Despite the considerable amount of literature that has been devoted to the notion of contrast, there is still controversy about how contrast should be defined. In the IS literature the notion of contrast has been mainly addressed from two different perspectives: syntactic and semantic. In syntactic analysis contrast is seen as a feature that can appear together with the topic or the focus of a sentence (Chafe 1976, Kiss 1998), giving rise to different types of focus and topic, like contrastive focus and contrastive topic. Since contrast is a feature that co-occurs with two different notions (focus and topic) there are at least two different types of focus: information focus (IF) and contrastive focus (CF), which are distinguished by their syntactic and semantic properties. From this view, CF occurs in a different syntactic position than IF and is interpreted as exhaustive. Some languages like Hungarian provide evidence for this claim: in (1a) CF occurs in preverbal position, whereas in (1b) IF is placed in final position (Kiss 1998).

From this approach additive focus particles like ‘also’ or ‘even’ cannot combine with a CF, since a CF is interpreted exhaustively.

However, this approach is problematic for languages where the marking of contrast is ambiguous, like English or German (Molnár 2002). Also, there are contrastive constructions that are not exhaustive, like parallel constructions. In (3) below there is a non-exhaustive parallel contrast between ‘linguist’ and ‘interpreter’, so the focus particle ‘also’ is completely acceptable.

From a semantic perspective contrast is understood as an independent category that is triggered by the existence of alternatives (cf. Rooth 1992, Vallduví & Vilkuna 1998, Molnár 2002, Krifka 2007, Repp 2016, Umbach 2004). The presence of contrast indicates that there are other relevant alternatives that can be added to the common ground (CG). In some of these works, the definition of contrast is the same as the definition of focus in Alternative Semantics (cf. Rooth 1992), where contrast and focus are unified into one single notion: focus. This framework states that focus is always contrastive, because it always triggers alternatives. Krifka (2007) further distinguishes between two types of focus depending on the type of set of alternatives they belong to: open focus vs. closed focus. The former triggers a set of alternatives that is not identifiable in the context; the latter triggers a closed set of alternatives that is delimited and easily identifiable in the context. This distinction is important in analysing LSC data since there is a specific marking for closed focus, which I identify as contrast. From this view, contrast/focus can also be present in topics whenever there are salient contrasted alternatives that are relevant in the CG (cf. Büring 2003).

The unifying approach of focus/contrast also poses some problems, though. Many languages, like Hungarian (see (1) above) and Finnish (4), have a specific marking for contrast (4b) which differs from the one for focus (4a), and thus they do not fit in a unifying approach.

In light of this, some works highlight the importance of separating new information from contrast (Vallduví & Vilkuna 1998, Vallduví 2016, Molnár 2009). In these studies contrast is defined as an orthogonal independent notion that is not obligatorily bound to focus and can overlap also with the topic of the sentence.

The diversity of views presented here regarding the definition of contrast can be explained by the heterogeneous forms for expressing it in the different languages studied to date, which leads to different conceptions of this notion. While in some languages contrast is marked by grammatical means, in other languages contrast does not have an explicit marking. Therefore, it is not an easy task to come up with a unified theory that accounts for all languages studied until now (Repp, 2016). In Section 2.2 some theories developed from a semantic perspective are presented that take into account this diversity and analyse contrast as a complex notion that conforms different types at different linguistic levels (Molnár 2002, Umbach 2004, Repp 2016).

SLs seem to contribute to the view of contrast as an independent notion, since they share some common features for the expression of contrast: most of the SLs studied to date show specific marking that only appears when contrast is present, and it is clearly distinguished from the marking of other IS notions, such as topic and focus (§ 2.3). In LSC contrast also shows clear independent prosodic marking that can be distinguished from the marking of focus and/or topic (§ 4) and correlates with different interpretations associated to different types of contrast (§ 5). However, it does not show a specific syntactic position for CF associated to exhaustivity. For this reason, I adopt a semantic-pragmatic perspective that allows us to better explain the expression of different types of contrast in LSC and the correlation between prosodic markers and pragmatic interpretations.

As mentioned in the previous section, some theories have developed more fine-grained analyses that present contrast as an independent notion that can be found in information and discourse structure and that involves different types (cf. Molnár 2002, Umbach 2004, Repp 2016).

Through crosslinguistic comparison, Molnár (2002) shows that contrast is an independent category superimposed on topic and focus. Contrast can be found at the pragmatic level, where it is found in a closed set of alternatives; and at the semantic level, where it involves an open set of alternatives. This work proposes the following labels for different contrast types:

At the pragmatic level, Molnár distinguishes CONTRAST and LD-CONTRAST (left dislocation contrast). CONTRAST is expressed in many languages moving a focus to the left periphery when there is explicit mentioning of the alternatives or the alternatives are salient in discourse. LD-CONTRAST is also moved to the left periphery and needs alternatives to belong to a closed set. It is used to mark a shift of attention from one topic to another.

At the semantic level, two types of contrast are connected to open sets of alternatives: FOCUS OPERATOR, which involves exhaustivity since it excludes all possible alternatives, and I-CONTRAST (intonation contrast), which does not involve exhaustivity, since it only excludes some of the alternatives. I-CONTRAST is expressed by intonation and is traditionally linked to topics; however, Molnár provides evidence from English that it can be superimposed on both topic and focus. Also, the type of contrast found in Finnish (see example (4) above), is always expressed by the syntactic position and can be superimposed on topics and focus too.

Umbach (2004) claims that the notion of contrast cannot be considered a unique or single notion, since there are different types of contrast both at the information structure (IS) and at the discourse structure (DS) level, which trigger different interpretations. At the IS level, contrast is divided into i) contrast due to similarity plus dissimilarity, which is a prerequisite for alternatives and thus for contrastiveness, and ii) contrast due to exclusion, which triggers exhaustivity. The former is exemplified in sentence (5) below. The words beer and port are contrasted referents that are similar, since they have a common integrator –both are drinks–, and are dissimilar, since both have semantic independence –both have different independent meanings.

As for contrast due to exclusion, it can be further divided into two varieties with semantic differences. The first variety is exemplified in (6) below, where the contrastive focus Ronald excludes the possibility that some other item instead of Ronald makes the proposition true. In example (7), instead, the only-phrase only Ronald excludes the possibility that some other item in addition to the focused one makes the proposition true.

Both varieties of contrast due to exclusion are exemplified only via the pragmatic context of a correction in Umbach’s work; however, as will be shown in § 5 this type of context could also be found in the context of a selection between two alternatives (selective contrast) in LSC, since the same ‘exclusive’ meaning could be derived.

At the DS level, Umbach analyses the discourse relations CONTRAST (8) and CORRECTION (9) through structures with but.

Both discourse relations behave similarly to contrast at the IS level and seem to be intertwined with contrast due to exclusion: both require the similarity plus dissimilarity condition and both indicate the exclusion of one of the alternatives; however, CONTRAST excludes the possibility that the second alternative is true in addition to the first one (like only-phrases), and CORRECTION excludes the possibility that the first alternative is true instead of the second one (like CF).

Similarly to Umbach, Repp (2016) analyses contrast taking into account two dimensions: i) the alternativeness of constituents and ii) the discourse relations between two segments. The former may involve different conditions for contrastiveness to hold: i) explicit alternative (10), ii) explicit alternative set (11), iii) implicit alternative set (12).

Regarding discourse relations contrast may involve different degrees of contrastiveness that may be categorized into different discourse relations: i) Smooth discourses (Q-A and SIMILAR), ii) OPPOSE, and iii) CORR. The discourse relation Q-A consists of a question and a congruent answer. The discourse relation SIMILAR is analogous to parallel constructions: there are two or more propositions that can be true in the world of evaluation, which make the same contribution to the current question under discussion (QUD).1 In the discourse relation OPPOSE both propositions can be true in the world of evaluation, but they make opposing contributions to the current QUD. Lastly, in the discourse relation CORR one proposition rejects the other either i) because both propositions cannot be true in the world of evaluation or ii) because the proposition corrected does not meet certain background assumptions needed for it to be considered felicitous. In Repp’s categorization the discourse relations Q-A and SIMILAR are not considered contrastive vis-à-vis OPPOSE and CORR, where CORR involves a higher degree of contrastiveness than OPPOSE. According to this proposal contrast may be triggered either i) by the alternativeness of constituents, ii) by the discourse relations or iii) by both.

In many studies, the so-called parallel contrast (SIMILAR in Repp 2016) is not considered a contrastive discourse relation, unlike other types of contrast such as the one found in a correction. One of the basic arguments for this claim is that there must be some kind of opposition between the alternatives, which is not found in parallel constructions like (10) above, in which both segments make the same contribution to the QUD (Repp 2016). However, at the level of alternatives they do involve contrast, since the constituents are explicitly contrasting with each other.

Regarding their marking, contrast types are often marked by (morpho)syntactic and/or prosodic strategies. Repp (2016) argues that different degrees of contrast should correlate with the application of additional or different grammatical means. The marking of the different types of contrast and their combinations depends on the specificities of the grammar of each language.

In general, IS notions tend to occupy positions of prosodic prominence in the sentence; however, there is no one-to-one correlation between prosodic markers and IS notions. Prosodic marking is considered just a tendency, since these notions can also be left unmarked or they can be marked by different means (Féry 2008). For instance, in Catalan and Spanish narrow IF2 there is a high correlation between focus in situ and rising intonational patterns, while falling intonational patterns seem to be associated to dislocation of the non-focal material and fronting (Vanrell & Soriano 2013). As for CF, an increase on rising intonational patterns correlates with an increase in the use of fronting for marking the CF constituent. Vanrell & Soriano (2013) propose that for broad focus declarative sentences in Catalan and Spanish there is a tendency for the intonational prominence to be located in clause-final position. However, in line with Féry (2008), they claim that these correlates of prosodic patterns with IS notions are just tendencies, since in IF declaratives, the intonational prominence can also be found in clause initial position in the sentence (in Eastern Catalan and Basque Spanish), and also in situ for both IF and CF declaratives.

Navarrete-González (2019) show three types of contrast in LSC that display a common prosodic marking that only differs in the addition of some nonmanual markers (NMMs) on types involving other interpretations, like exclusion or exhaustivity. This marking correlates with explicit or salient alternatives in discourse that share the similarity plus dissimilarity condition. Additional prosodic markers are found to express additional meaning components like exclusion, exhaustivity or violation of expectations that combine with the marking of similarity plus dissimilarity to express other discourse relations like, for instance, a correction.

However, the following questions are still open: are these common NMMs marking contrast at the level of alternatives or at the DS level? Are additional markers used to express different degrees of contrast? Moreover, do these prosodic markers appear just as tendencies that can be overridden by other factors? Sections 4, 5 and 6 intend to shed some light on these questions, taking as a basis the categorizations sketched above.

Studies conducted until now in SLs have generally assumed that contrast is not an independent category in IS but a feature of focus and topic (cf. Wilbur 1997, 1999, 2012; Neidle 2002; Van der Kooij et al. 2006; Nunes & Quadros 2008; Sze 2008, 2011, 2015; Crasborn & Van der Kooij 2013; Herrmann 2013, 2015; Kimmelman 2014; Kimmelman & Pfau 2016).

IS notions in SLs may be marked by syntactic or prosodic means (Kimmelman & Pfau 2016). Regarding syntactic strategies, different constructions have been attested to be used in order to express contrast. Wilbur (1997) claims that in American Sign Language (ASL) CF may be marked by question-answer pairs,3 which consist in a wh-question followed by an answer, as illustrated in (13).

Kimmelman (2014) points out that in Sign Language of the Netherlands (NGT) and Russian Sign Language (RSL), corrective focus can be marked by topicalization, moving the focused constituent to the left of the sentence (14), (15).

As for the prosodic strategies, SLs express intonation primarily through non-manual markers; namely, facial expressions, and body and head movements (Sandler 2012, Pfau & Quer 2010). Manual prosodic strategies such as the duration of the sign, the velocity of the sign’s movement, or the length of the sign’s movement trajectory, have also been attested as the way a sign is stressed in many SLs (Kimmelman & Pfau 2016).

Almost all SLs studied to date have shown that body leans are the most common strategy employed by SLs for the expression of contrast between two referents. In ASL, forward and backward body leans are consistently used for expressing contrastive relations (Wilbur & Patschke 1998, Schlenker et al. 2016). In NGT left and right body leans are used in order to express contrast (Van der Kooij et al. 2006, Kimmelman 2014). Forward movements are also found in some examples of NGT in combination with left and right body leans (16).

In ASL and NGT, besides expressing contrast, forward leans are associated with inclusion and affirmation, and backward leans with exclusion and negation at the lexical level and beyond (Figure 1) (Wilbur & Patschke 1998; Van der Kooij et al. 2006).

In NGT and RSL body leans are usually accompanied by the use of the opposite sides of signing space for placing contrasted referents, as illustrated in example (17) from RSL.

It is important to notice that the marking of contrast described above is spreading over both topics and foci in many examples, like (17) above. Kimmelman (2014) points out that body leans are used to mark contrast and not focus, since they are found co-occurring with both IS notions. This is a common pattern in some of the SLs studied until now. However, Kimmelman does not consider this evidence sufficient to argue for the notion of contrast as an independent category, since these markers in NGT and RSL are only used in a binary way, when only two alternatives are contrasted, and when they are explicitly mentioned. He argues that these markers may arise due to modality specificities: the use of space in SLs.

Crasborn & Van der Kooij (2013) also found a specific use of space, and manual and non-manual marking for CF in NGT. They argue though that the prosodic distinction between IF and CF seems to be gradient, based on the hypothesis that in larger stretches of discourse, body leans can also be used for IF. However, they acknowledge that the possibility that the co-occurrence of several cues may be a distinguishing factor between IF and CF is beyond the scope of their research.

Table 1 shows that body leans are a common strategy across SLs for marking contrastive relations. Head movements also appear in most of them as a way to express CF, mostly related with affirmation.

In sum, there seem to be evidence that SLs mark contrast independently from other IS notions. SLs can be argued to provide evidence that we are actually dealing with a different IS notion expressed by specific grammatical means, which overlaps with the topic and the focus of the sentence. The traditional approach adopted until now in most SL research has successfully served the purpose of providing the first descriptions of the expression of IS notions in SLs. However, most of them have consistently assumed traditional definitions for IS notions without questioning more theoretical issues like the ones posed in § 2.1 and § 2.2. This paper analyses in greater detail semantic-pragmatic interpretations of different types of contrast and their grammatical reflexes in order to contribute to a better understanding of the nature of this notion in both SLs and spoken languages. I adopt a semantic-pragmatic approach, since it allows us to broaden the paradigm of contrast markers in LSC. The definitions of IS notions in this perspective explain better the phenomena found in our data, allowing us to provide a more thorough explanation of the semantic features found in each type expressed through additional markers (head movements).

30 sentences were elicited using visual stimuli from the Questionnaire for Information Structure (QUIS) (Skopeteas et al. 2006). 10 sentences for each type of contrast: parallel, selective and corrective. Signers were shown a picture with two referents that were contrasted in some sense (Figure 2), and then they were asked a question in LSC related to that picture. Questions were asked by the author, who is a trained sign language interpreter fluent in LSC. To avoid biasing the answers the use of signing space was avoided in the questions.

Questions were designed to trigger different types of answers: i) some questions were seeking for plain information (‘What are the woman and the man doing?’), ii) some questions were forcing the signer to select between two alternatives that were included in the question (‘What is the woman riding: a horse or a bike?’), and iii) some questions contained false information that the signer needed to confirm or refute (‘The man is riding a bike, right?’). These different types of questions were thought to elicit minimal pairs between parallel, selective, and corrective contrast respectively.4 Given material in relation to the previous question was considered topic, while new material in relation to the previous question was considered focus. In the case of selective contrast, the answers contained referentially given material that was relationally new; this was considered focused material (based on Gundel & Freitheim 2006). Items were randomly presented, and results were consistent across both informants.

The goal of these tasks was to obtain more naturalistic data in the context of a monologue or a dialogue. Two storyboards from Totemfield Storyboards (Littell 2010), created specifically for the elicitation of CF and other IS notions, were used for the elicitation of this data. Signers had to create a dialogue between them based on the storyboards presented. 16 tokens that involved parallel contrast were analysed.

Moreover, two monologues were recorded in which signers had to retell two mute short films (Extinguished5 and Alike6). Both films were specifically selected for this research since the stories involved two referents that were consistently contrasted across the narration. Therefore, even though it was a freer discourse, they produced CTs and CFs as expected. 33 examples were obtained from both monologues: 26 instances of parallel focus, and 7 examples that contained some kind of opposition between the alternatives.

Signers were asked to translate a sentence that was presented in written Spanish.7 The procedure was the following: signers were provided with a context in LSC and afterwards they were asked to translate a sentence in relation to that context from Spanish to LSC. 215 tokens were obtained and analysed from this task: 120 involving parallel contrast (18), 34 involving selective (19) and 61 involving corrective contrast (20). The number of parallel contrast tokens is higher, since I also analysed structures involving additive focus particles, in addition to the basic parallel contrast with ‘and’.

All data collected was rated by the informants weeks later to ensure consistency in the results. Each piece of data was presented again with a signed context, and participants had to rate the felicity of the sentences in relation to that context in a binary way: felicitous vs. infelicitous. No descriptive statistics were conducted.

Finally, 20 sentences were collected from the Catalan Sign Language corpus (under development in the Institut d’Estudis Catalans, Barberà et al. 2015). Tokens were produced by three deaf native signers who narrated the story Frog, Where Are You? Same marking for contrast was observed in this data for parallel contrast across topics and focus that validate the results obtained through elicitation tasks. No selective or corrective focus was found in these narrations.

Focus in LSC can be expressed syntactically, through a change in the basic word order (SOV) as in (21) (Quer et al. 2005) to VO, placing the focused item in final position in the sentence (22), or prosodically through eyebrow raise when the focus remains in situ (Navarrete-González 2016, 2019). If contrastive, focus can also be fronted and marked with eyebrow raise (see § 4.2).8

Topic in LSC is typically placed in sentence initial position and marked with eyebrow raise and an intonational break after the topicalized constituent (Quer et al., 2005, Pfau & Quer 2010), as illustrated in (23) below.9

In line with previous research on contrast in SLs (cf. Crasborn & van der Kooij 2013, Kimmelman 2014), contrast in LSC is mainly expressed through a combination of prosodic and morphophonological markers: left-right body leans (bl) and head tilts (ht), and the use of the opposite sides of signing space (cf. Barberà 2012, Navarrete-González 2016, 2019, Zorzi 2018). These markers may spread over topic and focus constituents, as illustrated in (24), and are not found in IF, or simple aboutness topics, as shown in (21), (22) and (23) above.

In the example above, the topic and the focus of the contrasted sentences maintain their own markers, which overlap with the contrast markers described before. The topics GIORGIA and RAQUEL are marked with eyebrow raise, and the foci LINGUIST and INTERPRETER occur in clause-final position. On top of these markers, a leftward body lean spreads over the conjunct GIORGIA LINGUIST, which is articulated on the left side of signing space, and a rightward body lean spreads over the conjunct RAQUEL INTERPRETER, which is articulated on the right side of signing space.

Until now we have seen that contrast in LSC has been claimed to be expressed with this combination of markers, which overlap with other IS notions: topic and focus. However, is it always the case that these markers express a contrastive interpretation? Is there a one-to-one correlation between this combination of markers and the interpretation of contrast?

Mayol & Barberà (2018) analysed contrast in LSC in three different contexts: double contrast, implicit contrast, and weak contrast, based on corpus observation. They found that double contrast is primarily conveyed by bare nouns, implicit contrast is expressed through pointing signs addressed to space, and weak contrast conveys uncertainty through specific NMMs articulated in the lower part of the face. This study does not analyse body leans or head tilts but highlights that the use of signing space evokes a contrastive set of alternatives that are not explicit in the sentence.

In the next subsections I provide a paradigm of different types of contrast identified in LSC, and the different strategies that signers use to express this notion. I compare the markers of each of the types, and analyse the semantic interpretations triggered by each type and their correlation with some markers. Finally, I propose a classification of different types of restrictions that can affect (and change) the realization of these markers, giving rise to other marking strategies for the expression of contrast.

In LSC at least three different types of contrast can be identified: i) parallel contrast, ii) selective contrast, and iii) corrective contrast, each of which involves a different degree of contrastiveness (cf. Navarrete-González 2019).

NMMs in LSC are not only used for the expression of intonation. They can also have other grammatical functions, such as morphological and syntactic functions (Pfau & Quer 2010). Our hypothesis is that, since the meaning added through a morphological expression cannot be erased, when there is a clash between morphological markers and prosodic markers, the latter ones may get overridden.

Some lexical specifications of the signs of a sign language may be determined by the semantics of a verb or a noun. For instance, lexically specified body movements are attested in signs that involve a semantics of inclusion or exclusion. In different SLs, forward body leans have been associated with inclusion and affirmation, while backward leans have been associated with exclusion and negation (cf. Wilbur & Patschke 1998, van der Kooij et al. 2006, Pfau & Quer 2010). In ASL, for instance, the signs AVOID and reject involve a backward body lean while the signs INVOLVE and ENCOURAGE involve a forward body lean (Wilbur & Patschke 1998).

I took these claims as a baseline to test if LSC follows the same patterns, and the lexical level specifications affect or change the markers of contrast that work at a suprasegmental level. In example (43) below, the sign PARTICIPATE involves inclusion and the opposite STOP-PARTICIPATING involves exclusion.

We would expect that these specific lexical semantic features of inclusion and exclusion might influence the realization of the contrasted signs, which we would expect to be uttered with forward and backward body leans instead of left and right body leans (or at least with a combination of both). Surprisingly, the signs PARTICIPATE and STOP-PARTICIPATING are only uttered with left and right body leans, and the inclusive-exclusive meaning that we would find specified as a lexical feature is not found. We could arrive at the conclusion that this premise does not hold for LSC; however, the set of data used in this part is not very extensive, thus a more exhaustive study of verbs involving these features is needed in order to confirm the results and broaden the picture on this issue.

In sum, out of the three different contexts that were tested, only phonological and morphological restrictions seem to affect or modify the combination of markers found for the expression of contrast in LSC.