Invited Inferencing Theory
In this section, we discuss a theory of semantic change in grammaticalisation proposed by Elizabeth Closs Traugott and Richard Dasher, focusing on the presentation in Traugott and Dasher (2002). This theory is called the Invited Inferencing Theory of Semantic Change because its main claim is that the form-meaning reanalysis that characterises grammaticalisation arises as a result of situated language use. In other words, semantic change is usage-based. Traugott and Dasher argue that pragmatic meaning or inferences that arise in specific contexts come to be reanalysed as part of the conventional meaning associated with a given construction. Inferences of this kind are invited in the sense that they are suggested by the context.
From invited inference to coded meaning
According to Invited Inferencing Theory, semantic change occurs when invited inferences become generalised. The distinction between an invited inference and a generalised invited inference is that a generalised invited inference is not simply constructed on line, but is preferred without yet being conventionalised. Inferences that subsequently become conventionalised are called coded meanings. According to this model, semantic change follows the path indicated in Figure 21.3.
The difference between a generalised invited inference and a coded meaning is that while a generalised invited inference can be cancelled, a coded meaning cannot. The ability to be cancelled is a property of inferences that can be eradicated by subsequent context. Consider example (13).
An inference associated with the temporal expression after is that Lily felt tired as a result of the trip. In other words, there is an inference of causality: the trip to Paris made Lily tired. However, causality is not a coded meaning associated with after, because it can be cancelled. In example (14), further information is given relating to the cause of Lily’s tiredness.
In this example, the second sentence cancels the inference of causality associated with after by providing explicit information concerning the cause of Lily’s tiredness.
Subjectification
Traugott and Dasher (2002) argue that the range of semantic changes apparent in grammaticalisation are most insightfully conceived in terms of shifts from more objective to more subjective meaning. This process is called subjectification (not to be confused with Langacker’s approach to grammaticalisation, which we call the Subjectification Approach; this is discussed in the next section). In Traugott and Dasher’s sense of the term, subjectification involves a shift from a construction encoding some speaker-external event to a construction encoding the speaker’s perspective in terms of location in space and time, or in terms of the speaker’s attitude to what is being said. This is called the grounding of the speaker’s perspective, which is thereby lexicalised (becoming part of the coded meaning). For example, while the ALLATIVE meaning of be going to represents a concrete and objective event, the FUTURE meaning grounds the assertion with respect to the speaker’s subjective perspective. The FUTURE sense of the construction encodes the speaker’s ‘location’ in TIME relative to the event described in the utterance. Consider example (3a) again, which is repeated here as (15).
In this example, gonna indexes the speaker’s present location in TIME, marking the assertion as future-oriented from the speaker’s perspective. In this way, the grammaticalisation of be going to from ALLATIVE to FUTURE involves a shift from a more objective meaning to a more subjective meaning. Like other examples of subjective meaning, this involves deixis. As we have seen, deictic expressions encode information that is grounded in the speaker’s perspective. For example, spatial deixis grounds an entity relative to speaker location, as in expressions like here and there, whose reference can only be fully understood relative to the speaker’s location. Similarly, temporal deixis concerns the subjective grounding of speaker ‘location’ in TIME, as reflected in the use of tense and temporal adverbials such as yesterday and tomorrow, as well as in the future sense of the be going to construction. These expressions can only be fully interpreted if we know ‘where’ in time the speaker is located. As we have seen, person deixis governs the use of personal pronouns like I versus you, which are also grounded in speaker perspective. Another class of expressions that are subjective in this sense are the modal verbs, which encode information relating to possibility, necessity and obligation (among others), and thus encode these aspects of the speaker’s perspective.
Intersubjectification
A subsequent grammaticalisation process is intersubjectification. This relates to a shift from objective meaning to a meaning that grammatically encodes the relationship between speaker and hearer. For instance, Traugott and Dasher (2002) discuss social deixis in relation to the Japanese verb ageru, ‘give’. They note that until recently ageru was an honorific verb, which means that it had to be used by a speaker (giver) who was of an inferior social status to the (hearer) recipient. In other words, part of the meaning of the verb was to signal the recognition of differential social status. More recently, this verb has begun to be used to express politeness, regardless of the relative social status of the giver and recipient. In other words, a shift has occurred in which the expression has acquired a different intersubjective meaning, evolving from an honorific expression to a marker of politeness.
Other examples of intersubjective meaning include pronoun forms in languages like French, which has ‘polite’ and ‘familiar’ variants of the second person singular pronoun (vous and tu respectively). The choice of pronoun is grounded in intersubjective perspective and encodes the social relationship that holds between interlocutors. Explicit markers of the speaker’s attention to the hearer, including politeness markers like please and thank you and honorific titles like Doctor and Sir also express intersubjective meaning. Figure 21.4 summarises the evolution of subjectivity in the semantic change that underlies grammaticalisation.
As this diagram shows, objectivity and intersubjectivity represent the extreme poles of the continuum. The more objective an expression, the more likely it is to be unmarked for modality (speaker attitude) and the least likely it is to be dependent on inference for full interpretation. It follows that Grice’s (1975) maxim of quantity predominates in this type of expression: the hearer assumes that the speaker has given as much information as is required, and is licensed to infer that what is not said is not the case. In contrast, as an expression moves along the continuum from objectivity to subjectivity, the more likely it is to be marked for speaker perspective, including modality, spatial and temporal deixis, and discourse deixis. The latter relates to expressions that link back explicitly to portions of preceding discourse (recall our discussion of example (6b)), or link pieces of discourse by means of connectives such as so, if or because. Furthermore, the more subjective the expression, the more dependent it is upon inference. Traugott and Dasher (2002) argue that the Grice’s (1975) maxim of relevance therefore predominates in this type of expression: the hearer assumes that more is meant than is said. At the most subjective point on the continuum, intersubjectivity, expressions are characterised by overt social deixis (for example, honorifics) and overt politeness markers (for example, hedges like I suppose and expressions like please and thank you). Relevance also predominates in this type of expression.
As we mentioned earlier, Traugott and Dasher’s use of the terms ‘subjectification’ and ‘subjectivity’ differs from Langacker’s use of the same terms in Cognitive Grammar. Recall from Chapters 15 and 16 that, in Cognitive Grammar, subjectivity is related to perspective or vantage point, and is a property of ‘off-stage’ or implicit concepts, while objectivity is a property of ‘on-stage’ or explicit concepts. While subjectification relates to speaker perspective in both approaches, for Langacker subjectivity correlates with the absence of overt expression while for Traugott and Dasher subjectivity correlates with the presence of an overt expression that signals subjectivity. Furthermore, as Traugott and Dasher (2002: 98) point out, Langacker’s model focuses upon the conceptual representation of event structures and how they are construed by the speaker. In contrast, the invited inference model focuses upon discourse, and therefore subjectivity is seen as contextually determined rather than as an inherent property of constructions.
The status of metaphor in Invited Inferencing Theory
Traugott and Dasher (2002) observe that metaphor has sometimes been considered the predominant force behind the semantic change that underlies grammaticalisation. As we have already seen, because metaphor ‘was conceptualized as involving one domain of experience in terms of another and operating “between domains” . . . changes motivated by it were conceptualized as primarily discontinuous and abrupt’ (Traugott and Dasher 2002: 77). As we noted earlier, the linguistic evidence does not support this view, and Heine et al.’s metaphorical extension account has to allow the language user a significant role in grammaticalisation in order to overcome this potential problem. For Heine et al., metaphor represents a macrostructure in grammaticalisation: it is within the conceptual frame established by a conceptual metaphor that grammaticalisation occurs. From this perspective, a conceptual metaphor pro vides the underlying schema that facilitates context-induced semantic change.
In contrast, Traugott and Dasher argue that many, perhaps most, of the regular semantic changes involved in grammaticalisation do not involve metaphor. Instead, semantic change arises from the usage-based processes we described above, in which invited inferences become generalised before becoming conventionalised as coded meaning. From this perspective, the changes involved are smaller-scale, mediated by context and language use. These changes are therefore metonymic in the sense that one concept ‘stands for’ another closely related concept rather than one concept being understood in terms of another as a result of a metaphorical mapping from one domain to another (recall our discussion of the examples in (8)). Traugott and Dasher’s model is summarised in Figure 21.5.
While studies that focus on metaphor complement the Invited Inferencing approach, Traugott and Dasher argue that the predominance of metaphor-based explanations in theories of grammaticalisation results from a tendency to focus on the beginning and endpoints of the process of change (the bottom of the model in Figure 21.5), without fully investigating the pragmatic processes that drive the process of change (the top of the model in Figure 21.5). In this respect, Traugott and Dasher tend towards the view that metaphor is epiphenomenal in the context of grammaticalisation: a ‘side effect’ of the grammaticalisation process rather than an underlying cause.
