Grammaticalization is generally defined in terms of the diachronic change from a lexical item to a marker of a grammatical function, described since Meillet (1912: 139) as a process in which “the weakening of the meaning and the weakening of the form of the auxiliary word go hand-in-hand” (l’affaiblissement du sens et l’affaiblissement de la forme des mots accessoires vont de pair). Starting from this very strong universal prediction about the co-evolution of meaning and form, my talk will address two hypotheses on the universal homogeneity of processes of grammaticalization as discussed by Heine (2018), i.e., the Parallel-Reduction Hypothesis and the Meaning-First Hypothesis. The former can be addressed quantitatively, while the latter will be discussed qualitatively.

 

The Parallel-Reduction Hypothesis states that “form change parallels meaning change” (Heine 2018). This general statement is challenged by a quantitative cross-linguistic analysis of some 1.000 paths of grammaticalization presented by Bisang & Malchukov (2020). Based on our questionnaire for measuring grammaticalization in terms of eight parameters (similar to Lehmann 2015 [1982]) for each path, it was possible to show that the Parallel-Reduction Hypothesis is too coarse-grained for modeling cross-linguistic variation in scenarios of grammaticalization. The only parameters that significantly co–vary with the parameter of Semantic Integrity (semantic change) are Paradigmaticity and Syntagmatic Variability (morpheme order). What is particularly remarkable and contradicts general assumptions is that the covariation between Semantic Integrity and the purely form-related parameters of Phonetic Reduction, Bondedness and Allomorphy is very limited.

The Meaning–First Hypothesis claims that meaning change is primary and that it precedes form change in time. This is generally supported by the above observation that semantic change does not necessarily entail form change. However, proponents of the Meaning-First Hypothesis argue more specifically that the initial phase of meaning change is characterized by pragmatic inference, as modeled by the Invited Inference Theory (Traugott & Dasher 2002) or the Context Model of Grammaticalization (Heine 2018), with semantic conventionalization taking place at later stages. Given the difficulties of quantifying pragmatic effects, the Meaning-First Hypothesis will be evaluated from a qualitative perspective. Based on data from Chinese and some Southeast Asian languages on different domains of grammar (aspect, definiteness/ indefiniteness expressed by classifiers, kinship terms in pronominal function, etc.), it will be shown that pragmatics can play a more prominent role beyond the initial phase also at later stages of grammaticalization (Bisang 2020, 2021).

From the above analyses, I will conclude that:

• Grammaticalization scenarios vary cross-linguistically in terms of the parameters involved and the importance of pragmatic inference.
• If effects of pragmatic inference are important in processes of diachronic change but difficult to quantify, it is necessary to discuss the extent to which this limits the quantitative modeling of linguistic change.