From the beginnings of linguistic awareness of grammaticalization in the early 19th century, it has been viewed as a reductive process. The meaning of the lexical item affected is bleached, which means it looses some of its semantic features. In many language, this is accompanied by phonological erosion. In general, increasing subjection under rules of grammar implies loss of freedom. Contrariwise, towards the end of the 20th century, increasing emphasis was laid on the fact that grammaticalization creates grammar. It may innovate grammatical categories and structure; and even where it only renews constructions and categories that had been in the language, it produces new instances of these.

Convergence

Grammatical constructions and categories – parameters and their values – are elements of particular linguistic systems. A category like verbal aspect is present in one language, say Ancient Greek, Russian or Yucatec Maya, and absent from another one, say German. The same goes for case, present in all of these languages, but absent from Chinese and Yucatec Maya, and almost all other grammatical categories. Nevertheless, the concept of the interlingual (or cross-linguistic or comparative) grammatical category has considerable empirical import (Lehmann 2018).

For instance, both Russian () and Yucatec Maya () have a basic opposition between imperfective and perfective aspect.

The grammatical formatives coding these aspects are completely different in the two languages. The Russian aspects are prefixes or suffixes on the verb, while the Yucatec aspects are clause-initial auxiliaries. The Russian aspectual formatives are grammaticalized from preverbs and derivational suffixes, while the Yucatec auxiliaries illustrated are grammaticalized from adverbs (Lehmann 2017). Given the fundamental typological differences between the two languages, such structural disparity is not surprising. Nevertheless, the degree of functional similarity is amazing. In both languages, the imperfective aspect views the situation designated without any temporal bounds, while the perfective aspect views it as bounded. In both languages, the aspects have no temporal features, although the perfective aspect in most contexts invites the inference that the situation is located in the past.

The second example concerns the dative as the relator marking the recipient of some transfer. Mandarin uses the formative gěi governing the recipient ().

English uses the formative to in a comparable construction as shown by the translation of . Brazilian Portuguese uses para in what is the same construction as in English ().

Again, the formatives coding the recipient relation are very different in the three languages. In particular, Mandarin gěi is a coverb while English to and Portuguese para are prepositions. Also, they are grammaticalized from very different sources: Mandarin gěi is essentially the verb ‘give’; English to stems from a Germanic local preposition; and Portuguese para in dative function is a recent expansion of a basically benefactive preposition. Despite such differences, they converge in their function to code the recipient role and have therefore legitimately been subsumed under the comparative concept ‘dative’.

Given such evidence, there is no doubt that grammatical phenomena of different languages may justifiably be subsumed under a common concept – an interlingual grammatical category. Such a conceptual operation does not, of course, imply sameness of the various language-specific categories thus subsumed. It does imply, however, that they are sufficiently similar to justify their comparison. This is no different from conceptual operations in other disciplines, where sociology or ethnology designate certain social arrangements in the German and in the Chinese culture as marriage, although both the specific arrangements and their social contexts are entirely different. It suffices that they share a set of constitutive features.

To the extent that such analyses and conceptions are correct, it may be ascertained that grammaticalization processes that started independently of each other in different languages may converge on a common goal. This does not mean that there is a universal set of grammatical categories. It does mean, however, that human cognition and communication is based on a set of concepts and operations which are so fundamental that they constitute, again and again, the goal of grammaticalization processes. In this sense, a grammaticalization process may be considered as goal-directed. What matters for the language system are not the disparate sources of those aspectual and dative formatives, but rather the universal condition that situations may be temporally bounded or not and that acts of transfer involving a recipient are basic to human culture. No language grammaticalizes all such fundamental concepts and operations. However, to the extent that grammatical structure is useful, some cognitive and communicative functions are grammaticalized; and these are not random, but chosen because they are useful.

Apart from such convergence on a set of basic notions of communication and cognition which are useful for any language, grammaticalization paths also converge on a similar formal structure of their target. This phenomenon is probably best explained by formal reduction.

Persistence

The convergence of independent grammaticalization processes is one side of the coin. The opposite side is the basic structuralist tenet that a linguistic sign, be it a lexical item or a grammatical formative, is an element of a particular language system in the first place. Its properties derive from its position – Ferdinand de Saussure's valeur – in this system. Since the systems of different languages differ, signs of different languages cannot be, strictly speaking, synonymous. (The possible exception of terms which are rigorously defined in some formal theory may be ignored here.) While translation equivalents and interlingual categories are widely recognized both in linguistics and in the non-academic world, we are, at the same time, aware that every pair of translation equivalents may be one in certain contexts, while either member has further aspects not matched by the other.

When it comes to grammatical formatives, such differences concern, in the first place and trivially, their structural properties, including their distribution. As a first example, the formation of the periphrastic perfect in French and English may be considered. The following table shows the distribution of French avoir ‘have’ and être ‘be’ as opposed to English have in the formation of the perfect.

English uses have as the perfect auxiliary regardless of the verb class, as in transitive have seen, atelic have worked and telic have entered. French – illustrating with the translation equivalents to these three verbs – uses avoir with transitive verbs, as in avoir vu, and in atelic intransitive verbs as in avoir travaillé, but être with telic verbs, such as être entré. The telic verbs in question are mainly verbs of motion like entrer ‘enter’. Given these distributions, two statements follow:

• ‘have’ as a perfect auxiliary is more grammaticalized in English than in French, as it has a wider distribution.
• The distribution of avoir and être shows persistence of the original function of these verbs. The combination of ‘have’ with a perfect participle started out with transitive verbs (as in the other Germanic and Romance languages). It only gradually gained ground in intransitive verbs, starting with atelic ones. Since the perfect participle of telic verbs designates a state of the subject, ‘be’ had already been in use with these. The limited distribution of avoir in the perfect is, thus, a trait of persistence of its origin.

Every finite Russian verb is obligatorily either in perfective or in imperfective aspect; tertium non datur. This is not the case in Yucatec. The two aspects illustrated are only the basic ones. Beside them, the language has a rather large set of aspectual auxiliaries which occupy the same syntagmatic position and could be used instead of those in . Among the consequences of this difference in valeur is the fact that if progressive aspect is to be coded, Yucatec uses a different aspectual auxiliary, while the Russian imperfective covers progressivity.

Every Yucatec aspectual auxiliary triggers one value from a conjugation category called status. Most auxiliaries trigger the incompletive, which appears in a. The perfective auxiliary is the only one to trigger completive status (b). The diachronic explanation is that most auxiliaries stem from verbs which take the rest of the clause as a complement clause, and complement clauses are in imperfective status. The perfective auxiliary is not grammaticalized from a verb, but from an adverb. This just modifies a clause whose completive status originally coded perfective aspect, thus reinforcing the inherited aspect. (The same goes, incidentally and mutatis mutandis, for the imperfective aspect.) The different statuses triggered by the aspectual auxiliaries are, thus, a trait of persistence of a property that stems from their origin. Persistence is, as it were, the diachronic side of the synchronic valeur.

Again, the dative formatives reviewed in the preceding subsection differ in their distribution. Specifically, while the English and Portuguese formatives are typical prepositions, Mandarin gěi may be used as a main verb (a).

It may also be attached to the main verb of the clause. Instead of , we may have .

Such configurations are entirely alien to the English and Portuguese prepositions. The peculiar distribution of Mandarin gěi is, again, a persistent trait from its origin in a full verb.

Moreover, the three dative relators differ in their functional contexts. Mandarin gěi is also used in benefactive function (b). Likewise Portuguese para is used in the same function ().

The diachronic counterpart to this synchronic feature is that the source of para is a compond Ibero-Romance preposition per+ad ‘for+to’, which originally only had benefactive function. On the other hand, the English dative preposition to is not used in benefactive function and therefore does not appear in the translations of b and , true to its origin as a local preposition. All of this does not, of course, exclude the possibility that one day Mandarin gěi may be used in an allative sense (as Portuguese para already is). Persistence is also a question of the degree of grammaticalization reached by the item in question.

The last example of persistence comes from the grammaticalization of focus markers from the copula of an erstwhile cleft-construction. While the resulting constructions are functionally rather similar across languages, some of these focus markers precede the focal component while others follow it. This is simply a persistent trait from the original cleft-construction: there the focal component was the predicate complement of the copula. This precedes the copula in Jula, but follows it in Mandarin, with the result that Jula has a postposed focus marker while Mandarin has a preposed one.

Reference

Hopper, Paul J. 1991, „On some principles of grammaticization.“ Traugott, Elizabeth C. & Heine, Bernd (eds.), Approaches to grammaticalization. 2 vols. Amsterdam & Philadelphia: J. Benjamins (Typological Studies in Language, 19); 1: 17-35.

Lehmann, Christian 2017, “Grammaticalization of tense/aspect/mood marking in Yucatec Maya". Bisang, Walter & Malchukov, Andrej (eds.), Unity and diversity in grammaticalization scenarios. Berlin: Language Science Press (Studies in Diversity Linguistics, 16); 173-237. DOI:10.5281/zenodo.823244. [download ]

Lehmann, Christian 2018, “Linguistic concepts and categories in language description and comparison”. Chini, Marina & Cuzzolin, Pierluigi (eds.), Typology, acquisition, grammaticalization studies. Milano: Franco Angeli (Materiali Linguistici); 27-50. [download ]