2.1 discourse grammar

Discourse Grammar, as proposed by Kaltenböck et al. (Reference Kaltenböck, Heine and Kuteva2011) and Heine et al. (Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Carmen2013), is composed of all the linguistic resources that are available for constructing spoken or written (or signed) texts; an outline of its architecture is provided in Figure 1.^{Footnote 1} It is based on the assumption that there are two domains of discourse organization that need to be distinguished, referred to respectively as sentence grammar (SG) and thetical grammar (TG). Discourse Grammar thus differs from, and must not be confused with the model of Functional Discourse Grammar (FDG; Hengeveld & Mackenzie, Reference Mackenzie and Brady2008, Reference Hengeveld, Mackenzie, Heine and Narrog2011), which does not make such a distinction.^{Footnote 2}

Fig. 1.

A sketch of the architecture of Discourse Grammar.

The relationship between the two domains of Discourse Grammar in Figure 1 is complex; it is shaped most of all by cooptation, a mechanism whereby a chunk of SG, such as a clause, a phrase, a word, or any other unit, is deployed for use in TG (Kaltenböck et al., Reference Kaltenböck, Heine and Kuteva2011, pp. 874−875).

SG is well documented, having been the main or the only subject of theories of mainstream linguistics. It is based on propositional logic, and it is organized in terms of parts of speech or constituent types such as sentences, clauses, phrases, words, and morphemes plus the syntactic and morphological machinery to relate constituents to one another. The building blocks of TG are theticals, consisting on the one hand of thetical formulae and constructions and on the other hand of the ability to coopt information units of SG and deploy them for structuring discourse.^{Footnote 3} The main categories of theticals distinguished so far are illustrated in (1).^{Footnote 4}

Theticals differ from SG units in a principled way, their defining properties being listed in (2).^{Footnote 5} Note that this definition is prototypical rather than being based on necessary and sufficient criteria (see also Kaltenböck et al., Reference Kaltenböck, Heine and Kuteva2011, Section 2).

The term ‘non-restrictive’ meaning is taken from Huddleston and Pullum (2002, p. 1352). Restrictive meaning is a characteristic of SG; it is grounded in the semantic structure of a sentence or its constituents. Non-restrictive meaning, by contrast, concerns reasoning processes and inferential mechanisms grounded in the situation of discourse. To be sure, any act of linguistic communication requires a situation of discourse, but in the case of SG its impact is minimal, being restricted to a few factors such as spatial, temporal, and personal deixis, deontic modality, etc. Such restrictions do not appear to exist when TG is involved. The situation of discourse consists of a network of interlocking components, namely the ones listed in (3).

In accordance with this distinction, units of SG differ from those of TG (i.e., theticals) in their semantic−pragmatic scope potential: whereas the former have scope over some constituent of the utterance, theticals may have wider scope, typically extending over the entire situation of discourse (Heine et al., Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Carmen2013, Section 2.2; Kaltenböck et al., Reference Kaltenböck, Heine and Kuteva2011, p. 861). We may illustrate this with the English item frankly. It is an adverb of SG, determining the meaning of the predicate in (4a). In (4b), by contrast, it is a thetical, called a stance adverbial by Biber, Johansson, Leech, Conrad, and Finegan (1999, p. 133), a sentence adverb by Brinton and Traugott (2005, p. 139), or a disjunct by Quirk, Greenbaum, Leech, and Svartvik (1985, pp. 648, 613). As a thetical, it is syntactically detached, typically set off prosodically (marked by commas in writing), and its meaning is non-restrictive: rather than determining the meaning of the predicate, it relates to and may have scope over the situation of discourse, most of all to Speaker−Hearer Interaction.

It may not be surprising that the meaning of theticals has been described with reference to notions such as ‘subjectification’ (cf. the component Attitudes of the Speaker in (3)) or ‘intersubjectification’ (cf. Speaker−Hearer Interaction) by a number of authors (e.g., Traugott & Dasher, Reference Traugott and Dasher2002; Brinton, Reference Brinton2008), or as being ‘metalingual’ (Maschler, Reference Maschler1994), ‘procedural’, ‘metacommunicative’, ‘metatextual’, ‘metapragmatic’, ‘metadiscursive’, or ‘instructional’ (see Heine, Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Carmen2013).

2.2. the dual process model of van lancker sidtis (2009)

A detailed account of this model is found in various publications of Van Lancker Sidtis and associates (Sidtis, Canterucci, & Katsnelson, Reference Sidtis, Canterucci and Katsnelson2009; Van Lancker, Reference Van Lancker and Ellis1988, Reference Van Lancker1990, Reference Van Lancker1997; Van Lancker Sidtis, 2004, 2009, 2012; Van Lancker Sidtis & Postman, Reference Van Lancker Sidtis and Postman2006). Central to this model is the distinction between novel speech (or novel language, or newly created language, or propositional speech) and formulaic speech (or formulaic expressions or automatic speech). As these authors argue on the basis of substantial neurological and linguistic evidence, novel speech is represented in the left hemisphere, whereas formulaic speech is facilitated by a subcortical right hemisphere circuit.^{Footnote 6}

The dual process model features an analytic mode for the generation of novel speech and a holistic mode for processing formulaic speech, the distinction being one between newly created or propositional and fixed or non-propositional language, respectively. Unlike the former, the latter is not newly created from the operation of grammatical rules on lexical items (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, p. 445). Relying on this distinction, Van Lancker Sidtis summarizes a range of neurolinguistic findings thus:

According to Van Lancker Sidtis (2004, 2009, 2012, and other works), formulaic speech has the characteristics listed in (5) (see also Section 3 below).

Easily identified instances of formulaic speech are swear words, interjections, pause fillers (uh, um), discourse elements (well, so), non-literal lexical meanings for idioms (He was at the end of his tether), and proverbs. The functional criteria proposed pertain mostly to the repertory of speech formulas, such as Hello, Right, If you say so, How could you?, Here’s back atcha, and many others signaling “turn-taking, commentary, and assent, conveying countless attitudinal stances in conversational interaction” (Kreiman & Sidtis, Reference Kreiman and Sidtis2011; Van Lancker Sidtis, Reference Van Lancker Sidtis2012, p. 66).

In examining written transcripts of the spontaneous speech of patients who had suffered left or right hemisphere damage, Van Lancker Sidtis (2009, p. 452) distinguishes nine types of formulaic speech; we will deal with them in more detail in Section 3.

That brain lateralization shows positive correlations with two different modes of linguistic processing is an old hypothesis (see Van Lancker Sidtis, Reference Van Lancker Sidtis2004, 2009, p. 460, for detailed discussion). Already in the nineteenth century, the neurologist Hughlings Jackson (1874) provided examples of preserved aphasic speech, distinguishing between ‘propositional’ and ‘non-propositional’ (or ‘automatic’) speech associated with left and right hemisphere processing, respectively. Similar observations were made by subsequent authors. The neurologist Head (Reference Head1926) found that non-propositional speech appears first in both receptive and expressive aphasia, Bay (Reference Bay, De Reuck and O’Connor1964) described aphasia as an inability to propositionalize, and according to the Russian neuropsychologist Luria (Reference Luria1966), clinical observations showed that it was speech formulas, expletives, pause fillers, proper nouns, sentence stems, and serial speech that were preserved in aphasic speech.

But the most substantial evidence was presented in the work of Van Lancker Sidtis (2004, 2009, 2012, and other works), based on the distinction between novel and formulaic speech. Being extremely rare following right hemisphere damage, aphasia is almost exclusively associated with left hemisphere damage. Van Lancker Sidtis carried out a number of case studies to substantiate the hypothesis that brain lateralization shows positive correlations with two different modes of linguistic processing. In one kind of study she worked with three patients, where she made the following observations (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, pp. 451−452):

1. − Case 1 involved a patient who had sustained a large right hemisphere lesion. Although language abilities were intact, his conversational speech was often pragmatically inappropriate.

2. − The same observation was made in Case 2, who had suffered right-sided subcortical damage. This patient complained that she no longer produced the “little words” in conversational interaction, having difficulties with greeting and leave-taking.

3. − Case 3 was a left hemisphere damaged patient with the diagnosis of transcortical sensory aphasia, who spoke fluently but with numerous formulaic expressions.

A separate study confirmed that the rate of formulaic expressions was low in Case 1 (11%) and Case 2 (16.9%), but high in Case 3 (51.9%), compared to 20.1% in the normal control group (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, p. 454).

Examining written transcripts of the spontaneous speech of patients who had suffered left or right hemisphere damage, Van Lancker Sidtis and Postman (Reference Van Lancker Sidtis and Postman2006) found that persons with left hemisphere damage (LHD) use significantly more (30%) and persons with right hemisphere damage (RHD) significantly less formulaic utterances (17%) than normal subjects (25%). This finding compellingly implicates a role of the right hemisphere in the production of formulaic expressions (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, p. 452).

Further support for the dual process model was found in two speech samples of aphasic patients (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, pp. 454−457). In the first, taken from an aphasic subject who recovered some speech over a period of five treatments from early non-fluency, 64 percent of the text turned out to consist of formulaic expressions.

In the second speech sample, taken from a German aphasic patient before treatment, nearly all the speech products consisted of formulaic language. Of the thirty-nine information units uttered by the patient in conversation with a therapist, thirty, that is, 76.9 percent, were formulaic units (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, pp. 456−457); we will return to these two cases in Section 3.

The reciprocal effect of lateralization on comprehension was demonstrated by using the Formulaic and Novel Language Comprehension Test of Kempler and Van Lancker (Reference Kempler and Van Lancker1996). The result was a ‘double dissociation’ to the effect that left hemisphere damaged subjects performed poorly on literal expressions but relatively better on idiomatic and formulaic language, while right hemisphere damaged patients performed relatively worse on formulaic and idiomatic language than on novel expressions (Van Lancker Sidtis, Reference Van Lancker Sidtis2004, p. 26).

Support for the dual process model comes from some recent lines of linguistic research, in particular from Pawley’s (Reference Pawley, Corrigan, Moravcsik, Ouali and Wheatley2009) work on speech act formulas:

4.1. text organization

SG is determined by the syntactic and semantic compositionality of sentences. TG, by contrast, draws on inferential mechanisms that relate information units beyond their literal meaning to the speaker, the hearer, and to the situation in which speech operates − in short, TG is anchored in what is commonly described as pragmatics. For example, we saw in example (4a) that the meaning of the adverb frankly is determined by its function as an adverb modifying the predicate of the sentence. As a thetical in (4b), by contrast, where it is syntactically and prosodically detached, it relates the meaning of the utterance beyond the sentence to the attitudes and beliefs of the interlocutors. And if our hypothesis of a positive correlation between the use of TG and right hemisphere activity is correct, we will expect speakers suffering right hemisphere damage to have deficits in locating their speech appropriately within the situation of discourse and, more generally, in the world around them. TG is responsible in particular for the following functional goals:

1. (a) To design a coherent model of discourse.

2. (b) To anchor meanings in the situation of discourse rather than in the structure of sentences.

There is in fact neurolinguistic evidence in support of these two functional goals: both appear to be centrally associated with right hemisphere activity, rather than with the left hemisphere.

With reference to (a), there is a body of neurolinguistic observations suggesting that the left hemisphere is in charge of basic information (word recognition, syntactic processing). The right hemisphere, by contrast, tends to be activated to establish cohesive ties in narratives (Bloom, Reference Bloom, Bloom, Obler, De Santi and Ehrlich1994; Marini, Carlomagno, Caltagirone, & Nocentini, Reference Marini, Carlomagno, Caltagirone and Nocentini2005) and/or when the processing of higher-level information (integration of parts as a coherent whole), and what tends to be referred to as the ‘macrostructure’ of discourse, are involved (Robertson et al., Reference Robertson, Gernsbacher and Guidotti2000; see also Sherratt & Bryan, Reference Sherratt and Bryan2012, pp. 215−216), for instance the organizing and ordering of discourse structure (Lojek-Osiejuk, Reference Lojek-Osiejuk1996).

Confronting ten healthy, native English speaking volunteers with written texts which consisted on the one hand of titled and on the other hand of untitled paragraphs, St George, Kutas, Martinez, and Sereno (1999, pp. 1317, 1323) conclude that right hemisphere engagement occurs routinely as readers attempt to construct a unitary coherent model of a discourse and discover the producer’s intents. And it is especially the right middle temporal regions that appear to be important for the integrative processes needed to achieve global coherence during discourse processing, where ‘integration’ means that multiple pieces of information are integrated across sentences.

Right hemisphere damage following a stroke has been shown to lead to disturbances in communication skills, and these disturbances include difficulty in preserving the macro-structure and organization of discourse (Hough, Reference Hough1990; Joanette, Goulet, Ska, & Nespoulous, Reference Joanette, Goulet, Ska and Nespoulous1989). Accordingly, the content of discourse produced by RHD individuals tends to be characterized by reduced topic maintenance (Prutting & Kirchner, Reference Prutting and Kirchner1987) and to be incoherent, tangential, and self-oriented (Blake, Reference Blake2006).^{Footnote 13} And a number of authors argue that cognitive disorders in RHD individuals account for disturbed discourse skills (Bartels-Tobin & Hinckley, Reference Bartels-Tobin and Hinckley2005; Myers, Reference Myers1999; Penn, Reference Penn2000; Tompkins, Reference Tompkins1995).

That the integration task is of a different kind both between the two domains of Discourse Grammar and the two hemispheres can possibly be linked to the hypothesis proposed by Beeman (Reference Beeman, Beeman and Chiarello1998), according to which words are each associated with a large and diffuse semantic field in the right hemisphere but with a smaller, more focal, semantic field in the left hemisphere. Blonder et al. (1991, p. 1124) therefore suggest that activation in the left hemisphere is restricted to the target and its most closely linked associates, whereas in the right hemisphere many concepts give rise to weak activation for some time.

With regard to (b), the contribution of the right hemisphere is particularly evident in the domain of what has been referred to above as pragmatics (e.g., Bates, Reference Bates1976; Cutica, Bucciarelli, & Bara, Reference Cutica, Bucciarelli and Bara2006; Ferré, Ska, Lajoie, Bleau, & Joanette, Reference Ferré, Ska, Lajoie, Bleau and Joanette2011): left hemisphere patients typically exhibit primary impairment in comprehending and appropriately using syntactic and semantic aspects of language; persons with right hemisphere damage, by contrast, demonstrate great difficulty with pragmatic communication (Joanette, Goulet, & Hannequin, Reference Joanette, Goulet and Hannequin1990; Molloy, Brownell, & Gardner 1990; Moscovitch, Reference Moscovitch and Perecman1983; Ozonoff & Miller, Reference Ozonoff and Miller1996; Weylman, Brownell, Roman, & Gardner, Reference Weylman, Brownell, Roman and Gardner1989).

An impairment of RHD subjects has been referred to as a selective deficit in integrating pieces of information by means of inferences derived from the situational context (Carol, Baum, & Pell, Reference Carol, Baum and Pell2001; Delis, Wapner, Gardner, & Moses, Reference Delis, Wapner, Gardner and Moses1983; Jung-Beeman, Bowden, & Gernsbacher, Reference Jung-Beeman, Bowden and Gernsbacher2000;), and Marini et al. (2005, p. 53) speculate that the right hemisphere plays a relevant role in complex linguistic skills such as organizing a mental model for producing narratives.

Furthermore, patients with right hemisphere damage were found to have difficulty in interpreting indirect requests and commands and to rely on the literal meanings of conversations rather than pragmatic cues that involve deriving meaning from contextual information (Foldi, Reference Foldi1987; Hirst, LeDoux, & Stein, Reference Hirst, LeDoux and Stein1984; Weylman et al., Reference Weylman, Brownell, Roman and Gardner1989).

We observed in Section 3 that Van Lancker Sidtis (2009, pp. 451−452) found that one of the patients who had sustained a large right hemisphere lesion was characterized by conversational speech that was often pragmatically inappropriate, even though his language abilities were intact. Van Lancker and Cummings (1999, p. 96) observe that while the left hemisphere mediates most linguistic behaviors, the right hemisphere is important for broader aspects of communication. Other right hemisphere lesion patients were found to ignore context and were not able to fill in what was not present in the words (Myers, Reference Myers and Brookshire1978); Shields (Reference Shields1991) concludes:

The relationship between speech and pragmatics tends to be described in terms of inferential mechanisms and, in fact, some authors have pointed out that RHD individuals may lack the ‘mental flexibility’ for making inferences (Brownell, Potter, & Bihrle, Reference Brownell, Potter and Bihrle1986; but see also McDonald & Wales, Reference McDonald and Wales1986) or accessing indirect speech acts (Champagne-Lavau & Joanette, Reference Champagne-Lavau and Joanette2009). Note further that the right hemisphere has been argued to be relatively more involved in computing (non-linguistic) situation models that, in speaking, provide the input to specifying the propositional content of an utterance (see Menenti, Segaert, & Hagoort, Reference Menenti, Segaert and Hagoort2012, for discussion).

4.2. speaker−hearer interaction

Language structure provides a range of means to express interpersonal functions. But, as pointed out in the introduction to this section, the paradigm tools of expression are reserved for the domain of TG. One of the prominent functions of TG is to express interpersonal concepts: three of the five categories of this domain of grammar concern exclusively interpersonal communication, namely formulae of social exchange, vocatives, and imperatives (see Table 5). It would seem that theticals belonging to these three categories serve in particular the following functional goals:

1. (a) To establish and maintain contact with other speech participants.

2. (b) To create a social environment that is beneficial to all speech participants concerned.

3. (c) To address the hearer and ask him or her for action.

According to the ‘classical’ view surfacing from neurolinguistic analyses, the left temporal cortex is dominant in speech processing. The right cerebral hemisphere, by contrast, is more centrally associated with other functions. And it is in particular the above range of functions that appear to be more strongly associated with right hemisphere activity.

In accordance with (a), inappropriate social (and emotional) behavior is predominately associated with right frontal dysfunction (Joseph, Reference Joseph2000). People with right brain damage (RBD) are considered to be socially disconnected from the world around them (Myers, Reference Myers1999), and they have been found to have difficulties sharing the responsibility to develop and maintain adequately the exchange with the speaker (Hird & Kirsner, Reference Hird and Kirsner2003). In particular, RHD adults exhibit difficulties in governing verbal exchange since they take little account of their communicative partner. And they have problems with paralinguistic means of speaker−hearer interaction such as sending or receiving information via facial expression (Blonder et al., Reference Blonder, Bowers and Heilman1991), establishing and maintaining eye contact (Myers, Reference Myers and Chapey1994; Tompkins, Reference Tompkins1995), or spontaneously using gesture (Tompkins, Reference Tompkins1995). Note that many patients with acquired right hemisphere damage demonstrate paralinguistic deficits, including impairments in prosody and gesture (Joanette et al., Reference Joanette, Goulet and Hannequin1990).

Furthermore, persons with right hemisphere damage following stroke have been found to exhibit an impaired turn-taking and appreciation of the listener’s perspective (Chantraine, Joanette, & Ska, Reference Chantraine, Joanette and Ska1998; Kaplan, Brownell, Jacobs, & Gardner, Reference Molloy, Brownell, Gardner, Joanette and Brownell1990; Myers, Reference Myers and Chapey1994).

That the goal in (b) is distinctly more likely to involve activity in the right than in the left hemisphere has been pointed out by a number of researchers. The former hemisphere has been portrayed as providing the social context of linguistic communication (Berman, Mandelkern, Phan, & Zaidel, Reference Berman, Mandelkern, Phan and Zaidel2003) and serving successful social communication (Mitchell & Crow, Reference Mitchell and Crow2005). And a number of studies suggest a predilection for right hemisphere processing of social and real-world contextual associations for lexical items (Chiarello, Reference Delis, Wapner, Gardner and Moses1995; Drews, Reference Drews1987; see also Van Lancker, Reference Van Lancker1997; Van Lancker Sidtis, Reference Van Lancker Sidtis2004). As has been pointed out in some research findings, between 50 percent and 78 percent of individuals with right hemisphere damage may exhibit difficulties in one or more communication components, leading to inadequate social interactions (Ferré et al., Reference Ferré, Ska, Lajoie, Bleau and Joanette2011).

Interpersonal, as well as emotional, difficulties are among the main deficits experienced by patients who have suffered damage of the right hemisphere early in life or by inheritance (Shields, Reference Shields1991). It therefore comes as no surprise that for the two aphasic patients that we were concerned with in Section 3, formulae of social exchange were among the most frequently used speech units, topped only by interjections (Tables 1 and 3). And as we also saw in Section 3, one of the patients, having suffered right-sided subcortical damage, showed deficits in her command of information units concerning Speaker−Hearer Interaction, e.g., having difficulties with formulae of social exchange, such as greetings and leave-taking (Van Lancker Sidtis, Reference Van Lancker Sidtis, Corrigan, Moravcsik, Ouali and Wheatley2009, pp. 451−452).

4.3. attitudes of the speaker

The primary function of SG appears to be the structuring and expression of conceptual information in a propositional format. It relates primarily to what Jakobson (Reference Jakobson and Sebeok1960) calls the referential function, or Lyons (1977, pp. 50−51) the descriptive (or propositional, or ideational) function of language. But SG does not really dispose of dedicated tools, i.e., function-specific linguistic constructions, for the expression of emotions.

This is different in TG, which disposes of appropriate means for expressing speaker attitudes in general and emotional states in particular. In the sense of Jakobson (Reference Jakobson and Sebeok1960), there are dedicated categories in TG for the expression of conative, expressive, and phatic functions. Especially the thetical category of interjections (which also includes exclamatives; see Heine et al., Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Carmen2013, Section 4.6) provides cross-linguistically an ideal tool dedicated to the linguistic encoding of emotions (see Table 5).

That emotional behavior is strongly linked to the right hemisphere is an old observation in neurological research. Loss of propositional speech was reported in severely aphasic speech of patients already in the nineteenth century: these patients were found to be left only with expletives, interjections, and oaths (Van Lancker & Cummings, Reference Van Lancker and Cummings1999) − that is, with linguistic expressions that are all classified as interjections in the framework of Discourse Grammar (Heine et al., Reference Heine, Kaltenböck, Kuteva, Long, Bischoff and Carmen2013).

As has been established in a number of lesion studies, right brain damage usually results in deficits in both the linguistic and the non-linguistic comprehension and production of emotions (e.g., Borod, Andelman, Obler, Tweedy, & Welkowitz, Reference Borod, Andelman, Obler, Tweedy and Welkowitz1992; Borod et al., Reference Borod, Rorie, Haywood, Andelman, Obler and Welkowitz1996; Borod, Bloom, & Santschi Haywood, 1998; Borod et al., Reference Borod, Rorie, Pick, Bloom, Andelman and Campbell2000; Borod, Bloom, Brickman, Nakhutina, & Curko, Reference Borod, Bloom, Brickman, Nakhutina and Curko2002; Karow & Connors, Reference Karow and Connors2003; Myers, Reference Myers1999; Sherratt & Bryan, Reference Sherratt and Bryan2012; Wager, Phan, Liberzon, & Taylor, Reference Wager, Phan, Liberzon and Taylor2003). This generalization has received some support from brain-imaging studies, even though the latter have also shown that the expression of emotion also implicates left lateralization (Wager et al., Reference Wager, Phan, Liberzon and Taylor2003, p. 527).^{Footnote 14} Among the pragmatic aspects of language associated with the right cerebral hemisphere, emotions in speech have in fact a prominent place (e.g., Friederici & Alter Reference Friederici and Alter2004; Mitchell & Crow, Reference Mitchell and Crow2005). This hemisphere is said to be dominant in the processing of paralinguistic information, and to one’s affective state (Beeman & Chiarello, Reference Beeman and Chiarello1998; Devinsky, Reference Devinsky2000). Acknowledging that language and speech are typically related to the left hemisphere of the brain, Jakobson (1980, p. 23) adds that there are such verbal elements as interjections and exclamations that are typically associated with the right hemisphere (see also Tsur, Reference Tsur2010, p. 512). Kriendler and Fradis (1968, p. 111) observe that in all kinds of aphasia, motor articulation was dramatically better during ‘emotional speech’, and Blonder et al. (1991, p. 1116) conclude that the right hemisphere “houses a lexical representation of emotions”.

Furthermore, as we saw in the texts analyzed in Section 3, produced by two aphasic patients, interjections were clearly the linguistic units most frequently produced by both the English-speaking and the German-speaking subjects suffering left hemisphere damage (Tables 1 and 3).

A number of neurolinguistic studies suggest in fact that inappropriate emotional behavior is predominantly associated with right frontal dysfunction (e.g., Joseph, Reference Joseph2000), and Shammi and Stuss (Reference Shammi and Stuss1999) observe that individuals with RHD injury, especially when reaching the frontal cortex, do not react physically to emotions (laughing or smiling).

Both imaging and clinical evidence suggest in fact that the right hemisphere is highly relevant for the comprehension and production of emotional features in speech (Bloom, Borod, Obler, & Gerstman, Reference Bloom, Borod, Obler and Gerstman1992; Borod et al., Reference Borod, Bloom, Santschi Haywood, Beeman and Chiarello1998; Devinsky, Reference Devinsky2000; Rota, Reference Rota, Dogil and Reiterer2009). Ley (Reference Ley1980) and Ley and Bryden (Reference Ley, Bryden and Perecman1983) found that the presentation of emotional words during a list learning task selectively improved memory of stimuli directed to the right hemisphere.

Such observations are supported by studies on hemispheric differentiation in the activation of prosody. While prosodic processing requires a series of complex cognitive operations, there is evidence for a specialization of the right hemisphere for the processing of emotional prosody and of the left hemisphere for ‘linguistic’ prosody^{Footnote 15} (Ferré et al., Reference Ferré, Ska, Lajoie, Bleau and Joanette2011). Ross et al. (Reference Ross, Thompson and Yenkisky1997) analyzed the mechanisms underlying affective–prosodic deficits following left and right brain damage by testing the ability of subjects to repeat and comprehend affective prosody under progressively reduced verbal–articulatory conditions. They conclude that reducing verbal–articulatory conditions robustly improves the performance of left but not right brain damaged patients, thus supporting the supposition that affective prosody is strongly lateralized to the right hemisphere.

4.4. discussion

As the observations made in this section suggest, there are positive correlations between the functions served by the categories of TG and right hemisphere activity. However, there is so far still a wide gap between what the linguist expects to find located somewhere in the brain and what the neurologist actually observes. Accordingly, the correlations pointed out in the preceding paragraphs must be taken with care, in particular for the following reasons. First, exactly which functions are activated where in the brain is an area that is still largely ill-understood, and this applies crucially to linguistic functions. For example, language dominance in the left hemisphere is not an absolute human characteristic. As a study of 188 right-handed subjects by Knecht et al. (Reference Knecht, Deppe, Dräger, Bobe, Lohmann, Ringelstein and Henningsen2000) showed, using a functional imaging technique, altogether 92.5 percent of the subjects turned out to have left hemisphere language dominance, while 7.5 percent had right hemisphere language dominance.

Second, there appears to be widespread consensus that most, if not all, language components include both left and right hemisphere processes – in the wording of Beeman and Chiarello (Reference Beeman and Chiarello1998, p. 6): “the right hemisphere and left hemisphere conjointly process language at all levels” (e.g., Hagoort, Brown, & Swaab, Reference Hagoort, Brown and Swaab1996; Jung-Beeman, Reference Jung-Beeman2005, p. 513).

And third, the two hemispheres appear to have mutually supportive functions. For example, when a child has one hemisphere injured or removed at an early age, the remaining hemisphere may, with minor adjustments, be able to compensate for the injured or absent hemisphere’s function (Beeman & Chiarello, Reference Beeman, Beeman and Chiarello1998, p. 2). But there appears to be an asymmetry in lateralization − one that might be directly relevant to the subject matter of this paper. The right hemisphere exhibits a remarkable capacity to reorganize originally left hemisphere functions while there does not appear to be good evidence for the reverse pattern of reorganization, e.g., after right hemisphere lesions.^{Footnote 16} Helmstaedter, Kurthen, Linke, and Elger (1994, p. 735) suggest that even if evidence would be found for a left hemisphere reorganization of right hemisphere functions, this would not be at the expense of originally left hemisphere functions.

Furthermore, the correlations pointed out earlier in Sections 3 and 4 raise a number of additional problems. One has already been discussed by Luria (Reference Luria1974), namely that linguistic performances are distributed in widespread cortical constellations or assemblies: neither neural distinctions of brain activity nor the linguistic structures distinguished in this work are as neatly separated from one another as was implied above. There are complex overlaps between the two − both between the two hemispheres and between the two domains of Discourse Grammar. And there is constant and massive interaction between the two structures, and interaction takes place both between subareas of the two hemispheres of the brain and of the two linguistic domains. The nature of this interaction is the subject of ongoing research.

These observations raise the question of whether the generalizations proposed by us can in fact be accounted for in terms of functional modularity where specific areas of the cortex can be held responsible for particular language-specific functions, or whether such functions should not more profitably be analyzed in terms of network-based accounts. According to some lines of research, there are no significant correlations between particular brain regions and speech processing. When taking a set of putative ‘speech areas’ of the brain and looking at the non-linguistic processes that activate them, Price, Thierry, and Griffiths (2006) found no macro-anatomical structures in the human brain dedicated to speech. Rather than correlations between brain structures and speech, these authors suggest, speech-specific processing emerges at the level of functional connectivity among distributed brain regions, each of which participates in processes that are engaged in both speech and non-speech tasks (p. 271).

A final problem concerns the question of whether the correlations between brain lateralization and language-related functions are in fact symmetrical in the way implied in this paper. Neuroimaging evidence on speech comprehension suggests, for example, that functional dissociation may not concern the distinction between right and left hemisphere but rather between a distributed bilateral domain relating to general perceptual and cognitive processing on the one hand, and a more specialized left hemisphere domain supporting key grammatical language functions on the other (Bozic, Tyler, Ives, Randall, & Marslen-Wilson, Reference Bozic, Tyler, Ives, Randall and Marslen-Wilson2010).^{Footnote 17}

In spite of these and other caveats that have been voiced on hypotheses that assume a clear-cut division of language-related functions in brain lateralization, the evidence that has become available appears to allow for the following generalization on the neurological anchoring of Discourse Grammar: in the same way that SG, that is, ‘regular’ sentence structure, is unlikely to be activated without the participation of the left hemisphere, it is equally unlikely that the activation of TG phenomena, such as discourse markers, vocatives, interjections, or formulae of social exchange, can be achieved without any participation of the right hemisphere.