Grammaticalization Research Paper

As is shown by the example of Latin ille, grammaticalization may lead to the emergence of new grammatical categories (in the case in point, the article which did not exist in Latin). In all the adduced examples the grammatical output of the process has passed through a syntactic use of the lexeme—though this need not always be the case.

Lehmann (1995, p. 123) has suggested three relevant aspects of grammaticalization, which realize six parameters both at the paradigmatic and syntagmatic level (see Table 1).

While grammaticalization is a process, the six parameters are not processes but properties that linguistic forms possess in increasing or decreasing measure. A linguistic form shows integrity inasmuch as it maintains its phonological and semantic identity. Phonological erosion and semantic bleaching reduce the integrity of a form (Latin ille > Italian il, Spanish el, etc.). The scope is the structural size of a construction: when a full verb is grammaticalized as auxiliary it makes up one phrase with the main verb and thereby reduces its scope to just the main verb (see the example of habeo cultellum in entum ‘I have found a knife’ in Sect. 3.1). By paradigmatic cohesion is meant the formal and semantic integration of a form into a paradigm: the suppletive forms better, best are less paradigm integrated than longer, longest. Bondedness refers to the syntagmatic cohesion: the more a form is grammaticalized the more it will be syntagmatically constrained; French le, Spanish el, etc., are fully grammaticalized noun determiners which cannot appear in isolation. Paradigmatic variability refers, according to Lehmann, to the freedom with which the language user chooses a sign—and, again, variability decreases with increasing grammaticalization, if cases are weakly grammaticalized the case suffixes may vary with relative freedom:

(1) Hungarian

a gyerek az asztal-on jatsz-ik

The child the table-_{SUPERESSIVE CASE} plays

‘the child is playing on the table’

but asztal-nal (adessive) and asztallal (< asztal-val, instrumental–comitative) would also be possible. Another example of paradigmatic variability is: the use of Latin ille was not compulsory in preor postnominal position and was semantically governed, but in Romance languages it is in most cases impossible to have a noun without the article. Finally, the syntagmatic variability is the ease with which a form may be shifted in the linear order of a sentence: the more a form has undergone a grammaticalization process the less variable it will be; *I ha e a knife found or *I ha e a found knife are impossible in Modern English.

It has to be noticed that the processes alluded to under the three labels of weight, cohesion, and variability cut across different levels of language structure; therefore the same items (e.g., ille) turn out to be appropriate examples of different parameters (integrity, scope, etc.). Grammaticalization is a pervasive phenomenon whose effects may be seen at all linguistic levels. Consequently, ‘There is a growing interest in integrating grammaticalization with theoretical work in descriptive and historical linguistics’ (Giacalone Ramat and Hopper 1998, p. 1).

2. A Historical Sketch Of The Research

The classical examples of dicere habet and l’homme go back to Antoine Meillet (1916), who probably used first the term grammaticalization—though the concept is implicitly referred to by some previous linguists such as Wilhelm von Humboldt (Uber das Entstehen der grammatischen Formen und Ihren Einfluß auf die Ideenentwicklung, 1825), William Dwight Whitney (The Life and Growth of Language: An Outline of Linguistic Science, 1875), Georg von der Gabelentz (Die Sprachwissenschaft: Ihre Aufgaben, Methoden und bisherigen Ergebnisse, 1901) and others (see Lehmann 1995, Chap. 1, Heine et al. 1991, pp. 5–11).

As can be seen from the titles just quoted, grammaticalization has from the very beginning been considered in a large theoretical frame as one of the most characteristic developments languages may undergo. This holds also for the linguists of the next generations. Edward Sapir (1921, p. 102), for one, observed that ‘it is possible for a concrete concept, represented by a simple word, to lose its material significance entirely and pass over directly into the relational sphere [i.e., the expression of abstract morphosyntactic relations, PR] without at the same time losing its independence as a word’: forms live longer than their own conceptual content, which may be bleached and eventually fade completely. Sapir’s examples are the Chinese and Cambodian verbs for ‘give’ used in a very abstract sense to express the ‘indirecte object’ relation (see also Sect. 3).

Though not using the term grammaticalization, the French linguist Emile Benveniste (1968) took up again the case of habere, and called it a conservative mutation whereby a new (periphrastic) form takes over the function of an older one in the grammatical paradigm (e.g., habeo dictum ‘I have said’ instead of dixi ‘I said’).

3. The Present State Of The Discussion About Grammaticalization

3.1 The Continuum Of The Grammaticalization Process

Already Whitney and Benveniste have underlined the fact that linguistic changes do not happen abruptly: ‘Present possession often implies past action: habeo cultellum in entum, (…) habeo digitum ulneratum ‘I possess my knife (recovered after loss) [>‘I have found the/my/a knife,’ PR], (…) ‘I have a wounded finger’ [ >‘I have wounded my finger,’ PR].’ Thus, I ha e found the knife became the indicator of ‘a peculiar variety of past action contemplated as completed,’ (Whitney 1875, p. 91, cf. Heine et al. 1991, p. 7f, Benveniste gives an insightful philological description of how the periphrastic forms with habere came first to be used and later spread out.

As a matter of fact, grammaticalization, like all terms ending in- ization, alludes to a gradual process, not a state. Grammaticalization is different from grammar. Even what has recently been alluded to as ‘synchronic grammaticalization’ implies a process whose final stage cannot be simultaneous with the beginning moment. This is not in keeping with the assumption by generative transformational grammar of rigidly separated grammatical categories. Actually, the ‘new trend’ of grammaticalization studies which has flourished in linguistics since the 1980s is basically oriented toward a functionalist approach, with very few exceptions. Grammatical categories are not seen as waterproof boxes; they permit, on the contrary, the gradual leaking of forms from one category to another.

Being functionally oriented means that attention is paid to the internal mechanisms, which govern linguistic structures, and the external psychological cognitive causes of grammaticalization. Several linguists have hinted at the fact that grammaticalization very often originates in discourse strategies (i.e., in pragmatics), evolving to syntactic structures and later obligatory morphological markers (up to a possible total erosion of the markers):

Discourse < Syntax <Morphology <Morphophonemics <Zero (see Givon 1979, p. 209, cf. Lehmann 1995, p. 13, Traugott and Heine 1991, p. 3, Bybee et al. 1994, p. 40f).

Habeo cultellum in entum originates at the discourse level: the speaker wants to underline that he does really possess (habet) a knife which has previously been found (in entum) and uses a construct where every word keeps its own value. This pragmatic, effect-loaded expression evolves to a standard analytic construction of the syntax (with inversion and no longer with syntagmatic variability (see Sect. 1 above): habeo in entum cultellum) and finally enters the morphological paradigm of the verb in venire ‘to find’ as its periphrastic (agglutinated) past form (note that at this stage of the Romance languages it is no longer possible to have the word order habeo cultellum in entum (*j’ai le couteau trou e ): the auxiliary has agglutinated to the main verb). As an example of morphophonemic evolution one can quote the French future dirai, dir-a, etc. ‘I shall say,’ ‘s/he shall say’ from periphrastic future forms with habeo: dicere habeo, dicere habet: the auxiliary has undergone phonetic attrition and has become an inflectional ending.

There is a continuum between the phases of the grammaticalization process: Heine et al. (1991, p. 1) quote the case of Ewe (Kwa branch of the NigerCongo family):

(2) mena ga kofi

1Sg-give money Kofi

‘I gave Kofi money.’

(3) me-ple β tru na kofi

1Sg-buy door give Kofi

a ‘I bought a door and gave it to Kofi.’

b ‘I bought a door for Kofi.’

(4) me-w d vevie na dodokp la

1SG-do work hard give exam DEF

‘I worked hard for the exam.’

In (2) na has the function of a full verb, meaning ‘give’; in (4) na can be interpreted only as a preposition (‘for’), whereas in (3) it is ambiguous and allows two different readings. Should one day the Ewe language admit for (3) only the second reading, then we would have a categorial shift (_VERB>_PREPOSITION), much in the sense of the evolution of Old English hwil ‘time’ (_NOUN) to while (_CONJUNCTION) (see above, Sect. 1). At the end of an uninterrupted process there is really a break and a category shift.

Shifts like _VERB >_PREPOSITION, _NOUN>_CONJUNCTION are examples of so-called transcategorization. Items belonging to a main (and open) category (_VERB, _NOUN) pass to a functional (and restricted) category (e.g., _{ADPOSITION PRE}-or _PREPOSITION according to the structure of the language, concerned, _CONJUNCTION): these examples clearly hint at a move from a less grammatical to a more grammatical categorial status—it is however clear that the term transcategorization (or recategorization) is more appropriate than decategor(ial )ization used by Hopper (1991, p. 30f) and others.

3.2 The Dynamic Of The Grammaticalization Processes

In such a dynamic typology of grammaticalization it is possible to account also for interrupted grammaticalization processes: in Old Italian it was possible to use movement verbs (andare ‘go’ and enire ‘come’) as auxiliaries in far more cases than in Modern Italian (cf. Giacalone Ramat and Hopper 1998, p. 2f, with further references). Further, Hopper has drawn attention to what he calls ‘emergent grammar,’ i.e., to the incipient stages of grammaticalization at the level of discourse (see above): the negative French adverb pas meant originally ‘step’; therefore it was used first with movement verbs (‘he doesn’t go a step’), but later it spread to verbs such as ‘eat,’ ‘see,’ etc. and became the general mark of negation, ousting other nouns that denoted a least quantity and were used, like pas, to reinforce the negation (e.g., mie ‘crumb,’ gote ‘drop,’ etc.): specialization principle. Thereby the original meaning of pas was completely lost (bleaching process).

As noted by Hopper himself, the principles he suggests (specialization, persistence, divergence, etc.) characterize linguistic change in general and are not distinctive for grammaticalization (Hopper 1991, p. 21). This has raised the problem of defining the limits of grammaticalization vs. other dynamic linguistic processes (cf. Giacalone Ramat and Hopper 1998).

3.3 Unidirectionality Of Grammaticalization?

Unidirectionality is a much debated question in the literature on grammaticalization. Many linguists maintain that the evolution towards grammar cannot be reversed (Lehmann 1995, p. 19, Traugott and Heine 1991, p. 4f, Heine et al. 1991, p. 31f ). This is true if by this it is meant that once a lexeme has entered the grammar, it will hardly come back and assume again a lexical value. But there are many instances of items or morphemes leaving the realm of grammar and assuming an autonomous value as lexical entries: Hungarian, Spanish, and Italian have a large number of present participles used as nouns, e.g., Hungarian kolto (‘poet’ ( <kolt ‘to compose’), Spanish and Italian cantante ‘singer’ ( <cantar(e) ‘to sing’); nobody can fail to recognize in French seigneur ( >English sir) the comparative of the Latin adjective senex ‘old’ or in English elder ‘official in the Presbyterian church’ a comparative from Old English ald ‘old.’ Both were originally forms belonging to the grammatical paradigm of the adjectives, but later they developed an autonomous lexical meaning. Even syntactic constructs may develop to new lexemes: Dutch misschien ‘perhaps,’ ( <(het) mach schien literally ‘it may happen’), Italian forse ‘perhaps’ ( <Latin Fors sit (an) literally ‘be the case’), English perhaps literally ‘per chances (plural).’ Grammaticalization and lexicalization are two complementary processes (Moreno Cabrera 1998, Ramat 1992).

Degrammaticalization processes are certainly less frequent than grammaticalization processes, but they exist and have to be explained. The causes for their existence are to be sought in the overall tendency to use labels as economic symbols: e.g., suffix morphemes like -ade (from orangeade, lemonade, or even gatorade) can be separately manipulated and become autonomous lexemes: the label ade with the general meaning of ‘fruit juice’ is more economic than the phrase fruit juice, though not as transparent (see bus, in schoolbus, autobus, etc., as hyperonym for ‘public transportation vehicle’). Economic symbolism and noneconomic iconism are conflicting forces which are at work in every natural human language (cf. Ramat 1992, p. 557, Newmeyer 1994, p. 69). If we had just unidirectional evolution toward grammar we would expect that languages become more and more gram- maticalized, which by all evidence is not the case (see Moreno Cabrera 1998, p. 224). On the contrary, we have a double evolution towards grammar as well as towards lexicon (see Kuryłowicz 1975).

Phonogenesis is also a process which shows that loss of grammatical function may give rise to new lexemes (cf. Hopper 1990, 1992). As the German grammatical prefix ge-, used to indicate collectives (see Ge-beine ‘limbs,’ plural, Ge-drange ‘crowd’), was no longer felt as a prefix, it underwent phonetic erosion to gand gave rise to nouns such as Gluck ‘luck,’ Glied ‘limb,’ Gnade ‘grace’ which synchronically are no longer analyzable: phonetic junk, no longer recognizable as a functional morpheme, produced new lexemes. In the same vein of reasoning we have to consider also remnants of disappeared paradigms: this applies to dual forms which survived the loss of the dual declension: cf. Latin ambo ‘both’ (perhaps also octo ‘eight’) with the old dual marker -o.

Summing up, we may represent (see Table 2) the grammaticalization and degrammaticalization processes along a continuum which moves into two opposite directions. At both poles of the scale we have either clear lexemes (full words like habet, perhaps) or morphemes (inflectional –a or derivational –ade), while it is difficult to state when habet was delexicalized and the construct dicere habet became the norm to express the future, or ade was reanalyzed as a second member of a compound: that is why di(ce)re (h)a(be)t and lemon-ade appear in the previous scheme midway between (de)lexicalization and (de)morphologization. Transcategorization is a gradual process along a continuum; its final output represents a break in the categorial status of the input (see Sect. 3.1).

Bibliography:

1. Benveniste E 1968 Mutations of linguistic categories. In: Lehmann W P, Malkiel Y (eds.) Directions for Historical Linguistics: A Symposium. University of Texas Press, Austin, TX, pp. 83–94
2. Bybee J, Perkins R, Pagliuca W 1994 The Evolution of Grammar, Tense, Aspect, and Modality in the Languages of the World. University of Chicago Press, Chicago
3. Giacalone Ramat A, Hopper P (eds.) 1998 The Limits of Grammaticalization. Benjamins, Amsterdam
4. Givon T 1979 On Understanding Grammar. Academic Press, New York
5. Heine B, Claudi U, Hunnemeier F 1991 Grammaticalization. A Conceptual Framework. University of Chicago Press, Chicago
6. Hopper P 1987 Emergent grammar. Berkeley Linguistic Society 13: 139–57
7. Hopper P 1991 On some principles of grammaticization. In: Traugott E, Heine B (eds.) Approaches to Grammaticalization. Benjamin, Amsterdam Vol. 1, pp. 17–35
8. Hopper P 1992 Phonogenesis. In: Pagliuca P (ed.) Perspectives on Grammaticalization. Benjamins, Amsterdam, pp. 119–35
9. Kuryłowicz J 1975 The evolution of grammatical categories. In: J K (ed.) Esquisses Linguistiques Vol. 2. Fink, Munich, Germany, pp. 38–54
10. Lehmann Chr 1995 Thoughts on Grammaticalization. A Programmatic Sketch. (Arbeiten des Kolner Uni ersalien-Projekts Vol. 48). Lincom Europa, Munich, Germany
11. Meillet A 1948 L’evolution des formes grammaticales. In: A M (ed.) Linguistique Historique et Linguistique Generale. Champion, Paris, pp. 130–48
12. Moreno Cabrera J C 1998 On the relationships between grammaticalization and lexicalization. In: Giacalone Ramat A, Hopper P (eds.) The Limits of Grammaticalization. Benjamins, Amsterdam, pp. 211–27
13. Newmeyer Fr J 1994 Competing motivations and synchronic analysis. Sprachtypologie und Universalienforschung 47: 67–77
14. Ramat P 1991 Thoughts on degrammaticalization. Linguistics 30: 549–60
15. Traugott E C, Heine B (eds.) 1991 Approaches to Grammaticalization. Benjamins, Amsterdam
16. Traugott E C, Konig E 1991 On the expression of spatiotemporal relations. In: Greenberg J, Ferguson Ch, Moravcsik E (eds.) Universals of Human Language. Stanford University Press, Stanford, CA, Vol. 3, pp. 369–400