(Inter)subjectification and its limits in secondary grammaticalization

Language Sciences 47 (2015) 148–160

Contents lists available at ScienceDirect

Language Sciences journal homepage: www.elsevier.com/locate/langsci

(Inter)subjectification and its limits in secondary grammaticalization Heiko Narrog* GSICS, Tohoku University, Kawauchi 41, Aoba-ku, Sendai 980-8576, Japan

a r t i c l e i n f o

a b s t r a c t

Article history: Available online 18 August 2014

It is widely accepted that subjectification and intersubjectification are important processes of semantic change accompanying grammaticalization. However, typical changes of subjectification concern early stages of grammaticalization, and the role of subjectification and intersubjectification in late stages of grammaticalization is not yet fully explored. In this paper, I am looking for (1) regular changes other than (inter)subjectification in secondary grammaticalizations, and (2) counter-examples to the hypothesized directionalities of change. In doing so, I mainly look at the development of core case marking, at the development of elements with textual functions, and at presumptive counter-examples from other areas of grammar that have been proposed in the literature. I conclude that (1) the evolution of textual/discourse-functions is a significant development not captured in the canonical (inter-)subjectification scenario both in terms of concept and in terms of directionality, and (2) that de-subjectification and de-intersubjectification in terms of expressive (inter)subjectivity regularly takes place at later stages of grammaticalization. However, this de-(inter)subjectification mainly accompanies the development of highly paradigmatic and abstract meanings, and concerns a loss of subjectivity in terms of expressiveness. On the other hand, since meanings are increasingly appropriated for the expression of speaker-deixis and internal reasoning, the overall directionality of change can still be understood as speaker-, hearer- and text-orientation, and it might not be appropriate to label the overall development as ‘de-subjectification’ or ‘objectification’. Ó 2014 Elsevier Ltd. All rights reserved.

Keywords: (Inter-)subjectification Grammaticalization Speech-act orientation Textual functions De-(inter)subjectification Case functions

1. Introduction Subjectification and intersubjectification have been firmly established in the literature as semantic changes central to grammaticalization. On the other hand, their relationship to different stages of grammaticalization is not yet fully explored. It has been suggested that subjectification is widely involved in grammaticalization (Traugott, 1995; Visconti, 2013), and hypothesized that intersubjectification presupposes subjectification and follows it (Traugott, 2003), leading to an association of intersubjectification with a later stage of grammaticalization. There is a rich literature based on the observation of empirical data, which substantiates these claims, and it is only fair to assume that they have a certain validity. However, there may also be developments not conforming to this scenario. Recently, it has been suggested that a third tendency of ‘textual orientation’ is Abbreviations: 3, 3rd person; ACC, accusative; EPI, epistemic modality; F, feminine; GEN, genitive; INF, infinitive; LOC, locative; M, masculine; NEG, negation; NOM, nominative; Part, participle; PERF, perfect; PFV, perfective; PRG, progressive; PP, prepositional phrase; PST, past tense; QUE, interrogative; S, singular; TAG, tag; TOP, topic. * Tel.: þ81 22 221 5016. E-mail address: [email protected]. http://dx.doi.org/10.1016/j.langsci.2014.07.010 0388-0001/Ó 2014 Elsevier Ltd. All rights reserved.

H. Narrog / Language Sciences 47 (2015) 148–160

149

necessary to complement speaker-orientation (subjectification) and hearer-orientation (intersubjectification). It has been claimed that these three changes together combine for a tendency of ‘speech-act orientation’, and the order between these three changes is not settled (Narrog, 2012a,b,c). Also, it has been suggested that subjectification is more typical of primary than of secondary grammaticalization (Traugott, 2010), or even that secondary grammaticalization regularly involves desubjectification (Kranich, 2010b). In this paper, I will mainly explore semantic changes in late stages of grammaticalization that in my view potentially deviate from the expected “subjectification > intersubjectification” or “continuous subjectification” scenario. In this context, I will also critically discuss the concept of de-subjectification in secondary grammaticalization. 2. Key concepts and hypotheses 2.1. Grammaticalization, primary and secondary For a long time, the following quote by Kuryłowicz’s (1975(1965)) was cited as a standard definition of grammaticalization (e.g. Lehmann, 1986: 3; Heine et al., 1991: 3). “Grammaticalization consists in the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status, e.g. from a derivative formant to an inflectional one.” (Kuryłowicz, 1975: 52) The first part of “advancing from a lexical to a grammatical [status]” can be taken as a ‘primary’ form of grammaticalization, and the part of “advancing [.] from a less grammatical to a more grammatical status” as a ‘secondary’ form of grammaticalization. In citations, the phrase “e.g. from a derivative formant to an inflectional one” is sometimes omitted. This phrase reflects a traditional focus on morphology that is often not shared in modern studies. However, the basic idea has persisted. Givón (1991: 305) introduced the term ‘secondary grammaticalization’, not with a definition but by providing examples, such as past tense morphemes arising as the reanalysis of either the perfect or perfective aspects or the future arising as a reanalysis of irrealis. Furthermore, Hopper and Traugott (2003(2): 18) defined grammaticalization as a “change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions”, and Heine and Narrog (2010: 401), defined it as a “development from lexical to grammatical forms, and from grammatical to even more grammatical forms” (highlighting by author). Nevertheless, it is fair to say that change towards “more” or “new” grammatical forms has been explored less in terms of grammaticalization than the change from the lexical to the grammatical, and the differences between ‘primary’ and ‘secondary’ grammaticalization are not well-known yet. 2.2. Subjectification, intersubjectification and increase in speech-act orientation 2.2.1. The concepts The concept of subjectification has been developed and brought to the main stream of historic oriented functional linguistics chiefly by Traugott, who worked on it from the 1980s (e.g. Traugott, 1980, 1982, 1995, 2003, 2010). The major competing concept, which differs from Traugott’s both in its mainly synchronic perspective, and in its very conceptualization of ‘subjectivity’ and ‘subjectification’, has been proposed by Langacker (1990, 1998). In this paper, we are concerned with the diachronic concept and possible variations on it. The current “standard” definition of ‘subjectification’ goes back to Traugott (2003), and runs as follows: (1)

“subjectification is the mechanism whereby meanings come over time to encode or externalize the SP/W’s perspectives and attitudes as constrained by the communicative world of the speech event, rather than socalled ‘real-world’ characteristics of the event or situation referred to.” (Traugott, 2003: 126)

This definition, and the concept behind it is the result of a number of developments in Traugott’s theory that we will come to later in this paper where they are relevant. It was also Traugott who developed a companion concept of ‘intersubjectification’ in the early 2000s, which was defined as in (2): (2)

“intersubjectification is the semasiological process whereby meanings come over time to encode or externalise implicatures regarding SP/W’s attention to the ‘self’ of AD/R in both an epistemic and a social sense.” (Traugott, 2003: 129–130)

The hypothesized diachronic relationship between subjectification and intersubjectification was stated clearly on several occasions: (3)

(4)

“The hypothesis is that, for any lexeme L, intersubjectification is historically later than and arises out of subjectification” (Traugott, 2003: 130). “There cannot be intersubjectification without some degree of subjectification” (Traugott, 2003: 134) “In my view [. ] intersubjectification [is] the mechanism by which meanings [. ] once subjectified may be recruited to encode meanings centered on the addressee” (Traugott, 2010: 35)

150

H. Narrog / Language Sciences 47 (2015) 148–160

While a number of concepts of ‘subjectivity’ and ‘intersubjectivity’ have been proposed (cf. Narrog, 2012c for a synopsis), to my knowledge Traugott’s are the only concepts of ‘subjectification’ and ‘intersubjectification’ with a diachronic import and broader circulation. However, recently I have proposed an alternative that aims to be more comprehensive by generalizing over subjectification, intersubjectification and related changes, and partially incorporating other concepts of subjectivity and intersubjectivity in their diachronic dimension Narrog (2012a,b,c). I have labeled this alternative concept ‘speech-act orientation’, and its diachronic dimension ‘increase in speech act orientation’. An increase in speech act orientation may consist of one of the following three tendencies: (i) increasing orientation towards the speaker’s perspective (increased speaker orientation; corresponding to subjectification) (ii) increasing orientation towards the speech situation including the hearer (increased hearer orientation; corresponding to intersubjectification) (iii) increasing orientation towards discourse itself (increased discourse/textual orientation) These three tendencies together mark an increase in ‘speech act orientation’, which is an increase in orientation towards the participants in the speech act, and its setting. Semantic change is hypothesized to proceed always towards an increase, or at least, no decrease, in speech act orientation. Speaker-orientation and hearer-orientation in the sense used here are characterized mainly by ‘performativity’ (Narrog, 2012a: 40–46), as opposed to ‘evidentiality’ (Nuyts, 2001), or construal (Langacker, 1990, 1998). Discourse/textual orientation mainly refers to the development of means to organize chunks of discourse into larger discourse (text), and the formation of coherent discourse. In contrast to Traugott’s theory of subjectification and intersubjectification, none of the three tendencies is taken to entail another one. It is assumed that the actual sequence of changes still needs to be empirically established. I believe that our knowledge of historical changes is still limited, and I propose that several avenues of change between increased speaker orientation, increased hearer orientation, and increased discourse orientation are possible. This is in fact something that I would like to demonstrate in the main Section 3 of this paper. 2.2.2. Status assigned to (inter)subjectification Scholars who actively engage in the study of (inter)subjectification and related concepts have generally expressed confidence that this is an important, if not the most important type of semantic change in grammaticalization. Thus, Traugott and Dasher (2002: 279) asserted that “the main mechanism of semantic change is subjectification (including intersubjectification). This follows from the hypothesis that the seeds of semantic change are to be found in SP/Ws, drawing on and exploiting pragmatic meanings that arise in negotiated interaction.” Likewise, Narrog (2012a: 104–5) hypothesized that “semantic change always proceeds in the direction of the same or a higher degree of speech act orientation”. Indeed, there is a long trail of case studies across a large range of typologically diverse languages, and theoretical literature that underscores the validity of the concepts. Besides Traugott and Dasher (2002), and Narrog (2012a), which also contain a number of case studies, there are for example Cornillie and Delbeque (eds) (2006), Onodera and Suzuki (2007), Davidse et al. (eds) (2010), Devos and van der Wal (2012) and Van der Auwera and Nuyts (2012) with collections of studies, there are Visconti (2004) on the subjectification of conditionals in Germanic and Romance languages, Fitzmaurice (2004a) on subjective discourse markers eventually becoming intersubjective, Aaron and Torres Cacoullos (2005) quantifying subjectification on a Spanish counter-expectation marker, Breban (2006, 2008) on subjectification of adjectives in English NPs, López-Couso (2010) with a theoretical piece on the concepts, and Rhee (2012) of (inter)subjectification of sentence-final particles in Korean. This list could be made even much longer without too much effort. However, most of the literature focuses on actual cases of subjectification and intersubjectification (or increased speech act orientation). Studies so far suggest that (inter-)subjectification and increased speech act orientation are very frequent, and may indeed be the dominant type of semantic change. Arguments against (inter-)subjectification as an important or dominant type of change, or for reverse tendencies of changes in individual cases, have been comparatively sparse so far. They include Adamson (1995), who made a case for the desubjectivization of emphatic deictic elements in literary prose, Kranich (2010a,b), who suggested that secondary grammaticalization of the English progressive and other tense–aspect related categories involves de-subjectification, and Abraham (2005), who presents a fundamental rejection of the concept as such. However, since there are no quantitative studies systematically analyzing semantic changes in one or more than one languages, not much is really known yet as to how dominant (inter-)subjectification and increased speech act orientation really are, or conversely, where exactly its limits lie. Narrog (2012a: 107) suggested that the actual sequence of changes may vary by domain of grammar and should first be established for each domain of grammar, and Kranich (this volume) suggests that (inter-)subjectification may be more typical for some domains of grammar than for others. These suggestions will also be reflected in Section 3, the main part of this paper. 2.2.3. The correlation between grammaticalization and (inter)subjectification (Inter-)subjectification and increased speech act orientation have mostly been studied in the context of grammaticalization. However, it is important to note that, at least to my knowledge, nobody has claimed that (inter)subjectification and grammaticalization entail each other. With respect to (inter)subjectification potentially entailing grammaticalization, this is

H. Narrog / Language Sciences 47 (2015) 148–160

151

an almost trivial fact. (Inter)subjectification can also occur in the domain of the lexicon (cf. Traugott, 1989), and is not bound to the domain of grammar. So it is only natural to assume, as Traugott does, that “[n]either subjectification or intersubjectification entails grammaticalization” (Traugott, 2010: 38). The reverse case is less trivial. The claim that grammaticalization is always accompanied by some form of subjectification or intersubjectification would seem not unreasonable, but I am not aware that such a claim has been actually advanced by anyone. As Visconti (2013: 13) writes, “subjectification is pervasive, as it is motivated by the very subjectivity of the speech event.” Indeed, Traugott herself suggested that “[a] related topic that deserves further study is whether subjectification is characteristic of all domains of grammaticalisation. To date it appears that it may be” (1995: 46). She also offered an explanation for the apparent pervasiveness of the process: “The reason for the ubiquity of subjectification presumably lies in the speaker’s attempts to communicate the relevance of what is said to the communicative event, which includes hearers as well as speakers, but which ultimately depends for its occurrence on the speaker.” However, eventually Traugott became more cautious and later wrote that “not all grammaticalization is equally likely to involve equal degrees of subjectification, and some may involve no subjectification at all” (Traugott, 2010: 40). It does not seem that anyone else has claimed that grammaticalization is always accompanied (inter-)subjectification either. Overall then, the view offered here is that (inter-)subjectification does occur in grammaticalization and, as it is assumed to be the most common type of semantic change overall, it must be inferred that it occurs in a large number of cases of grammaticalization that involve semantic change. However, it does not occur necessarily. As for possible exceptions, Traugott (1995: 45) suggested that “some counter examples do exist, but they all involve later stages of grammaticalisation,” and later (2010: 40) that “Subjectification is more likely to occur in primary grammaticalization (the shift from lexical/constructional to grammatical) than in secondary grammaticalization (the development of already grammatical material into more grammatical material)”. Intersubjectification, in contrast, appears to be instead involved in secondary grammaticalization, if it is involved at all. This follows from the hypothesis that intersubjectification presupposes subjectification: “Intersubjectification intersects less extensively with grammaticalization. In most languages it is grammaticalized only into some discourse markers and interjections.” (Traugott, 2010: 41). Based on her study of the history of the English progressive, Kranich (2010b: 118) goes even much further in claiming that “secondary grammaticalization processes (or later stages of grammaticalization) commonly lead to a type of semantic change that can be labeled objectification.” This is seconded by Norde (2009: 88–9), who cites this as a case of change counterdirectional to the parameters of grammaticalization. The idea that subjectification is less present in secondary than in primary grammaticalization was thus also one of the leading ideas pertaining to (inter-)subjectification in the Workshop ‘Secondary Grammaticalization’ at the ICHL XXI in Oslo, which most of the papers in this volume originate from. In contrast, in my own approach to increase in speech act orientation, I have claimed that this is a unidirectional type of change: “semantic change always proceeds in the direction of the same or a higher degree of speech act orientation” (Narrog, 2012a: 104–5). This means that I am assuming that de-subjectification or ‘objectification’ occur only exceptionally and cannot be regular types of change. I have also assumed that an increase in speaker orientation is typical for early stages of grammaticalization, while advanced grammaticalization tends to involve an increase in hearer orientation or textual orientation (Narrog, 2012a: 106). I have especially highlighted textual/discourse-orientation as a last stage of development, and I intended to demonstrate that hearer-orientation (intersubjectification) clearly is not always the endpoint of development (Narrog, 2012b). However, there is no direct correlation between increase in speech-act orientation and grammaticalization (Narrog, 2012a: 109). I have further pointed out that the development of textual functions may be accompanied by a decrease in (inter-)subjective properties (p. 106). On the other hand, I have also emphasized that our knowledge about the sequence of changes is far from being known, and must be empirically established for each domain of grammar (Narrog, 2012a: 107, 2012b: 108–9). In this paper, I wish to further explore the actual directionality of changes. 3. The limits of (inter)subjectification in grammaticalization Logically speaking, there are two major possible limitations to the role of (inter-)subjectification, or an increase in speechact orientation in grammaticalization (a) no (inter-)subjectification takes place; there is no increase of speech act orientation. This could entail that (a1) no particular type of semantic change seems to be at work; (a2) other apparently regular types of change that cannot be subsumed under (inter-)subjectification or increased speech-act orientation are at work (b) the diachronic development runs counter to (inter-)subjectification or speech-act orientation. This could concern either (b1) the overall directionality, or (b2) the order of specific sub-directionalities, e.g. intersubjectification following subjectification The changes in (a) do not really constitute a counterexample to unidirectionality, and they also do not contradict claims about the relationship between (inter-)subjectification and grammaticalization, since apparently nobody has claimed that grammaticalization must always be accompanied by (inter-)subjectification, or an increase in speech-act orientation (cf. 2.2.3). However, if these types of change are regular and significant, they deserve investigation. Therefore I will devote the two subsections 3.1 and 3.2 to presumptive examples of such changes. The two other subsections of this section are devoted to (b1) presumptive de-subjectification, and (b2) presumptive counterexamples to claims about the ordering of changes.

152

H. Narrog / Language Sciences 47 (2015) 148–160

Fig. 1. Chain of increasing grammaticalization of case functions (Heine et al., 1991: 159).

As we will see below, the borderline between cases of (a2) tendencies other than (inter-)subjectification and (b1) counterdirectionality is not necessarily clear. In other words, some of the tendencies that do not count as (inter-)subjectification may also be interpreted as de-subjectification. 3.1. Other tendencies than (inter-)subjectification (1): development of abstract grammatical meaning The first case of recurring semantic change in, mainly secondary, grammaticalization, that is difficult to conceptualize as (inter-)subjectification broached here is the development of abstract grammatical meanings in the domain of case. The domain of case was long held to behave regularly with respect to grammaticalization and semantic change, or as even being a paradigm example. Heine et al. (1991: 159) famously proposed the “chain of increasing grammaticalization of case functions” that is displayed in Fig. 1. The basic assumptions behind this chain are that it is (a) semantic and (b) based on metaphor. The metaphorical change runs from more concrete to more abstract categories along the super-categories of location (1), human participants (2), participants in human activity (3) to more abstract categories of time and human reasoning (4), (5), (6). On the face of it, this is highly compatible with the concept of subjectification, as speakers gradually appropriate meanings pertaining to objectively observable phenomena in the external world for human actors, and finally internal reasoning, as meanings extend along this chain. However, our empirical knowledge about changes in meaning and function of case markers has considerably increased in the past 30 years, calling for a revision of this chain in the light of the currently available cross-linguistic data. This is what I did in Narrog (2014), where I tried to redraw the map based on a 200 languages-sample of semantic-role polysemy (Narrog, 2009, 2010), plus historical data meanwhile available in the research literature. The main finding was that the unidirectionality of the map cannot be maintained and that at least two directionalities have to be distinguished and represented separately, with (1) the development of marking of core participants, and (2) development of inanimate abstract functions, as endpoints. The resulting map is displayed in Fig. 2. As mentioned above, the development of human participant concepts out of spatial concepts, and the developments internal to human reasoning out of the human participant, or participant in human activity concepts are arguably cases of subjectification. However, the development of case functions of the core participants, namely various agents and patients are not obvious cases of subjectification. The most abstract core case markings at the end of the chain, namely, nominative, accusative, absolutive and ergative are often fairly semantically void and cover a wide range of meaning relationships. Crucially, the development of these core case functions from peripheral, more clearly semantically motivated case functions is often not the result of a semantic change of the case marker but rather of constructional change in which the semantics of the case marker plays no apparent role. These include for example the following changes: -

From From From From From From From

attributive possessor to ergative agent attributive possessor to nominative attributive possessor to absolutive and accusative possessor in a possessive predication to ergative agent possessor to passive agent instrument etc. as passive agent to ergative agent recipient to causee

The changes from possessor to agent are mainly due to the reanalysis of a nominalized clause construction to a main clause construction. For example, in a possessive construction of locative–existential origin with a non-canonically marked possessor subject, this subject possessor may be reanalyzed as a general subject if the construction is grammaticalized to express temporal–aspectual distinctions (particularly, perfect or perfective) (cf. Creissels, 2008: 23–25). According to Jung

1 Note that like in Heine et al.’s (1991) chain, a category on the left or bottom that extends to, or shifts to a category on the right or the top does not necessarily have to go through all intermediate steps. Furthermore, the fact that the agentive functions are placed to the right of the patient functions is a pure convenience solution for graphical representation in a two-dimensional space. It is not assumed that one of these two domains is more ‘advanced’ with respect to the horizontal directionality of change than the other.

H. Narrog / Language Sciences 47 (2015) 148–160

153

Fig. 2. Map of increasing grammaticalization of case functions.1

(2009), this seems to have happened in some Northern Russian dialects. Sentence (5) from the 16th century presumably represents an early example of this development. (5)

U carja pereloz-en-o na se l eto rat:-Ø at tsar:PPGEN undertake-PART-N.SG for this summer troops-NOM.F.SG svoj-a na moskovskuju ukrajnu posla-ti. own-NOM.F.SG to Moscow region send-INF ‘By the tsar it was undertaken to send his troops to the Moscow region for this summer.’ (Jung, 2009: 214)

While the possessive subject in the Russian construction is not a possessive in terms of case (i.e. a genitive) but a locative preposition, the same development may also be assumed for other languages where the possessive is expressed directly through a genitive. The change from instrument etc. to ergative agent is mainly due to the reanalysis of a passive as an ergative. The hypothesis that many ergatives are derived from passives is fairly old and well-established (e.g. Anderson, 1977; Trask, 1979). The development of the construction including the ergative agent in the target structure can be described as follows: “When [the passive] construction becomes more current and the Agent becomes increasingly obligatory, it is reinterpreted as a transitive ergative construction, the instrumental serving also as the case of the transitive subject” (Lehmann, 2002: 98). In this manner a whole construction, in which the instrumental already had agentive function, is reanalyzed, instead of a direct semantically-motivated extension from instrumental to ergative function. Arguably, in any kind of change, a linguistic item is not reanalyzed or extended in isolation but in a constructional context. However, in cases as these, there is no apparent semantic contribution by the case marker whose function changed as a result of the syntactic reanalysis. Now interestingly, in a minority of cases, the same sort of change may nevertheless occur when the semantics of the case marker is actually at stake. I am referring here to the following cases: - From instrument, dative etc. to passive agent - From instrument, dative etc. directly to ergative agent Passive constructions are in most cases originally agentless. The possibility to add agents is often only added eventually over time. At that point it is reasonable to assume that semantic fit does play a role in the choice of marker. The same is presumably the case with some ergative constructions which do not originally have an agent phrase. Passive and ergative agents are still more concrete and bound to specific semantic roles than nominatives or accusatives. However, it is not clear that a passive agent would be more ‘subjective’ than an instrument or a recipient (dative). Both tendencies, the one on the vertical axis towards core participants, and the development of adverbial functions on the horizontal axis have in common that they are developments towards abstract, semantically highly generalized functions. At least the change on the vertical axis is difficult to conceptualize as a subjectification, and is often the result of syntactic reanalysis rather than semantic extension. However, as we saw, in some cases it is semantic extension. It is also not particularly meaningful to speak of ‘de-subjectification’ or ‘objectification’, since it is not clear that an agent would be more objective than an instrument or a recipient, for example. In the end, both tendencies can be understood as the evolution of abstract core resources of grammar.

154

H. Narrog / Language Sciences 47 (2015) 148–160

3.2. Other tendencies than (inter-)subjectification (2): the evolution of textual meaning The second regularly occurring tendency of change that I consider as not covered by the concept of (inter-)subjectification, while it is part of the concept of ‘increase in speech act orientation’, is the development of textual meanings. Traugott, the main proponent of the concept of (inter-)subjectification from a diachronic perspective, did have it as part of her portfolio of changes at an earlier stage. Traugott was influenced by the linguistic thought of Halliday, for whom the ‘textual metafunction’ was of great importance as one of the three metafunctions of language (the other being the ‘ideational’ ¼ propositional and the ‘interpersonal’ ¼ expressive). In the first version of her theory of subjectification, she defined ‘textual elements’ as “resources available for creating a cohesive discourse” (Traugott, 1982: 248). Such textual elements included connectives, anaphoric & cataphoric pronouns, topicalizers, relativizers, complementizers etc. At this stage she proposed the following sequence of changes: (6)

propositional > (textual) > expressive (Traugott, 1980: 54, 1982: 256–57, 1989: 31)

‘Textual’ here is an optional transitional stage from the ‘propositional’ to the ‘expressive’, i.e. the ‘subjective’ in more recent terminology. In 1989, Traugott proposed the pair of terms “textual & metalinguistic”, and defined the latter as follows: “By ‘metalinguistic situation’ I mean the situation of performing a linguistic act. Examples include the shift from a mental-state to a speech-act verb meaning” (Traugott, 1989: 35). However, thereafter shift towards textual meanings was eliminated from the scenario of subjectification for the following two reasons (Traugott, 1995: 47): (i) There were “question[s] about the ordering of changes [between textual and expressive]”. In this context, an apparent counterexample by Herring (1991) (see below) was mentioned. (ii) “[T]he theoretical basis for the ordering no longer exists”, because the term ‘metalinguistic’ was used in too many divergent senses. It appears, though, that the shift towards textual function was partially absorbed into ‘subjectification’. This is most clearly stated in the description of the subjectification of in fact: “To the extent that speakers/writers choose textual connectives of various kinds to signal their perspective on the relationship between clauses, sentences, and discourses, they choose clause linkers such as causals and conditionals to signal coherence and create textuality, thus expressing their subjectivity” (Traugott, 2007: 298) Thus, the textual domain was removed primarily because of unclear ordering with respect to other changes, and change towards textual meaning was included in subjectification. However, it is questionable whether the textual domain is indeed identical with, or part of, the subjective/expressive domain. The logic of including it into the subjective is expressed in the phrase “signal coherence and create textuality, thus expressing their subjectivity”, but “signaling of coherence and creation of textuality” are not very obvious expressions of subjectivity. Signaling of coherence and creation of textuality may be an end in its own. This has led some scholars concerned with the development of textual meanings to call for a revival of ‘textual meaning’ as a tendency of change outside of subjectification. Ghesquiere (2010: 286, 309) reasoned that the earlier 1982 pathway “is semantically more fine-grained.and seems to capture better the semantic development of the adjectives of completeness”. Furthermore, “textual meanings can be both subjective and intersubjective”. Narrog (2012b,c) also suggested that textual functions cannot be simply subsumed under subjective or intersubjective (or objective) functions. (Inter)subjective functions cover textual functions only partially. Visconti (2013), aiming to keep the concept of subjectification more precise, argued for excluding change towards textual functions from ‘subjectification’ as a process distinct from the original idea of subjectification. The ‘concessive’ uses of may (cf. Palmer, 1990) are a case in point. Concessive may is found in so-called ‘universal concessive conditionals’, in although-concessives, in as-concessives with pre-posed predicates, and in while-concessives (cf. Narrog, 2012b: 35–6). It is based on the epistemic (subjectified) use of may and functions to construct logical coherence in complex sentences, and beyond that, discourse. (7) is an example. (7)

I looked at some of my portraits and grotesque as they may be, they capture some aspects of reality (Coates, 1983: 135)

The concessive use is arguably not only subjective (inherited from the epistemic use) but also intersubjective, as it takes into account an imaginary objection or criticism by the hearer. It would be insufficient to describe it as a case of subjectification (subsuming textual uses under subjective ones), since the epistemic use is already subjectified, and it is not clear at all that the concessive use is ‘more’ subjectified than the epistemic one. It has a textual/discourse function, and it is exactly this textual/ discourse function, which is more salient than the epistemic meaning in the concessive construction, and which distinguishes it from the epistemic use. A description of the change of may without a proper place for the development of such a textual/ discourse function is arguably incomplete. ‘Conditional’ uses of imperatives are similar in this respect. I am referring here to complex sentences as the one in (8) in which an imperative clause is connected by the coordinating conjunctions and or or to a declarative clause:

H. Narrog / Language Sciences 47 (2015) 148–160

(8)

155

Give him enough rope and he’ll hang himself (cf. Quirk et al., 1985: 832)

The imperative is originally an independent speech act but here it is subordinate to a following clause and gives a condition, that is, it serves as the protasis of a conditional. This is a cross-linguistically quite common construction with different degrees of grammaticalization in different languages. The imperative is a paradigm example for intersubjectivity in the sense of orientation towards the hearer but it has further developed a function of connecting pieces of discourse here. If I say the extension to conditional use is subjectification, in the sense of evaluating the logical relationship of propositions, I miss the most salient part of this change towards subordination, namely towards formation of complex/subordinate clauses. Similarly, not in English but in many other languages, one can also find imperatives used to construct concessive conditionals (cf. Narrog, 2012b: 42–4). In conclusion, secondary grammaticalization often entails an extension towards textual/discourse functions in secondary grammaticalizations, which cannot be omitted from any description, and simply subsuming them under the label of (inter-) subjectification, is also a very unsatisfactory description. This is thus a tendency outside the (inter-)subjectification scenario. However, it is included as a tendency in the model of increased speech-act orientation. 3.3. Counterdirectionality (1): de-subjectification/objectification Theories espousing (inter-)subjectification assume that this is a unidirectional process. “Unidirectional” in context of a functional view of language means that the overwhelming majority of changes are expected to conform with the hypothesized unidirectionality, and counterdirectional developments are expected to be exceptions. Traugott herself left the door open for such exceptions, as in the following quote: “Does objectification then occur? Indeed. In semantic change it is the highly conscious and deliberately interventive sort of change that comes about when ordinary words are preempted for technical or legal purposes, [.]. But these are cases of “external” “non-natural” change not of “internal” “natural” change, which is the central concern of historical cognitive semantics.” (Traugott, 1999: 190) According to this rationale, changes towards more subjectivity are natural, while counterdirectional change is the result of an artificial manipulation of language material. Notwithstanding, some scholars have argued for “naturally-occurring” counterexamples to subjectification. Two cases proposed in the literature concern question markers. In a study on the grammaticalization of historical questions in Tamil, Herring (1991) proposed a chain of grammaticalization from question markers to text cohesion markers and clause-linking markers, i.e. relativizers and adverbial conjunction. Example (9) shows the question marker - e indicating a rhetorical tag question in narration, and as a relative clause marker in (10). The latter is hypothesized to be derived from the former. (9)

(10)

Inta aracan kannai mutikkittiruntan- e (Herring, 1991: 259) this king eye.ACC close.PRG.PST.3M.S.TAG ‘The king was keeping his eyes closed, right?’ Nan poy aval ninriruntal-e anta itattil. (Herring, 1991: 275) I go she stand.PERF.PST.3.F.S.TAG that place.LOC ‘I went and stood in the place where she had stood.’

Although this analysis is not strictly based on historical data, I am not aware of research questioning the plausibility of this scenario of change. According to Herring (1991), the original development of a tag question marker is a case of subjectification (in current terms, we would add that it also crucially involves ‘inter’subjectification, or hearer-orientation), while further development towards a relativizer is the opposite. She speaks of “pragmatic unmarking” and “gradual diminishing of [the markers’] expressive impact” (pp. 277–8). While they are used subjectively as rhetorical question markers, they end up as grammatical markers of conjunction and subordination devoid of expressiveness. Herring (1991) concludes that, “Thus the history of the conjunctions.and the relativizer may be said to involve both subjectification - in the original extension [.] from true to rhetorical questions - and de-subjectification, in the grammaticalization of pragmatic devices as autonomous clause-linking elements” (p. 279) A similar development has taken place in Japanese, in this case a language that is well-documented historically. Japanese originally had no indirect question markers. The direct question marker ka developed into an indirect one from the 15th century on. Kinuhata (2012) sketched this development. (11) shows an example of the direct question use which is already documented in the early stage (8th century), while example (12) shows an example of the indirect question use from Modern Japanese. (11)

(12)

Wa-ga puru sode-wo imo mi-tu-ramu-ka I-GEN wave sleeve-ACC wife see-PFV-EPI-QUE ‘Did my wife see me wave my sleeves? (8thc) (Kinuhata, 2012: 800) Watasi-wa dare-ga paatii-ni ki-ta-ka siranai I-TOP who-NOM party come-PST-QUE know-NEG ‘I don’t know who came to the party’ (Kinuhata, 2012: 806)

156

H. Narrog / Language Sciences 47 (2015) 148–160

According to Kinuhata (2012), indirect questions are not only semantically less expressive of the speaker’s doubt (subjectivity), but also syntactic properties of subjectivity are lost, since the indirect questions are now embedded in past tense, negation, etc. He concludes that the development of indirect question functions “constitute[s] the counterevidence to subjectification” and instead labels them as cases of ‘syntacticization’ (pp. 809–810). Last but not least, Kranich (2010a,b) has argued for interpreting the historical development of the English progressive in Early and Late Modern English as a case of de-subjectification, counter to previous research that has presented it as a paradigm case of subjectification (especially Fitzmaurice (Wright) in Wright (1995), Fitzmaurice (2004b), and elsewhere). The main reason is that in the decisive period of grammaticalization of the English progressive to a common aspect marker in the second half of the 18th century subjective uses significantly decline in number. Interestingly, though, they rose again from the 19th century on when the aspectual uses were firmly established (cf. Kranich, 2010a; ch. 7.5, 8.2). Specifically, certain types of subjective use, evaluative and emphatic ones, as exemplified in (13) declined, while uses as in example (14), which Kranich (2010a) labels as ‘interpretative’, have been on the rise. (13)

(14)

They were now able to speak to each other and consult. That Louisa must remain where she was, however distressing to her friends to be involving the Harvilles in such trouble, did not admit a doubt. Her removal was impossible. (archern1800-49.bren1818aust.f5) (Kranich, 2010a: 221) HARRY. Why, it is possible you may yet receive a valentine. SOPHIA. Nay, now, but don’t you go to think that I am asking for one; for that would be very wrong of me, and I know better. (archern1750-99.bren1792holc.d3) (Kranich, 2010a: 225)

Thus, Kranich (2010a: 245) observes that “the rise in the subjective uses that one sees in the 19th and 20th centuries only affects a very specific meaning of the progressive”, namely the interpretative function. In her view, this function “was built on the aspectual use” (p. 246). As for the temporary decline of subjective uses of the progressive, Kranich suggests that, “[.] in order for a form to be recruited for the expression of subjective meaning, the speaker has to be free to decide whether or not to use the marker in question [.]. Now if the use of a particular marker is felt to be appropriate (or even obligatory) in a particular grammatical context, regardless of speaker involvement, attitude, or desire for emphasis, then such subjective meanings can no longer be associated with this marker in such a context.” This is a quite plausible explanation. If a grammatical marker becomes extremely frequent and obligatory in certain contexts, it obviously cannot retain its ‘subjective expressiveness’. At each instance of use the speaker would be forced to be ‘subjective’ if intended or not. One can also put it reversely in more traditional grammaticalization terms, namely that the expressiveness of a marker will ‘wear off’ or ‘bleach off’ through very frequent use. While the English progressive developed new subjective uses instead, this is not necessarily the case, as Kranich (2010b) argues. It seems that in the course of the development of tense–aspect markers in other European languages, expressive subjective uses simply wore off, without being “replaced”. 3.4. Counterdirectionality (2): ordering between subjectification, intersubjectification, and textual orientation As explained in 3.2, in the recent writings by Traugott, textual orientation has been eliminated or absorbed into subjectification. Instead, there are now two tendencies, subjectification and intersubjectification, whereby the former is hypothesized to precede the latter. Furthermore, in earlier Traugott writing, there was textual orientation as an early stage of change. Narrog (2012a,c) posits speaker-orientation, hearer-orientation and textual/discourse-orientation, claiming that the order between these changes is not yet settled but needs to be empirically investigated for different domains of grammar. I will start here with the question of the relative position of textual orientation vis-à-vis speaker-orientation/ subjectification and hearer-orientation/intersubjectification because relevant examples were already presented in the preceding paragraphs. In the case of ‘concessive’ may (cf. Section 3.2), we first find the development of subjective (epistemic) uses, then uses that imply a concession, that is are intersubjective, and finally clearly marked concessive constructions (cf. Narrog, 2012b: 39–40). In the case of the imperatives grammaticalizing into conditional or concessive conditional protases, it is clear that the intersubjective use formed the basis for the extension to the text/discourse-oriented uses. Concerning the development of question markers into textual markers presented in Section 3.3, we find the same tendency. The Japanese direct question marker, which has a clear intersubjective dimension of meaning historically unambiguously precedes the text- and discourse-organizing indirect question use. Concerning the Tamil rhetorical question markers developing various subordinating functions, Herring (1991) depicts the following scenario. The rhetorical questions are “a means of engaging the attention of the interlocutor”. For that purpose, the narrator constructs a hypothetical listener with whom he “interacts”, even speaking at times in this other listener’s “voice”” (p. 260). In our current understanding of (inter-) subjectification, we would clearly identify such uses as ‘intersubjective’. From this, according to Herring (1991), the ‘discourse-organizational’ and ‘textual’ function developed, that is textual orientation followed and eventually “replaced” (p. 264) intersubjective uses. Note that Herring (1991) labels those intersubjective uses as “expressive” and “subjective” because this was the terminology at that point. Traugott (1995: 47) commented on Herring (1991) that “these changes [.] do evidence a shift from attention directed primarily to the hearer toward more internally oriented, more solely speaker-based text organisation.” This comment is interesting because Traugott seems to admit here a shift from the intersubjective (which

H. Narrog / Language Sciences 47 (2015) 148–160

157

was not yet distinguished from the subjective) to the textual or the subjective (if the textual is included in the subjective). This leads to the next question, namely the ordering between the subjective and the intersubjective. In this area, caution is in place, because intersubjectivity in Traugott’s theory of semantic change differs from the concept of intersubjectivity in the work of a number of other linguistic scholars. Traugott (2003: 128) states that intersubjectivity as “involves SP/W’s attention to AD/R[addressee/reader] as a participant in the speech event” and that “expressions of intersubjectivity are expressions the prime semantic or pragmatic meaning of which is to index speaker’s attention to addressee self-image” (Traugott, 2010: 32). For Nuyts, another proponent of ‘intersubjectivity’ in functionally-oriented linguistics, in intersubjective expressions, “the speaker “indicate[s] that the evidence is known to (or accessible by) a larger group of people who share the conclusion based on it”, thus leading to “shared responsibility” (Nuyts, 2001: 34). Finally, for Verhagen (2005), intersubjectivity is the basic setting of the speech situation involving two ‘conceptualizers’, the speaker and the hearer. Neither Nuyts’ nor Verhagen’s intersubjectivity have an explicit diachronic dimension (intersubjectification). Crucially, intersubjectivity in the Nuytsian sense is close to objectivity, and therefore comes out as almost the opposite of Traugott’s intersubjectivity. Diachronically we would therefore expect intersubjective uses in Nuyts’ sense to precede subjective ones, if we follow Traugott’s theory. This is what was actually observed by Cornillie (2004, 2005, 2008), in his study of the development of evidential verbs in Spanish. According to his study, historically older constructions with parecer and resultar (parecer que and resultar que) are more ‘intersubjective’, the historically newer ones more ‘subjective’. (15) is an example of the older construction, and (16) of the newer one. (15)

(16)

Decidi respirar un poco y viajar fuera del pais por unas semanas. Parece que el viaje me hizo mucho bien; segun elias, tenia mejor semblante. ‘I decided to breathe a bit and travel abroad for a couple of weeks. It seems that the journey did me good; according to them, I looked much better.’ (Cornillie, 2008: 72) DeIa una parte d’el veran alisos, que de lexos paresce estar cayendo sabre una tabla d’agua tan hermosa, tan clara. ‘On this side you will see alders, which from far seem to be falling over a water table, so beautiful and so clear.’ (Cornillie, 2008: 66)

It is reasonable, then, to conclude that “intersubjective modal meanings seem to precede the more subjective ones” (Cornillie, 2008: 73–4). However, ‘intersubjectivity’ here is intersubjectivity in the sense of Nuyts, as shared evidence, and not necessarily in Traugott’s sense. For example, in (14), the writer describes something that is based on the opinion of others, or shared between self and other, while in (15) the writer describes an impression solely based on his own judgment. Therefore Traugott (2010: 33–4) argued that “intersubjectivity in my view is an entirely different view of subjectivity and intersubjectivity from that of Nuyts (2001, 2005) and Cornillie (2004), and indeed orthogonal to theirs.” Furthermore with respect to the diachronic dimension, she suggested that “[i]f this view of intersubjectivity were dynamicized and taken as the basis of intersubjectification one would have to argue that intersubjectification precedes subjectification historically” (Traugott, 2010: 34). Something similar may apply to Ghesquière’s use of the term ‘intersubjective’ in her study of determiners in the NP (Ghesquière, 2010, 2011). In this study, she showed that secondary so-called intersubjective textual uses of secondary determiners such as whole, particular, such, or Dutch zulk precede noun-intensifying, that is, subjective, uses. (17) would be an example of an intersubjective use, and (18) of a subjective one. (17) (18)

Numerous though we of Middle England are, we are a minority of the whole population. (identifying; CB times) (Ghesquière, 2011: 94) In Detroit downtown there’s just not a whole lot of people there. It’s just pretty ugly generally. (intensifying; CB ukspok) (Ghesquière, 2011: 148)

Again, like in the case of Cornillie, the diachronic facts seem to be very solid, and the observations are valid. However, they may not be examples for intersubjective uses preceding subjective ones in the sense of Traugott. On the other hand, as stated above, intersubjectivity in the sense of Nuyts is indeed likely to precede subjectivity, so we can expect to find developments such as those outlined by Cornillie and Ghesquière also in other languages and areas of grammar. If we try to identify possible counterdirectional developments from intersubjective to subjective strictly in the sense of Traugott, what we do find is the development from intersubjective to textual/discourse-oriented markers outlined above, since in Traugott’s recent writing the textual/discourse uses are not acknowledged as such but instead as ‘subjective’ (see above in this section). The only other case that I am aware of concerns certain parentheticals, i.e. constructions that can be used instantaneously (without undergoing much grammaticalization) for high-level grammatical and discourse functions (cf. Kaltenböck et al., 2011). (19) (20)

Mary d don’t forget d is coming over to visit. Because John is, you say, a spy, we should be careful what we say to him. (examples from Kaltenböck et al., 2011: 864)

Don’t forget in (19) and you say in (20) are apparently not grammaticalized but used instantaneously. Nevertheless, these phrases immediately assume intersubjective ((19), (20)) and evidential (20) functions.

158

H. Narrog / Language Sciences 47 (2015) 148–160

It is thus reasonable to assume that also more established expressions like you know are intrinsically intersubjective, and do not need to go through a phase of subjectification. Fitzmaurice (2004a) tried to preempt such arguments against subjectification necessarily preceding intersubjectification in claiming that second-person you know (intersubjective) must be derived from first-person I know (subjective). This is historically not really clear but cannot be excluded either. Avoiding this problem, Brinton (2007) chose to study a first-person intersubjective parenthetical instead, namely, ‘I mean’. Based on the observation of the historical data, she stated that “in the case of I mean one cannot show that subjective meanings clearly precede intersubjective meanings” (p. 68). Traugott (2010) argues against this that “parenthetical I mean has always been pragmatically intersubjective. Over time it has been used more intersubjectively to express emphasis and assertion of the veracity of an utterance (Brinton, 2007) but it has not been intersubjectified.” In conclusion, there are definitely good examples for the development of textual meanings that do not only follow intersubjectification but even need intersubjective meanings as a basis for their development. On the other hand, cases of intersubjectification without subjectification, or subjectification following intersubjectification, according to our present knowledge, may be restricted to parentheticals. 4. Discussion The previous section was concerned with identifying areas of grammar and processes of change that in one way or the other appear not to conform with prevalent hypotheses about (inter-)subjectification or increase in speech-act orientation. Throughout these sections we have seen that for the identification of such cases of limitation the conceptual understanding of subjectification and intersubjectification is crucial. This problem was especially salient with presumptive cases of intersubjectivity preceding subjectification but it is also latent elsewhere. In particular, it concerns the question of so-called ‘de-subjectifications’ or ‘objectifications’, which has even more fundamental consequences for idea of unidirectionality. The rest of this section is therefore devoted to this issue. With ‘de-subjectification’ as well, the question what we conceptualize as ‘subjective’ is crucial. Naturally, the same holds for ‘de-intersubjectification’, a term that has probably not been used yet but that could be applied to some developments discussed in this paper. So far, the term ‘de-subjectification’ was only applied in 3.3, to cases where this label was proposed in the literature. However, depending on the interpretation, most of the developments sketched in 3.1 and 3.2 that I labeled as the ‘development of abstract grammatical meaning’ and ‘development of textual meanings’ out of primarily intersubjective meanings, can also be conceived of as cases of de-subjectification or de-intersubjectification. The development of abstract core case marking out of semantic cases may be conceptualized as a ‘de-subjectification’, the development of concessive may entails a backgrounding of the original subjective epistemic meaning in favor of intersubjective and textual meanings, and therefore may be labeled as ‘de-intersubjectification’. The development of subordinating devices out of questions and commands arguably involves ‘de-intersubjectification’. Many of these changes have in common with the changes presented in 3.3 that were explicitly labeled in the literature as ‘de-subjectification’ or ‘objectification’ that expressive subjective or intersubjective meanings wear off in highly frequent use in certain contexts at late stages of grammaticalization, and give way to meanings and functions that organize text and grammar. Thus, if this wearing off of expressively subjective and intersubjective uses is an appropriate understanding of de-subjectification and de-intersubjectification, probably all cases cited in 3.1, 3.2 and 3.3, neglecting finer differences, should be subsumed under this label. We may then conclude that ‘de-subjectification’, and in certain constructions de-intersubjectification, is the rule rather than the exception in secondary grammaticalization. The reasons for this were very accurately formulated by Kranich (2010a: 245–6; cf. Section 3.3). It is only to be expected that highly automatized and abstract grammatical meanings cannot be subjectively expressive anymore at every instance of use. Presumably they were subjective (intersubjective) when speakers chose them to enhance expressiveness (cf. Haspelmath, 1999: 1057; Hopper and Traugott, 2003: 73) at an earlier stage of grammaticalization, but with frequent use, and in certain constructions, this expressiveness gets diluted. On the other hand, the question is if the concept of (inter-)subjectivity, or of speech-act orientation should be confined to an expressive type of (inter-)subjectivity. For one thing, within the concept of ‘increase in speech-act orientation’, if linguistic expressions assume textual functions as the last, or one of the last, stages of their development, and the hearer- and speakeroriented functions are backgrounded in the course of this process, I still conceive of it as a change towards speech-act orientation, since speech-act orientation includes textual orientation. On the other hand, it is also possible to argue that if speakers appropriate linguistic forms that are (still) relatively rich in conceptual and emotive–evaluative contents to express concepts that have mainly a language-internal function we are dealing with a higher form of subjectification in which these forms are entirely internalized to the speaker’s reasoning, and her or his creation of discourse. I would therefore include most of the cases that other authors may identify with ‘de-subjectification’ in the tendency towards increase in speaker-orientation and textual orientation within the concept of speech-act orientation. Perhaps the trickiest case is the temporary prevalence of aspectual meanings ‘objectively’ describing a temporal situation in the development of the English progressive, which can also be extended to other tense–aspect markers, as described by Kranich (2010a,b). While the older expressively subjective meanings of the progressive were fading away, and the newer ‘interpretative’ subjective meaning had not yet developed, at least for a while, ‘objective’ aspectual meanings prevailed. On the other hand, if we look at the development of temporal meanings in general from a broader perspective, we see very regular the development from lexical constructions, to lower-order aspect, i.e. phasal aspect and (im)perfectivity, to higherorder aspect, i.e. event quantification and relative tense, to absolute tense, that is from less speaker-deictic to increasingly

H. Narrog / Language Sciences 47 (2015) 148–160

159

exclusively speaker-deictic meanings (cf. Hengeveld, 2011). That is, even if the expressively subjective meanings wear off, a tendency towards subjectivity on a more abstract plane holds very regularly. Therefore, while loss of expressive meaning does take place regularly, neither ‘de-subjectification’ and much less the term ‘objectification’ seem appropriate to me to characterize the overall development. 5. Conclusion Probing into possible limits to the role of the semantic processes of (inter-)subjectification (Traugott, 2003, 2010, and elsewhere), or an increase in speech-act orientation (Narrog, 2012a,c) in grammaticalization, I investigated the following two possibilities: first, regularly occurring changes not apparently involving (inter-)subjectification, and second, counterdirectionality against (inter-)subjectification or increasing speech-act orientation, including counterdirectionality of subchanges within (inter-)subjectification. Potential cases of these types of change, according to the literature and to the author’s own research, turned out to primarily involve cases of secondary grammaticalization. This confirms suggestions in the previous literature that if exceptions to (inter-)subjectification exist, they are primarily associated with secondary grammaticalization (cf. Traugott, 2010; Kranich, 2010b). As for regularly occurring changes not obviously involving subjectification or intersubjectification, I identified (1) the development of highly abstract grammatical meaning, which can place both through extension of meaning and through constructional reanalysis, and (2) the development of textual meanings, insufficiently captured by either subjectification or intersubjectification. However, these development of textual meanings are well-captured in the three tendencies comprising increase in speech-act orientation according to Narrog (2012a,c). As for counterdirectionality to presumably unidirectional sequences of changes, we found (1) apparent de-subjectification and de-intersubjectification, and (2) textual orientation as a later or end stage of grammaticalization rather than as an early stage as previously predicted. De-(inter)subjectification primarily concerns the fading away of expressive subjectivity once grammatical forms get automatized, extremely frequent and express increasingly abstract concepts. This is obviously a very regular process in secondary grammaticalization. However, I have argued that the overall tendency towards an increase in speech-act orientation is maintained if we acknowledge that the process of the development of textual meanings and other highly abstract meanings that serve to organize grammar and discourse are understood as the speaker’s internalizing concrete meanings for the creation of speaker- and text-deictic meaning. Overall then, while a certain type of subjective, and also intersubjective, meaning regularly decreases at late stages of grammaticalization, other speaker-oriented and text-oriented meanings take their place. Acknowledgments I am grateful to Svenja Kranich, Tine Breban, the participants in the workshop, and two anonymous reviewers for their valuable comments that were of great help in improving this paper. References Aaron, Jessi Elena, Torres Cacoullos, Rena, 2005. Quantitative measures of subjectification: a variationist study of Spanish salir(se)*. Cogn. Linguist. 16 (4), 607–633. Abraham, Werner, 2005. An intersubjective note on the notion of ‘subjectification’. In: Broekhuis, H., Corver, N., Huybregts, R., Kleinhenz, U., Koster, J. (Eds.), Organizing Grammar: Linguistic Studies in Honor of Henk van Riemsdijk. Mouton de Gruyter, Berlin, pp. 1–12. Adamson, Sylvia, 1995. From empathetic deixis to empathetic narrative: stylisation and (de-)subjectivization as processes of language change. In: Stein, Dieter, Wright, Susan (Eds.), Subjectivity and Subjectivisation. Cambridge University Press, Cambridge, pp. 195–224. Anderson, Stephen R., 1977. On mechanisms by which languages become ergative. In: Li, Charles N. (Ed.), Mechanisms of Semantic Change. University of Texas Press, Austin, pp. 317–363. Breban, Tine, 2006. Grammaticalization and subjectification of the English adjectives of general comparison. In: Athanasiadou, Angeliki, et al. (Eds.), Subjectification: Various Paths to Subjectivity. Mouton de Gruyter, Berlin, pp. 241–277. Breban, Tine, 2008. Grammaticalization, subjectification and leftward movement of English adjectives of difference in the noun phrase. Folia Linguist. 42 (2), 259–306. Brinton, Laurel J., 2007. The development of I mean: implications for the study of historical pragmatics. In: Fitzmaurice, Susan, Taavitsainen, Irma (Eds.), Methods in Historical Pragmatics. Mouton de Gruyter, Berlin, pp. 37–79. Coates, Jennifer, 1983. The Semantics of the Modal Auxiliaries. Croom Helm, London. Cornillie, Bert, 2004. The shift from lexical to subjective readings of Spanish prometer ‘to promise’ and amenazar ‘to threaten’. A corpus-based account. Pragmatics 14 (1), 1–30. Cornillie, Bert, 2005. On modal grounding, reference points, and subjectification. The case of the Spanish epistemic modals. Annu. Rev. Cogn. Linguist. 3, 56–77. Cornillie, Bert, 2008. On the grammaticalization and (inter)subjectivity of evidential (semi-)auxiliaries in Spanish. In: Seoane, Elena, López-Couso, María José (Eds.), Theoretical and Empirical Issues in Grammaticalization. Benjamins, Amsterdam, pp. 55–76. Cornillie, Bert, Delbeque, Nicole (Eds.), 2006. Topics in subjectification and modalization. Belg. J. Linguist. 20. Creissels, Denis, 2008. Direct and indirect explanations of typological regularities: the case of alignment variations. Folia Linguist. 42 (1), 1–38. Davidse, Kristin, Cuyckens, Hubert, Vandelanotte, Lieven (Eds.), 2010. Grammaticalization, Subjectification and Intersubjectification. Mouton de Gruyter, Berlin. Devos, Maud, van der Wal, Jenneke (Eds.), 2012. Grammaticalization and (inter)subjectification in African languages. Afr. Linguist. 18 (special issue). Fitzmaurice, Susan, 2004a. Subjectivity, intersubjectivity and the historical construction of interlocutor stance: from stance markers to discourse markers. Discourse Stud. 6 (4), 427–448. Fitzmaurice, Susan, 2004b. The meanings and uses of the progressive construction in an early eighteenth-century English network. In: Curzan, A., Emmons, K. (Eds.), Studies in the History of the English Language II. Unfolding Conversations. De Gruyter, Berlin, pp. 131–173.

160

H. Narrog / Language Sciences 47 (2015) 148–160

Ghesquière, Lobke, 2010. On the subjectification and intersubjectification paths followed by adjectives of completeness. In: Davidse, Kristin, Vandelanotte, Lieven, Cuyckens, Hubert (Eds.), Subjectification, Intersubjectification and Grammaticalization. Mouton de Gruyter, Berlin, pp. 277–314. Ghesquière, Lobke, 2011. The Directionality of (Inter)subjectification in the English Noun Phrase: Pathways of Change (PhD dissertation). KU Leuven. Givón, Talmy, 1991. The evolution of dependent clause morpho-syntax in Biblical Hebrew. In: Traugott, Elizabeth C., Heine, Bernd (Eds.), Approaches to Grammaticalization, Types of Grammatical Markers, vol. 2. John Benjamins, Amsterdam, pp. 257–310. Haspelmath, Martin, 1999. Why is grammaticalization irreversible? Linguistics 37 (67), 1043–1068. Heine, Bernd, Claudi, Ulrike, Hünnemeyer, Friederike, 1991. Grammaticalization. A Conceptual Framework. Chicago University Press, Chicago. Heine, Bernd, Narrog, Heiko, 2010. Grammaticalization and linguistic analysis. In: Heine, Bernd, Narrog, Heiko (Eds.), Oxford Handbook of Linguistic Analysis. Oxford University Press, Oxford, pp. 401–423. Hengeveld, Kees, 2011. The grammaticalization of tense and aspect. In: Narrog, Heiko, Heine, Bernd (Eds.), The Oxford Handbook of Grammaticalization. Oxford University Press, Oxford, pp. 580–594. Herring, Susan C., 1991. The grammaticalization of rhetorical questions in Tamil. In: Traugott, E.C., Heine, B. (Eds.), Approaches to Grammaticalization, vol. I. Benjamins, Amsterdam, pp. 253–284. Hopper, Paul, Traugott, Elizabeth C., 2003. Grammaticalization, second ed. Cambridge University Press, Cambridge. Jung, Hakyung, 2009. Possessive subjects, nominalization, and ergativity in North Russia. In: Bubenik, Vit, et al. (Eds.), Grammatical Change in IndoEuropean Languages. Papers Presented at the Workshop on Indo-European Linguistics at the XVIIIth International Conference on Historical Linguistics, Montreal, 2007. Benjamins, Amsterdam, pp. 207–220. Kaltenböck, Gunther, Heine, Bernd, Kuteva, Tania, 2011. On thetical grammar. Stud. Lang. 35 (4), 852–897. Kinuhata, Tomohide, 2012. Historical development from subjective to objective meaning: evidence from the Japanese question particle ka. J. Pragmat. 44, 798–814. Kranich, Svenja, 2010a. The Progressive in Modern English. A Corpus-Based Study of Grammaticalization and Related Changes. Rodopi, Amsterdam. Kranich, Svenja, 2010b. Grammaticalization, subjectification and objectification. In: Stathi, Katerina, et al. (Eds.), Grammaticalization. Current Views and Issues. Benjamins, Amsterdam, pp. 101–212. Kuryłowicz, Jerzy, 1975[1965]. The evolution of grammatical categories. Esquisses Linguist. 2, 38–54 (Diogenes 1965: 55–71). Langacker, Ronald W., 1990. Subjectification. Cogn. Linguist. 1 (1), 5–38. Langacker, Ronald W., 1998. On subjectification and grammaticization. In: Koenig, Jean-Pierre (Ed.), Discourse and Cognition – Bridging the Gap. CSLI, Stanford, pp. 71–89. Lehmann, Christian, 1986. Grammaticalization and linguistic typology. Gen. Linguist. 26 (1), 3–22. Lehmann, Christian, 2002. Thoughts on Grammaticalization. In: Arbeitspapiere des Seminars für Sprachwissenschaft der Universität Erfurt, second, revised ed., 9. Seminar für Sprachwissenschaft der Universität, Erfurt. López-Couso, María José, 2010. Subjectification and intersubjectification. In: Jucker, Andreas H., Irma, Taavitsainen (Eds.), Historical Pragmatics. Mouton de Gruyter, Berlin, pp. 127–163. Narrog, Heiko, 2009. Varieties of instrumental. In: Malchukov, Andrej, Spencer, Andrew (Eds.), The Oxford Handbook of Case. Oxford University Press, Oxford, pp. 593–600. Narrog, Heiko, 2010. A diachronic dimension in maps of case functions. Linguist. Discov. 8 (1), 233–254. Narrog, Heiko, 2012a. Modality, Subjectivity and Semantic Change. A Cross-linguistic Perspective. Oxford University Press, Oxford. Narrog, Heiko, 2012b. Beyond intersubjectification – textual uses of modality and mood in subordinate clauses as part of speaker-orientation. Engl. Text Linguist. 5 (1), 29–52. Narrog, Heiko, 2012c. Modality and speech-act-orientation. In: van der Auwera, Johan, Nuyts, Jan (Eds.), Grammaticalization and (Inter-)subjectification. Contact Forum, Brussels, pp. 21–36. Narrog, Heiko, 2014. The grammaticalization chain of case functions – extensions of semantic roles vs. universals of grammaticalization. In: Luraghi, Silvia, Narrog, Heiko (Eds.), Perspectives on Semantic Roles. Benjamins, Amsterdam, pp. 71–99. Norde, Muriel, 2009. Degrammaticalization. Oxford University Press, Oxford. Nuyts, Jan, 2001. Epistemic Modality, Language, and Conceptualization. John Benjamins, Amsterdam. Onodera, Noriko, Suzuki, Ryoko (Eds.), 2007. Historical changes in Japanese: with special focus on subjectivity and intersubjectivity. J. Hist. Pragmat. 8 (2) (special issue). Palmer, F.R., 1990. Modality and the English Modals, second ed. Longman, London. Quirk, Randolph, Greenbaum, Sidney, Leech, Geoffrey, Svartvik, Jan, 1985. A Comprehensive Grammar of the English Language. Longman, London. Rhee, Seongha, 2012. Context-induced reinterpretation and (inter)subjectification: the case of grammaticalization of sentence-final particles. Lang. Sci. 34, 284–300. Trask, R.L., 1979. On the origins of ergativity. In: Plank, Frans (Ed.), Ergativity: Towards a Theory of Grammatical Relations. Academic Press, London, pp. 385–404. Traugott, Elizabeth C., 1980. Meaning-change in the development of grammatical markers. Lang. Sci. 2 (1), 44–61. Traugott, Elizabeth C., 1982. From propositional to textual and expressive meanings: some semantic-pragmatic aspects of grammaticalization. In: Lehmann, Winfred P., Malkiel, Yakov (Eds.), Perspectives on Historical Linguistics. John Benjamins, Amsterdam, pp. 245–271. Traugott, Elizabeth C., 1989. On the rise of epistemic meanings in English: an example of subjectification in semantic change. Language 65 (1), 31–55. Traugott, Elizabeth C., 1995. Subjectification in grammaticalisation. In: Stein, Dieter, Wright, Susan (Eds.), Subjectivity and Subjectivisation. Cambridge University Press, Cambridge, pp. 31–54. Traugott, Elizabeth C., 1999. The rhetoric of counter-expectation in semantic change: a study in subjectification. In: Blank, Andreas, Peter, Koch (Eds.), Historical Semantics and Cognition. Mouton de Gruyter, Berlin, pp. 179–196. Traugott, Elizabeth C., 2003. From subjectification to intersubjectification. In: Hickey, Raymond (Ed.), Motives for Language Change. Cambridge University Press, Cambridge, pp. 124–139. Traugott, Elizabeth C., 2007. (Inter)subjectification and unidirectionality. Journal of Historical Pragmatics 8/2, 295–309. Traugott, Elizabeth C., 2010. (lnter)subjectivity and (inter)subjectification: a reassessment. In: Davidse, Kristin, Vandelotte, Lieven, Cuyckens, Hubert (Eds.), Subjectification, Intersubjectification and Grammaticalization. Mouton de Gruyter, Berlin, pp. 29–71. Traugott, Elizabeth, Dasher, Richard, 2002. Regularity in Semantic Change. Cambridge University Press, Cambridge. Van der Auwera, Johan, Nuyts, Jan (Eds.), 2012. Grammaticalization and (Inter-)subjectification. KVAB, Brussel. Verhagen, Arie, 2005. Constructions of Intersubjectivity. Discourse, Syntax and Cognition. Oxford University Press, Oxford. Visconti, Jacqueline, 2004. Conditionals and subjectification: implications for a theory of semantic change. In: Fischer, Olga, Norde, Muriel, Perridon, Harry (Eds.), Up and Down the Cline d the Nature of Grammaticalization. John Benjamins, Amsterdam, pp. 169–192. Visconti, Jacqueline, 2013. Facets of subjectification. Lang. Sci. 36, 7–17. Wright, Susan, 1995. Subjectivity and experiential syntax. In: Stein, Dieter, Wright, Susan (Eds.), Subjectivity and Subjectivisation. Cambridge University Press, Cambridge, pp. 151–172.