same position as that in which they occur in a clause that contains an overt subject, either as non-manually marked topics as in (8) or in a post-verbal object position, as in (9).

(8) American Sign Language (Rankin 2013: 47)
  ‘His bike is being repaired.’
(9) American Sign Language (Rankin 2013: 63)
  ‘Receipts should be given to the secretary tomorrow.’

The agent omission occurs in ASL with any kind of verbs: plain verbs such as FIX in (8), indicating verbs such as GIVE➔right in (9), as well as depicting verbs. Rankin (2013: 43) proposes that omission with plain verbs is the most complete way of defocusing the agent. She suggests that the defocusing is more complete when the verb does not refer to the agent at all (Rankin 2013: 58–65). In Rankin’s data, indicating verbs were used in two manners, leading to two different levels of defocusing. Either the beginning location of the movement was at the center of the signing space, not referring to a specific agent (see Liddell 2003: 117), as in (9), or the beginning place was a specific location. The former defocuses the agent to the full extent, the same way as plain verbs do, but the latter gives the agent some focus on a structural level while the agent is semantically non-referential.

A construction that omits the agent without using CA has been identified also in other sign languages (Cuxac 2000: 199; Guitteny 2006; Barberà & Cabredo Hofherr 2017). For French Sign Language, Guitteny (2006: 310–312) and Cuxac (2000: 199) have proposed that this can be done by modifying the indicating verb so that the location for agent is in the neutral space, not associated with any referent. This form appears to be similar to the first option of using indicating verbs with omitted agent in ASL (Rankin 2013), as in (9). Interestingly, Guitteny (2006: 311) found no evidence of agent defocusing by omission in clauses containing a verb articulated on the body of the signer, corresponding to plain verbs. In Catalan Sign Language, by contrast, agent omission (Non-agreeing central construction) has been analyzed as being limited to plain verbs (Barberà & Cabredo Hofherr 2017: 780).

Interestingly, Rankin’s (2013) data also contained utterances in which the event was told from the agent’s perspective using CA (CA:agent) even though the agent was omitted and non-referential (Rankin 2013: 65–68, 91; see also Barberà & Quer 2013: 250–252). In these utterances, the agent’s perspective was marked through the eye gaze (towards the patient) and facial expression (representing the emotions of the agent). These structures were much more common in narrative data than in single elicited utterances. To summarize, in ASL, in a construction in which the agent is defocused by not overtly expressing it, the event can be told from a neutral perspective, from the patient’s perspective or from the agent’s perspective. This implies, contrary to the suggestion by Janzen et al. (2001), that the use of non-manuals to show the perspective is not a crucial property of agent defocusing. The analysis carried out by Barberà & Cabredo Hofherr (2017) supports this observation.

Overt A arguments used non-referentially have been found in Catalan Sign Language (Barberà & Quer 2013) as well as American Sign Language (Rankin 2013). For Catalan Sign Language, Barberà & Quer (2013: 244–245) have found non-referential use of the lexical sign WHO (together with the third person plural pronominal form (pointing) or with the determiner SOME, accompanied with the mouthing alguien ‘someone’). Lexemes used this way are also “a sign that consists of a derived form of the lexical noun PERSON”, and the sign ONE. Signs used non-referentially are usually produced in the upper area of the signing space which, according to the researchers, is reserved for non-referential use, and the modification of the verbs is consistent with that location, as described earlier (Barberà & Cabredo Hofherr 2017). Barberà & Quer (2013: 245) also found that the third person plural pronominal pointing sign alone can be used non-referentially. As for American Sign Language, Rankin (2013: 68–73) has identified the use of impersonal pronouns such as NO-ONE, a plural noun with generic reference (PEOPLE), pronominal pointing signs used non-referentially, and a metonymic use of an inanimate entity as the agent, although these were not common in her data. It has been reported that the first person pronominal pointing sign is used for non-specific reference in Swedish Sign Language, referring anaphorically to a nominal with non-specific reference (Nilsson 2004).

It has been argued for some sign languages that the use of different handshape types of depicting verbs serves to focus on different participants in the event. The use of handling handshapes focuses on the agent, while the use of entity handshapes focuses on the patient, thus defocusing the agent (Benedicto & Brentari 2004; Börstell 2017). Börstell (2017) found that the use of handshapes and perspective interact as argument focusing strategies in Swedish Sign Language. According to the elicited data, agent perspective and handling handshapes are preferred when describing a two-participant situation. However, the agent can be defocused using an instrument handshape or patient perspective, or both.

3 Data and methodology

The main data of this study is a sample lasting 11.5 minutes (approximately 1200 sign tokens) from The Language Policy Programme for the National Sign Languages in Finland 2010. It contains three video clips of signing (6, 2 and 3.5 minutes, respectively). The main data contains approximately 50 clauses defocusing the agent, the number of two-participant clauses being 105. A random set of clips was quickly examined, and those that seemed to contain a lot of two-participant clauses were selected for careful analysis. They were annotated in ELAN digital annotation software (Crasborn & Sloetjes 2008). In addition to the main data, some supporting evidence has been sought in other parts of the Kipo Corpus (Finnish Association of the Deaf 2016), which contains the whole Language Policy Programme for the National Sign Languages in Finland 2010, annotated after the annotation of the main data of this study with ID glosses (a unique gloss identifying a lexeme) and a translation into Finnish. It can be found in the Language Bank of Finland. The main data comprises clips number 08, 09 and 47.

The data is a sample of informational texts, signed by one person at a time. It has two different signers, who present the vast majority of the language policy programme. They are both female, adult, native signers of FinSL, and right-handed. They are experienced translators and are known in the Finnish deaf community. The data is a free translation from Finnish. The text has been edited in FinSL rather than translated and, according to the publisher, special attention has been paid to the language, which suggests that it contains less interference from Finnish than translations generally. Sometimes the order of presenting things has been changed, and the perspective has been often changed from third person to first person. The text is targeted at the deaf community of Finland, and the aim is that all members of the community could easily understand it. It can be considered a good example of a public, videotaped, well prepared informational text in FinSL. The text also includes some short quotations from interviews conducted with signers of FinSL, as well as some proposals for action. The quotations have been first translated into Finnish and then translated back into FinSL.

The data was treated as an original text in FinSL, not as a translation. The search for two-participant clauses and agent-defocusing strategies was carried out on the sign language version, without reference to the original Finnish text, and the translations from FinSL into Finnish were made independently of the original text. The main data was annotated in multiple layers, each feature on a separate tier.

First, all the signs in the main data were annotated with glosses. The glosses are written in capital letters (e.g. SIGN), as is normal practice in sign language research. Pauses were annotated with a slash (/). The data was also segmented into utterances (signing between raising the hands and putting them down again, which was a common strategy in this data), and the utterances were translated into Finnish. After this, all the verbal predicates were identified, and the verbs containing semantically at least two distinct participants were searched for. The clauses built around these verbs were identified as two-participant clauses, and these clauses were analyzed in detail. A clause was thus analyzed as a two-participant clause if the verbal predicate in the clause expresses an event which semantically involves at least two participants, one of them being agent, whether the participants are overtly expressed or not. Also clauses including a clause as the P argument were considered to be two-participant clauses. Any kind of two-participant clause where the agent is not fully referential or the identity of the agent is somehow blurred was seen as an agent defocusing structure (Shibatani 1985; Siewierska 2008). The features of the two-participant clauses were annotated on several levels including information on, for example, the use of space, the use of non-manual articulators, semantics and sign order. Figure 1 illustrates the annotation tiers and annotations used for the analysis.

Figure 1
Figure 1

A screenshot from the ELAN annotation tool and annotations used in the analysis. The screenshot presents the sequence shown also in Figure 3.

Whether the agent was expressed with a noun phrase (NP) or a pronominal pointing sign (PT) or whether it was not overtly expressed (Ø) were all marked. The nominals or pronominal pointing signs expressing the participants of the predicate were marked either as A (the agent argument) or P (the patient argument), and the sign order of the clauses and possible omission of arguments were identified. With regard to pointing signs, markings were made for whether the pointing was directed towards the signer or another location and whether it contained an arch-like movement to indicate plurality. Whether a pointing sign or a null referred to a previously mentioned referent or whether it was non-referential were also analyzed. A non-first person pointing sign or a null was analyzed as anaphoric if there was a previous mention of a referent which could be understood as an antecedent. If there was no overt antecedent, it was analyzed as non-referential. Since the data used for this study contains only monologues signed to the camera in a studio, there are no referents physically present other than the signer. A third person referent cannot be denoted using a deictic pointing towards a present referent in the way described in Section 2.2.1. A non-first person pointing sign or a null can thus be fully referential in this data only if there is an overtly mentioned antecedent.

The verb classes of predicates (see Section 2.2.1) were coded as plain verbs, locatable plain verbs and indicating verbs. Depicting verbs were not found in the main data. It was marked whether the verb sign is modified and if it is, whether it indicates the agent and/or the patient.

The non-manual features used in the two-participant clauses defocusing the agent were annotated separately. The direction of the head, body and gaze were annotated. There were two kinds of body leans: sideways (right or left) and backwards. There was also one tier labeled as Facial expression, reserved for annotations on the facial expressions expressing the emotions or thoughts of a participant. The use of this kind of expression was rare. Finally, it was marked whether the non-manual features were used for expressing the perspective of the agent or the patient.

The annotations and observations on the agent-defocusing two-participant clauses were gathered in an Excel table, with the help of which they were classified and analyzed further. Clauses defocusing the agent were compared to clauses containing a referential agent. This study explores the possibilities for defocusing the agent in FinSL. Some tendencies can be stated, but the frequencies of each strategy are not given: numbers could be misleading, since the amount of data is relatively small, and many cases are somehow ambiguous.

4 Agent defocusing strategies in the data

The data shows that there are several strategies for defocusing the agent in FinSL. The agent can be omitted or expressed with a pronominal pointing sign used non-referentially. The data of this study indicates that the most common strategy for agent defocusing is omission. The pronominal pointing sign used non-referentially as the A argument can be either a non-first person plural pointing sign or a first person pointing sign. Constructed action (CA) may be used with omitted agents so that the signer represents either the patient or the agent, but more often CA is not used. It seems that CA can be used with pronominal pointing signs too, representing the patient with a non-first person pointing sign, and representing the agent with a first person pointing sign. This section presents the different agent defocusing strategies found in the data. The section is divided into two main parts: omission of the agent, and pronominal pointing signs used non-referentially.

4.1 Omission of the agent

4.1.1 Simple omission

The most common strategy for defocusing the agent in the data is omission of the agent. Usually it involves no CA. In these clauses, the agent is not mentioned at all, and the clause is similar in form to a clause where a referential agent is omitted. However, there is no antecedent to which a null anaphora could refer, and there is no physically present referent that could be denoted. The agent is unknown, irrelevant or obvious. Example (10) is a clear example of an omitted agent.

  ‘The first deaf club was founded in Turku in 1886.’ (Kipo 08, 1:23)11

In (10), the patient is expressed at the beginning of the clause with the noun phrase FIRST DEAF-CLUB, the verb FOUND follows it, and the peripheral elements expressing place and time are expressed at the end of the clause. The predicate is a locatable plain verb which is not modified spatially and thus tells nothing about the semantic roles or the identity of the participants. The agent is completely defocused.

The patient is usually expressed before the verb, as in (10), the sign order being PV, but it can just as well follow it (sign order VP), or the patient can be omitted. Also in clauses containing a null anaphora, these same orders occur (Ala-Sippola 2012; Jantunen 2013), so the sign order does not make a difference.

Agent omission is used both with plain verbs and with indicating verbs, but plain verbs occur more often in the data. If a locatable plain verb is located meaningfully, the location in which the sign is articulated indicates the location of the event rather than the agent. Indicating verbs can be modified or not. If the patient is a non-first person referent, the verb can indicate the patient through the end location of the movement or the orientation.

In clauses containing a simple omission, the patient is usually a non-first person referent, but can also be the first person. The data contained one clause with a simple omission where the patient was the first person. It is shown in (11). Figure 2 represents the sequence LOOKright➔1 TRUST OBJPRO➔1.

Figure 2
Figure 2

The signs of the sequence LOOKright➔1’ TRUST OBJPRO➔1 ‘we are not trusted’.

  rh: LOOKright➔1’ TRUST OBJPRO➔1
  lh: LOOKleft➔1’
  ‘Even when we get employment, we are not trusted.’(Kipo 47, 0:56)12

Example (11) contains two clauses defocusing the agent, built around the predicates LOOK and TRUST, where the patient is the first person and the agent is omitted. The verb LOOK is produced with both hands, carrying the meaning that the agent is plural. It is directed so that the beginning place is to the right for the right hand and to the left for the left hand. No referent is associated with these locations. The end location is close to the signer. The agent is understood to be someone other than the signer and the patient is first person plural, appearing here as a null anaphora, the antecedent of which is expressed at the beginning of the example with the sign PT➔1-pl. The second clause contains the predicate TRUST, which is a plain verb and cannot be modified. It is followed by the sign glossed as OBJPRO➔1 (see Börstell 2017: 155–173), expressing the patient. Being articulated on the body of the signer, it indicates that the signer is the patient. As can be seen from Figure 2, during the predicate LOOK and TRUST, the eye gaze is directed towards the camera and the body posture is neutral. Also the facial expression of the signer is rather neutral, i.e., there is no CA representing the patient’s perspective. The negation is expressed with a headshake.

Simple omission is used with both animate and inanimate patients, most patients in the data being inanimate. In (10) the patient is inanimate, while in (11) it is animate. There are also cases where the P argument is a clause.

4.1.2 Omission with constructed action representing the patient

Some clauses with an omitted agent include non-manual elements that can be analyzed as reduced or subtle CA (Cormier et al. 2015) representing the patient (CA:patient). These two types of CA are not systematically separated here, and they could equally well be referred to as non-overt CA (see Cormier et al. 2015). The CA is not overt since the hands are always used to produce lexical signs, i.e., the predicate of the clause, while in overt CA, all the articulators of the signer would be used to represent the actions and emotions of the participant (Cormier et al. 2015). In these clauses, the agent is omitted and non-referential, as in the simple omission described in 4.1.1. In addition, non-manual elements are used to represent the perspective of the patient. Non-manual features that are used include body lean sideways, body lean backwards, head directed towards the agent and eye gaze directed towards the agent. The facial expression tends to be rather neutral. None of the features is used in every clause of this type, and the number and combination of non-manual articulators vary. Sometimes the non-manual elements are so weak that it is open to interpretation whether there is any CA or not.

One clear instance of the use of CA, which can be analyzed as reduced CA, is presented in Figure 3.

Figure 3
Figure 3

A clause defocusing the agent with the omission of the agent and constructed action representing the patient.

The verbal predicate NOTIFY (framed in Figure 3) is articulated so that the hand touches the signer’s right shoulder. The agent is associated with the space to the right of the signer. The non-manual elements start already during the sign DAY and continue during the articulation of the predicate, gradually neutralizing. Eye gaze and head are directed to the right of the signer, towards the place of the non-referential agent. The body leans left and backwards, away from the agent. The eyes are widened, which can be interpreted as expressing the emotions of the patient, such as being surprised.

In these clauses, an indicating verb is directed from a semantically empty space towards the body of the signer. The modification of the verb indicates that the signer (or the referent associated with the signer) is the patient, but the patient is usually also lexically unexpressed. The beginning space of the movement can be in a high location in the signing space, but it does not need to be. In Figure 3 it is at a neutral level. In two instances the agent seems to be associated with a high location in the signing space. The upper space is analyzed as being used iconically to reflect the real or metaphorical relations of the situation. For example, with the predicate TEACH, the signer iconically reflects pupils’ perspective: pupils look up to their teachers.

In all the clauses where the agent is omitted and CA is used to represent the patient, the patient is actually the first person referent. Either the meaning is ‘me’, when the signer is quoting a deaf person interviewed, or the meaning is ‘we the deaf’, which is expressed by the signs PT➔1-pl DEAF. The leaning of the body does not change the referent associated with the signer. The data contains no cases where the signer chooses to represent a third person referent using CA:patient in order to omit the agent.

Example (12) shows that the modification of the verb in agent defocusing constructions with omission of the agent and CA:patient is similar to the modification used in clauses where there is a referential agent and CA:patient is used. The relevant parts of the example are illustrated with still images in Figure 4.

Figure 4
Figure 4

The signs WORK+GIVE ‘employer’ and REFUSE ‘refuse’.

  ‘The employer restricted a deaf employee’s use of an interpreter. He needed an interpreter for some of the tasks, but the employer wouldn’t allow it.’ (Kipo 47, 02:46)13

In (12), the referent ‘deaf’ is associated with the body of the signer using the signs DEAF INDIVIDUAL➔1. With the compound WORK+GIVEright➔1, the referent ‘employer’ is associated with the space to the right of the signer (Figure 4a). In the next utterance, an indicating verb REFUSE is directed from the right towards the signer (Figure 4b), indicating that the agent is the employer and the patient is the deaf employee. In this clause, the signer represents a third person referent. The facial expressions of the signer represent the emotions of the deaf employee, the patient of the predicate REFUSE.

This clause resembles the clause in Figure 3. In both clauses, the verbal predicate is directed from the agent associated with the space to the right of the signer towards the signer, who represents or is the patient. The CA is used to represent the patient. In the clause in Figure 3, the agent is non-referential, and in the clause in (12) the agent is referential. To conclude, in clauses with omission of the agent and CA:patient, there seems to be no formal difference between clauses with a referential agent and clauses without one. The modification of the verb is similar, CA is used to represent the patient, and both agent and patient are omitted.

However, since the signer has chosen to represent the situation from the perspective of ‘the deaf employee’, the patient of the predicate REFUSE, the patient is more focused than is the agent ‘the employer’. It can thus be analyzed that the agent is defocused to some extent. There is a functional similarity with the English passive, where the agent is expressed in a by-phrase: He was forbidden by the employer to bring an interpreter.

4.1.3 Omission with constructed action representing the agent

Some clauses can be analyzed as including reduced or subtle CA representing the omitted, non-referential agent (CA:agent). The example in Figure 5 is the clearest instance of this. It contains subtle or reduced CA used with the predicate.

Figure 5
Figure 5

A clause defocusing the agent with the omission of the agent and constructed action representing the agent.14

The predicate is analyzed as consisting of two verbs, ANNOUNCE and NOTIFY. ANNOUNCE is a plain verb. NOTIFY is an indicating verb, the same as was used in the example in Figure 3, but here it is directed in the opposite direction. It is directed towards the location to the right of the signer, which associates the referent ‘population register’, the patient, with that location. The verb is directed slightly toward the upper area of the signing space. Simultaneously with producing the predicate, the signer turns her head to the right, and leans her body slightly to the left or backwards, away from the patient, associating herself not with the patient but with the agent. This can be analyzed as CA:agent. However, the agent is not mentioned in this clause and there is no antecedent. The agent of the predicate ANNOUNCE NOTIFY can be understood as anyone (or everyone) who has registered their mother tongue as sign language in the population register.

The predicate is followed by a clause complement, MY MOTHER+LANGUAGE PI SIGN+LANGUAGE, which is an instance of quotational CA. The person quoted is the non-referential, defocused agent. However, the non-manual elements expressing CA during this clause are very few. In some clauses, only the head expresses CA, being directed towards the space associated with the patient.

4.2 Pronominal pointing signs used non-referentially

The data contains some clauses in which the A argument is a pronominal pointing sign that is being used non-referentially. These correspond to the R-impersonals (Siewierska 2011): the clause has the form of a normal personal construction, but the A argument is human and non-referential. The data of this study suggests that at least the non-first person plural pointing sign and first person singular pronominal pointing sign, and arguably also first person plural pointing sign are used this way.

4.2.1 Non-first person plural pointing sign as the A argument

The data contains some clauses in which a non-first person plural pointing sign is used non-referentially as the A argument. An example is given in Figure 6.

Figure 6
Figure 6

A clause with a non-first person pointing sign as the A argument and constructed action representing the patient.

In Figure 6, the first sign SOCIETY is analyzed as the topic, marked with widened eyes, raised eyebrows and an eye blink, that sets a spatial interpretative framework for the following clause, the comment (Jantunen 2013: 313). The comment clause begins with an A argument, which is a non-first person pronominal pointing sign, an arch-like pointing to the right. The movement begins slightly to the left and it is directed slightly upwards. It has the form of a pointing that points either anaphorically to a referent that is associated with that space or to a physically present referent, but there is no lexically expressed referent associated with that space or a physically present referent to which the pointing would be directed. This means that the pointing sign is non-referential, i.e., the agent is defocused. The upper space is not analyzed as carrying a special non-specific meaning (cf. Barberà & Cabredo Hofherr 2017). Probably the people who see the deaf as handicapped are conceptualized as metaphorically higher, and also the motoric easiness may affect the articulation. As the example in Figure 5 shows, referential agents can be associated with the upper space as well as non-referential agents, and therefore referential pointing signs can be directed upwards as well. The verbal predicate LOOK is produced with both hands, expressing plurality. It is directed from the space to the right of the signer with the right hand and from the space to the left of the signer with the left hand towards the signer, the patient. The agent is understood to be someone other than the signer, and plural. The patient is expressed after the predicate with the signs PT➔1-pl DEAF ‘we, the deaf’.

Figure 6 also includes CA. While articulating the verbal predicate LOOK, the head turns to the right, towards the space associated with the agent. Also the body turns in that direction and leans back, away from the agent. These non-manual features are similar to those used with omitted agents when CA is used to represent the patient. Therefore it seems that CA:patient can also be used when the agent is expressed with a non-first person pointing used non-referentially. However, this was the only clause of this type. Other clauses containing a non-first person pointing sign include no CA.

The non-first person pointing sign is used with both plain and indicating verbs. When an indicating verb is used, it is modified the same way as in Figure 6 in both of the two clauses in the data. One of the pointing signs was produced with both hands, the right hand directed to the right and the left hand directed to the left. This sign also had a mouthing with the meaning ‘others’, and it could be argued that this is actually a lexicalized sign. The other non-first person pointing signs were directed to the right. It seems that the patient can be animate or inanimate, and first person or non-first person. The sign order varies a lot.

The agent expressed with a non-first person plural pointing sign is interpreted as excluding at least the signer, possibly also the interlocutor as the third person plural subject cross-linguistically (Siewierska & Papastathi 2011). In the clauses in the data, the agent is understood to be human, corresponding to the definition of R-impersonals (Siewierska 2011), but so are all the agents in the data.

4.2.2 First person pointing sign as the A argument

The data also contains some instances of a first person pronominal pointing sign used non-referentially as the A argument. The clause is constructed in such a way that the agent is expressed with a pronominal pointing sign towards the chest of the signer, denoting typically the signer herself or another referent associated with the signer’s body through CA. However, it can be inferred from the context that the signer does not mean herself, nor is another referent associated with the signer’s body. Example (13) is an excerpt from a section where the signer is giving some recommendations to improve the situation of sign language users. The first person singular pointing sign used in the example is illustrated in Figure 7.

Figure 7
Figure 7

First person singular pointing sign (PT➔1) used in Example (13).

  ‘[The teaching of sign language as a mother tongue to children using sign language must be developed.] If the mother tongue is properly acquired, it is easier to learn other languages.’ (Kipo 35, 0:06)

This excerpt is from the very beginning of the section. The first utterance is presented only as a translation in square brackets. The first person pronominal pointing, i.e., pointing towards the chest of the signer, occurs in the second utterance. The clause begins with the noun phrase MOTHER+LANGUAGE ‘mother tongue’ expressing the patient. It is followed by the expression of an agent, a first person singular pointing sign, which is followed by the predicate, the plain verb ACQUIRE. It is clear from the context that the signer is not talking about herself but about potentially anyone. The agent can be anyone, including the signer and the interlocutor. This usage is similar to the non-referential usage of the first person pronoun in Finnish (Helasvuo 2008). It also has a conditional context, which is characteristic of a non-referential interpretation in Finnish.

When a non-referential first person pointing sign is followed by a null anaphora, this also has a non-referential reading. The last clause in (13) contains a P argument, PT➔front-pl LANGUAGE ‘other languages’, expressing the patient, a predicate LEARN, and a null anaphora expressing the agent. The null anaphora refers back to the non-referential first person singular pointing sign in the previous sentence, and therefore also the null anaphora is non-referential.

In addition to a first person singular pronominal pointing, there are one or two occurrences of a first person plural pointing sign used non-referentially. In one of them it is not clear from the articulation whether the pointing is intended to be singular or plural. This clause also contains CA:agent. It seems thus that CA:agent can be used with the first person pointing sign as the A argument, when the pointing sign is used non-referentially.

5 Discussion

This study has shown that Finnish Sign Language has similar agent-defocusing strategies to spoken languages (Shibatani 1985; Siewierska 2011): omission of the agent and the use of pronominal pointing signs, which are the functional equivalents of pronouns in spoken languages. An important finding is that the most common strategy, according to this data, seems to be simple omission: it appears to be common to simply omit the agent without any modifications to the structure of the clause or the use of constructed action (CA). This is in line with Rankin’s (2013) findings and contrary to the suggestion of Janzen et al. (2001) for ASL that the use of CA would be necessary for omission to occur.

The pronominal pointing signs that are used non-referentially include at least the non-first person plural pronominal pointing sign and the first person singular pointing sign. Apparently also the first person plural pointing sign is used this way, but more research is needed to confirm this. The non-referential use of the first person pronoun appears not to be very common cross-linguistically (see Siewierska 2004). It is possible that the use of the first person pronoun for agent defocusing in Finnish (Helasvuo 2008) has influenced FinSL. Also, the non-specific use of the first person pronominal pointing sign in Swedish Sign Language (Nilsson 2004), which is related to FinSL, may have influenced FinSL. Another possibility is that CA, in which the signer identifies herself with a referent and which is a common strategy in sign languages, has an effect on the use of the first person pronominal pointing sign non-referentially, allowing the interlocutor to identify herself with the agent. It would be interesting to examine whether the first person pointing sign is commonly used for agent defocusing in other sign languages, and to what extent it is combined with CA:agent.

The use of a third person plural pronoun, by contrast, is common in spoken languages (Siewierska 2004; Siewierska & Papastathi 2011), and the omission of the agent is not uncommon, either (Shibatani 1985: 836; Siewierska 2008; 2013). In this respect, Finnish Sign Language does not seem to differ dramatically from spoken languages in terms of agent defocusing. Even though the pronominal pointing signs can be used gesturally, pointing to the physically present referents that are denoted (see Cormier et al. 2013; Johnston 2013), their use in agent defocusing appears to be very similar to the use of pronouns in agent defocusing. At least the use of pronominal pointing signs, possibly also the omission, are comparable to so called R-impersonals (Siewierska 2011): the form of the clause is the same as in a clause containing a referential agent, but the form that fills the place of an A argument (pronominal pointing sign or a null) does not refer to a specific referent. As for the omission, the analysis here is based on the assumption that the verbs in these clauses are transitive. However, the possibility that at least some of the verbs are ambitransitive, with the possibility of having only one argument referring to the patient, cannot be excluded (see also Hansen 2007). In this case there would be no omission of the agent. More research on these verbs is needed.

One issue that needs to be addressed here is the existence of a passive construction in FinSL. This paper argues that FinSL does not have a morphological passive construction. The agent defocusing strategy in which the agent is omitted and the event is told from the patient’s perspective using CA:patient was the best candidate for analysis as passive, since the same kind of construction has been analyzed as passive in ASL (Janzen et al. 2001), French Sign Language (Guitteny 2006) and Irish Sign Language (Leeson 2001). However, there are several characteristics that do not fit e.g. the definition of a prototypical passive put forward by Shibatani (1985: 837). According to Shibatani, a prototypical passive has the following characteristics:

a. Primary pragmatic function: Defocusing of agent.
b. Semantic properties:
  (i) Semantic valence: Predicate (agent, patient).
  (ii) Subject is affected.
c. Syntactic properties:
  (i) Syntactic encoding: agent ➔ (not encoded).
                                 patient ➔ subject.
  (ii) Valence of P[redicate]: Active = P/n;15
                                      Passive = P/n–1.
d. Morphological property: Active = P;
                                        Passive = P[+passive].

Let us compare the English clause in (1) (The house was built last year) and the FinSL clause in Figure 3 (SOMETIMES SAME DAY NOTIFYright➔1 AFTERNOON MEETING) with this characterization of the prototypical passive. The English clause is a prototypical passive. First of all, the agent is not encoded, and the patient is coded as the subject. In the FinSL clause, the agent is not encoded either, but the patient is not encoded as the subject. The patient is retrieved on the basis of the modification of the verb and the use of CA, but it cannot be analyzed as a subject, for two reasons. One reason is that the patient is not overtly expressed at all. The other reason is that the criteria for a grammatical subject in FinSL have not yet been identified by researchers, and the feature might not even be useful for the description of FinSL (see e.g. Jantunen 2008). Secondly, in the English passive, the valence of the predicate is reduced so that the clause is structurally intransitive. In FinSL, the valence is not decreased. Thirdly, in the English passive, the verb has a morphological passive marking, while in FinSL there is no morphological passive marking on the verb. The form of the verb is similar to that of a verb in a clause containing a referential agent, whether overtly expressed or omitted. CA is also used in the same way with referential and non-referential agents. In addition, since CA is considered a gestural phenomenon (Cormier et al. 2015), not a morphological one, elements of CA cannot be considered to be markers of a morphological passive.

Semantically and pragmatically, the FinSL construction in which the agent is omitted and CA:patient is used has similarities with the prototypical passive defined by Shibatani: the primary function of the clause is defocusing of the agent, and the predicate contains semantically an agent and a patient. It cannot be said that the subject is affected since the clause does not contain an overt grammatical subject, as was stated in the previous paragraph.

Functionally, the construction has clear similarities with a morphological passive. The agent is defocused, and the agent is not overtly expressed. Further, the event is told from the patient’s perspective, which, it has been claimed, is one function of a passive (Shibatani 1985: 832). Choosing to represent either the patient or the agent through CA, the signer can choose to what extent she wishes to defocus the agent or underline the patient. The strategy in which the signer represents the patient through CA corresponds relatively well with the function of a morphological passive. Whether the agent is overtly expressed or not, the perspective is that of a patient, and the patient is more prominent than the agent, as in morphological passives in spoken languages.

Even if the agent is referential, as in (12), the clause corresponds functionally to an English passive in which the agent is mentioned in a by-phrase (e.g. The house was built by Joe). In both structures, the agent is mentioned but it is not in a prominent position. If the use of CA:agent is understood as the norm, the use of CA:patient diminishes “the prominence or salience from what is assumed to be the norm” (see Siewierska 2008). This can be compared with not mentioning the agent as the subject, which is the norm in accusative languages. However, further investigation is required to discover whether the use of CA:agent is indeed more frequent than the use of CA:patient. The claim that FinSL has no morphological passive should not be surprising given that, like other sign languages, FinSL makes use of space and gesturality (such as pointing, verb modification and CA) to express meanings that are generally expressed in spoken languages using morphology and syntax. This also means that earlier studies on “passives” in sign languages need to be reconsidered to see whether the constructions defocusing the agent should be analyzed as morphological passives or something else.

It should be noted that none of the strategies for defocusing the agent in FinSL differs essentially from the corresponding structure that contains a fully referential agent. Syntactically, the sign order varies with non-referential agents just as with referential agents, whether overtly expressed or not, and in both cases it is typical to leave the agent unexpressed. There is no promotion of the patient as subject. The pronominal pointing as A argument is similar in form whether it is referential or not. There is no morphological marking of the verb in any of the strategies, nor does the spatial modification of the verb seem to be different. For example, no evidence was found that the upper area of the signing space is restricted to or specially kept for non-specific referents, as has been suggested for Catalan Sign Language (Barberà & Cabredo Hofherr 2017). The upper space can be used both with referential and non-referential agents and is probably linked to the habit of associating authorities with upper space, not with non-specificity. However, it could be useful to analyze the modification of verbs and the sign order more closely to see what tendencies there might be, for example whether it is more typical to express the patient before the predicate when the agent is defocused by omission. Also CA can be used to represent either the agent or the patient with non-referential agents as well as with referential agents. None of this is exceptional cross-linguistically. The non-referential use of pronouns is common also in spoken languages (Siewierska 2011), the only difference from the referential use of pronouns being the non-existence of an antecedent, or the need to interpret the difference from the context.

This study has shown that the use of CA is not necessary for the omission of the agent in FinSL. Rather, it seems to be more typical to omit the agent without CA, even if the patient is animate, which contrasts to the analysis of Barberà & Cabredo Hofherr (2017) on Catalan Sign Language. Simple omission was the most common strategy with third person patients, contrary to the proposition of Janzen et al. (2001) for ASL. Omission with CA representing the patient is used in the data only with first person patients, as in Rankin’s (2013) data on ASL. However, a clause where the patient is the signer can also be expressed without CA. No environment was found in which the use of CA is obligatory. It has been argued that the use of CA is an optional strategy in sign languages (Ferrara & Johnston 2014; Jantunen 2017a), and this study suggests that this applies also to non-referential agents in FinSL. Since CA seems to be very similar across the sign languages studied so far, it is reasonable to question how obligatory CA:patient is in clauses defocusing the agent in other sign languages, too.

Further, this study suggests that CA can be used in agent defocusing constructions in FinSL to represent not only the patient but also the agent, similarly to ASL (Rankin 2013). It seems, too, that the same construction, in which the event is told from the patient’s perspective and the agent is omitted, can be used whether the agent has been previously mentioned or not, i.e., whether it is referential or non-referential. This means that this construction could be analyzed as R-impersonal (see Barberà & Cabredo Hofherr 2017). The use of CA with different agent defocusing strategies and its role in agent defocusing need to be studied in depth with a larger set of data.

The data used in this study was quite small and contained only one text type. Also, it lacks the third main type of verbs, depicting verbs, altogether. However, it shows a relatively rich range of different agent defocusing strategies in FinSL. As a first study focusing on this subject, it gives a good overview of how an agent can be defocused in FinSL. It also leaves some questions open for further research, such as the distribution and semantic/pragmatic differences of different strategies, the frequency of different constructions, the role of CA in agent defocusing constructions and the use of different handshape types for defocusing with depicting verbs. These questions should be investigated using much wider data comprising a range of data types. With a different kind of data, the frequency of different strategies might look very different, and there might even be other strategies that were not found here at all. The themes discussed in The Language Policy Programme for the National Sign Languages in Finland 2010 are quite abstract, there might be interference from Finnish, since the original text was written in Finnish, and e.g. CA might be used more when more concrete events are described and depicting verbs are also used (see Jantunen 2017a). Also, with depicting verbs, it is possible that the choice of handshape type is used as a means of defocusing (see Benedicto & Brentari 2004; Börstell 2017).

6 Conclusion

This study has shown that there are several strategies for defocusing the agent in FinSL. The agent can be left unexpressed or it can be expressed with a pronominal pointing sign used non-referentially. The most common strategy for agent defocusing in the data was simple omission. Pronominal pointing signs that can be used non-referentially include at least the non-first person plural pronominal pointing sign and the first person singular pointing sign, possibly also the first person plural pointing sign.

In addition, the event can be represented from the perspective of a patient using CA (CA:patient), or it can be represented from the perspective of the agent (CA:agent) even though the agent is non-referential. The study has shown that CA can be combined with different agent defocusing strategies. It is argued, however, that the use of CA is an optional strategy that is rarely used in the data of informational texts. Most often, the event is told from a neutral perspective without CA. In sum, in line with what has been argued about CA in general, CA seems to be an optional strategy also with non-referential agents in FinSL (Jantunen 2017a; see also Ferrara & Johnston 2014).

The form of clauses defocusing the agent is similar to that of clauses including a referential agent. None of the strategies is defined as a passive in this study. They all defocus the agent, which is the principal function of a prototypical passive (Shibatani 1985), but there is no promotion of the patient to subject, and no morphological marking of the verb or valence reduction, all of which are defining properties of a morphological passive.

This study is the first step towards understanding agent defocusing in FinSL, and contributes to the study of such constructions in other sign languages. More study with more data and frequencies is needed for deeper understanding of the variety of agent defocusing strategies in FinSL and in other sign languages.


1SG = first person singular, 3SG = third person singular, A = A argument, ASL = American Sign Language, Auslan = Australian Sign Language, CA = constructed action, CA:agent = constructed action used for representing the agent, CA:patient = constructed action used for representing the patient, fbuoy = fragment buoy, FinSL = Finnish Sign Language, INESS = inessive, INF = infinitive, lh = left hand, OBJPRO = object pronoun, P = P argument, pl = plural, POSS = possessive pointing sign, PT = pointing sign, PTV = partitive, Q = question particle, rh = right hand, V= verbal predicate


  1. See footnote 6. [^]
  2. Agent defocusing in German Sign Language has been discussed by Hansen (2007). Unfortunately, only the dissertation abstract is available in English to the international research community. [^]
  3. In FinSL research, the term verbal is used to refer to one of the two main sign classes and it is broader than the term verb, see Jantunen (2010). In this article, the term verb is used for the sake of simplicity. Also, the classification of FinSL verb(al)s (Jantunen 2010) differs in some respects from the classification of e.g. Liddell (2003) for ASL verbs, but the differences do not affect the analysis of this study. [^]
  4. Max Planck Institute for Psycholinguistics, The Language Archive, Nijmegen, The Netherlands: see https://tla.mpi.nl/tools/tla-tools/elan/. [^]
  5. The glossing of the example has been slightly modified from the original in order to be consistent with the glossing in this paper. [^]
  6. A refers to the A argument, the lexical unit which expresses the agent, P refers to the P argument, the lexical unit which expresses the patient, and V refers to the verb. [^]
  7. The convention of sign language linguistics is to gloss a sign with capital letters. [^]
  8. Here the locatable verbs of FinSL are classified as plain verbs. In some studies, they are classified as indicating verbs (see e.g. de Beuzeville et al. 2009 for Auslan). There is effectively no research on locatable verbs in FinSL and the use of space with them. [^]
  9. The abbreviation CA:patient is used when the CA is used to represent the patient, and CA:agent is used when the CA is used to represent the agent. [^]
  10. The notation of other researchers has been slightly modified to fit the notation of this article. The modification of the verbs is marked in superscript. The starting location is marked before the arrow if the verb is modified for agent, and the ending location is marked after the arrow if the verb is modified for patient. The actual side (right or left) is marked when known, but as the original example here contains only the glosses and the location is not specified, notation “a” for a location in the signing space is kept. Example (7) is presented only with glosses since the original example exists only with glosses. In the analysis of the data (Section 4), still images from the video data are used with many examples in order to give a better understanding of the form of the constructions. Letters separated by a dash are used to represent the use of manual alphabets. [^]
  11. A hyphen between words is used when more than one (English) word is needed to gloss one sign (this is the convention in sign language linguistics). Square brackets in glossing are used to mark the participants and the predicate: P stands for patient, V for predicate and A for agent. In brackets, the clip number in the Kipo corpus and the beginning time (m:ss) of an example are given. [^]
  12. A plus sign is used between consecutive signs in compounds. An apostrophe in the glossing means some kind of a change in prosody, i.e., in nonmanual behavior (Jantunen 2008). [^]
  13. The sign class of the compound glossed as WORK+GIVEright➔1 is ambiguous. The sign GIVEright➔1 is modified like an indicating verb. However, the overall interpretation is nominal, the meaning being ‘employer’. This is interpreted as a nominalization (without overt marking). [^]
  14. The sign glossed as PI is a modal element expressing certainty that can be used in equative sentences, see Jantunen (2007). POPULATION-REGISTER is here marked as P. The analysis is based on the fact that the modification of the indicating verb and CA are motivated by the location of the referent in question. Simplification is used here, since the focus is on agents. Alternatively, the sign could be interpreted as an R argument. Also, in Figure 3, the addressee of the verb NOTIFY is analyzed as patient. [^]
  15. Here, P refers to predicate. The citation has been kept in its original form, although elsewhere in this article P is used to refer to P argument (expressing the patient). [^]


The author wishes to thank Eleanor Underwood for checking the English of the paper. The author also wishes to thank the three anonymous reviewers for many helpful comments on an earlier version of this paper.

Funding Information

This research was supported by funding from Kone Foundation.

Competing Interests

The author has no competing interests to declare.


Ala-Sippola, Sanna. 2012. Agentin ilmaiseminen suomalaisella viittomakielellä tuotetussa asiatekstissä [Expressing agent in an informational text in FinSL]. Helsinki: University of Helsinki MA thesis.

Barberà, Gemma & Joseph Quer. 2013. Impersonal reference in Catalan Sign Language (LSC). In Laurence Meurant, Aurélie Sinte, Mieke Van Herreweghe & Myriam Vermeerbergen. Sign language research, uses and practices: Crossing views on theoretical and applied sign language linguistics, 237–258. Boston/Berlin: De Gruyter Mouton. DOI:  http://doi.org/10.1515/9781614511472.237

Barberà, Gemma & Patricia Cabredo Hofherr. 2017. Backgrounded agents in Catalan Sign Language (LSC): Passives, middles, or impersonals? Language 93(4). 767–798. DOI:  http://doi.org/10.1353/lan.0.0028

Benedicto, Elena & Diane Brentari. 2004. Where did all the arguments go?: Argument-changing properties of classifiers in ASL. Natural Language & Linguistic Theory 22(4). 743–810. DOI:  http://doi.org/10.1007/s11049-003-4698-2

Börstell, Carl. 2017. Object marking in the signed modality: Verbal and nominal strategies in Swedish Sign Language and other sign languages. Stockholm: Stockholm University dissertation. Available: http://su.diva-portal.org/smash/get/diva2:1088175/FULLTEXT01.pdf

Comrie, Bernard. 1977. In defense of spontaneous demotion: The impersonal passive. In Peter Cole & Jerrold Sadock (eds.), Grammatical relations (Syntax and Semantics 8), 47–58. New York: Academic Press.

Cormier, Kearsy, Adam Schembri & Bencie Woll. 2013. Pronouns and pointing in sign languages. Lingua 137. 230–247. DOI:  http://doi.org/10.1016/j.lingua.2013.09.010

Cormier, Kearsy, Sandra Smith & Zed Sevcikova-Sehyr. 2015. Rethinking constructed action. Sign Language & Linguistics 18(2). 167–204. DOI:  http://doi.org/10.1075/sll.18.2.01cor

Crasborn, Onno & Han Sloetjes. 2008. Enhanced ELAN functionality for sign language corpora. In Proceedings of LREC 2008, Sixth International Conference on Language Resources and Evaluation.

Cuxac, Christian. 2000. La Langue Des Signes Française (LSF): Les voies de l’iconicité (Faits de Langues). Paris: Ophrys.

de Beuzeville, Louise, Trevor Johnston & Adam Schembri. 2009. The use of space with indicating verbs in Auslan: A corpus-based investigation. Sign Language & Linguistics 12(1). 53–82. DOI:  http://doi.org/10.1075/sll.12.1.03deb

Dixon, Robert M. W. 1979. Ergativity. Language 55. 59–138. DOI:  http://doi.org/10.2307/412519

Dixon, Robert M. W. 2010. Basic Linguistic Theory vol. 1 Methodology. Oxford; New York: Oxford University Press.

Dowty, David. 1991. Thematic protoroles and argument selection. Language 67. 547–619. DOI:  http://doi.org/10.1353/lan.1991.0021

Dudis, Paul G. 2004. Body partitioning and real-space blends. Cognitive Linguistics 15(2). 223–238. DOI:  http://doi.org/10.1515/cogl.2004.009

Fenlon, Jordan, Adam Schembri & Kearsy Cormier. 2018. Modification of indicating verbs in British Sign Language: A corpus-based study. Language 94(1). 84–118. DOI:  http://doi.org/10.1353/lan.2018.0002

Ferrara, Lindsay & Trevor Johnston. 2014. Elaborating who’s what: A study of constructed action and clause structure in Auslan (Australian Sign Language). Australian Journal of Linguistics 34(2). 193–215. DOI:  http://doi.org/10.1080/07268602.2014.887405

Finnish Association of the Deaf. 2016. Kipo Corpus. The Language Policy Programme for the National Sign Languages in Finland 2010 Corpus, version 2. [cited 21.02.2018]. Available: http://hdl.handle.net/11113/00-0000-0000-0000-340D-2@view.

Guitteny, Pierre. 2006. Le passif en langue des signes. Bordeaux: Université Michel de Montaigne Bordeaux 3 dissertation.

Hansen, Martje. 2007. Warum braucht die Deutsche Gebärdensprache kein Passiv? Verfahren der Markierung semantischer Rollen in der DGS [Why can German Sign Language (DGS) do without a passive construction? Ways of marking semantic roles in DGS]. Sign Language & Linguistics 10(2). 213–222. DOI:  http://doi.org/10.1075/sll.10.2.09han

Helasvuo, Marja-Liisa. 2008. Minä ja muut: Puhujaviitteisyys ja kontekstuaalinen tulkinta [Me and the others: Speaker reference and contextual interpretation]. Virittäjä 112(2). 186–206. Available: https://journal.fi/virittaja/article/view/40660.

Huang, Yan. 2000. Anaphora: A cross-linguistic study (Oxford Studies in Typology and Linguistic Theory). Oxford: Oxford University Press.

Jantunen, Tommi. 2007. The equative sentence in FinSL. Sign Language & Linguistics 10(2). 113–143. DOI:  http://doi.org/10.1075/sll.10.2.04jan

Jantunen, Tommi. 2008. Fixed and free: Order of the verbal predicate and its core arguments in declarative transitive clauses in Finnish Sign Language. SKY Journal of Linguistics 21. 83–123. Available: http://www.linguistics.fi/julkaisut/SKY2008/Jantunen_NETTIVERSIO.pdf.

Jantunen, Tommi. 2010. Suomalaisen viittomakielen pääsanaluokat [The main parts of speech in FinSL]. In Tommi Jantunen (ed.), Näkökulmia viittomaan ja viittomistoon [Perspectives on sign and lexicon] (Soveltavan Kielentutkimuksen Teoriaa ja Käytäntöä 5), 57–78. Jyväskylä: Jyväskylän yliopisto.

Jantunen, Tommi. 2013. Ellipsis in Finnish Sign Language. Nordic Journal of Linguistics 36. 303–332. DOI:  http://doi.org/10.1017/S0332586513000292

Jantunen, Tommi. 2017a. Constructed action, the clause and the nature of syntax in Finnish Sign Language. Open Linguistics 3. 65–85. DOI:  http://doi.org/10.1515/opli-2017-0004

Jantunen, Tommi. 2017b. Fixed and NOT free: Revisiting the order of the main clausal constituents in Finnish Sign Language from a corpus perspective. SKY Journal of Linguistics 30. 137–149. Available: http://www.linguistics.fi/julkaisut/SKY2017/SKYJol30_jantunen.pdf.

Janzen, Terry, Barbara O’Dea & Barbara Shaffer. 2001. The construal of events: Passives in American Sign Language. Sign Language Studies 1(3). 281–310. DOI:  http://doi.org/10.1353/sls.2001.0009

Johnston, Trevor. 2013. Formational and functional characteristics of pointing signs in a corpus of Auslan (Australian sign language): Are the data sufficient to posit a grammatical class of ‘pronouns’ in Auslan? Corpus Linguistics and Linguistic Theory, 109–159. DOI:  http://doi.org/10.1515/cllt-2013-0012

Johnston, Trevor & Adam Schembri. 2007. Australian sign language (Auslan): An introduction to sign language linguistics. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511607479

Kegl, Judy. 1990. Predicate 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.

Laitinen, Lea. 2006. Zero person in Finnish: A grammatical resource for construing human reference. In Marja-Liisa Helasvuo & Lyle Campbell (eds.), Grammar from the human perspective: Case, space and person in Finnish, 209–231. Amsterdam; Philadelphia: John Benjamins Publishing Company. DOI:  http://doi.org/10.1075/cilt.277.15lai

Langacker, Ronald W. 2006. Dimensions of defocusing. In Masayoshi Shibatani, Tasaku Tsunoda & Tarō Kageyama. Voice and grammatical relations: In honor of Masayoshi Shibatani (Typological Studies in Language 65), 115–137. Amsterdam; Philadelphia: John Benjamins Publishing Company. DOI:  http://doi.org/10.1075/tsl.65.08lan

Leeson, Lorraine. 2001. Aspects of verbal valency in Irish Sign Language. Dublin: Trinity College Dublin dissertation. Available: http://hdl.handle.net/2262/79586.

Liddell, Scott. 1995. Real, surrogate, and token space: Grammatical consequences in ASL. In Karen Emmorey & Judy Reilly (eds.), Language, gesture and space, 19–41. Hillsdale, NJ: Lawrence Erlbaum Associates.

Liddell, Scott. 2003. Grammar, gesture, and meaning in American Sign Language. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511615054

Mathur, Gaurav & Christian Rathmann. 2012. Verb agreement. In Roland Pfau, Markus Steinbach & Bencie Woll (eds.), Sign language: An international handbook, 136–157. Berlin: Mouton de Gruyter. DOI:  http://doi.org/10.1515/9783110261325.136

Næss, Åshild. 2007. Prototypical transitivity. Philadelphia: John Benjamins Publishing Company. DOI:  http://doi.org/10.1075/tsl.72

Neidle, Carol, Judy Kegl, Dawn MacLaughlin, Ben Bahan & Robert Lee. 2000. The syntax of American Sign Language. Cambridge, MA: MIT Press.

Nilsson, Anna-Lena. 2004. Form and discourse function of the pointing toward the chest in Swedish Sign Language. Sign Language & Linguistics 7(1). 3–30. DOI:  http://doi.org/10.1075/sll.7.1.03nil

Rankin, Miako. 2013. Form, meaning, and focus in American Sign Language. Washington, D.C.: Gallaudet University Press.

Rissanen, Terhi. 1985. Viittomakielen perusrakenne [The basic structure of sign language] (Helsingin Yliopiston Yleisen Kielitieteen Laitoksen Julkaisuja 12). Helsinki: University of Helsinki.

Rissanen, Terhi. 1998. The categories of nominals and verbals and their morphology in Finnish Sign Language. Turku: University of Turku licentiate thesis.

Schembri, Adam, Kearsy Cormier & Jordan Fenlon. 2018. Indicating verbs as typologically unique constructions: Reconsidering verb ‘agreement’ in sign languages. Glossa: A Journal of General Linguistics 3(1). 89. 1–40. DOI:  http://doi.org/10.5334/gjgl.468

Shibatani, Masayoshi. 1985. Passives and related constructions: A prototype analysis. Language 61. 821–848. DOI:  http://doi.org/10.2307/414491

Siewierska, Anna. 2004. Person. Cambridge: Cambridge University Press. DOI:  http://doi.org/10.1017/CBO9780511812729

Siewierska, Anna. 2008. Introduction: Impersonalization from a subject-centred vs. agent?centred perspective. Transactions of the Philological Society 106(2). 115–137. DOI:  http://doi.org/10.1111/j.1467-968X.2008.00211.x

Siewierska, Anna. 2011. Overlap and complementarity in reference impersonals. Man-constructions vs. third person plural-impersonals in the languages of Europe. In Andrej Malchukov & Anna Siewierska (eds.), Impersonal constructions. A cross-linguistic perspective (Studies in Language Companion Series 124), 57–90. Amsterdam: John Benjamins Publishing Company. DOI:  http://doi.org/10.1075/slcs.124.03sie

Siewierska, Anna. 2013. Passive constructions. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online. Leipzig: Max Planck Institute for Evolutionary Anthropology. Available: http://wals.info/chapter/107.

Siewierska, Anna & Maria Papastathi. 2011. Towards a typology of third person plural impersonals. Linguistics 49(3). 575–610. DOI:  http://doi.org/10.1515/ling.2011.018

Wulf, Alyssa, Paul Dudis, Robert Bailey & Ceil Lucas. 2002. Variable subject presence in ASL narratives. Sign Language Studies 3(1). 54–76. DOI:  http://doi.org/10.1353/sls.2002.0027