On the Rythm of Grammaticalization Across Languages

24 Degrees of Grammaticalization across Languages

Béatrice Lamiroy and Walter De Mulder

1 Introduction

The central hypothesis of this paper is that an essential property of grammaticalization, viz. its gradual character, [1] also applies within a genealogical family. Thus, several grammaticalization processes may be more advanced in one language than in the other languages of the same family. We will provide evidence for Romance, by comparing French to two other Romance languages, Italian and Spanish. The cline goes like this:

(1)FrenchItalianSpanish

Various approaches have been proposed to account for grammaticalization phenomena, i.e. the development from lexical to grammatical items, and from grammatical to more grammatical forms. We will assume here, following Heine and Narrog (2010), that grammaticalization results from the interaction between phonetic, morphosyntactic, semantic and pragmatic factors, and that the following parameters can be used as mechanisms to uncover instances of grammaticalization:

- extension, both to new contexts and to other speakers of the linguistic community;

- desemanticization;

- decategorialization, i.e. the loss of the typical morphosyntactic properties of lexical items (or of less grammaticalized forms);

- phonetic reduction.

These parameters not only correspond to each of the above mentioned linguistic components, but also reflect the diachronic order followed by a grammaticalization process. Phonetic erosion thus represents the final stage of the process, and is not a sine qua non precondition for grammaticalization to occur. Most of the phenomena we will deal with are characteristic of grammaticalization which has already attained a fairly advanced stage in the process.[2]

As pointed out by Heine and Narrog (2010), the ultimate motivation behind any grammaticalization process is successful communication. This entails the development of simpler structures on the one hand (speaker’s economy) and of more “extravagant” structures (hearer’s economy) on the other (Geurts 2000; Haspelmath 2000). In both cases, frequent repetition by users is necessary for the propagation of the new form throughout the linguistic community (Bybee et al. 1994:8; Krug 2001), a factor which is subsumed under the extension mechanism in Heine and Narrog (2010).

Although Heine and Narrog (2010) do not consider obligatorification, one of Lehmann’s (1995) parameters, as a central concept but rather as a predictable by-product of decategorialization, we tend to believe that it plays an important role in grammaticalization, along with paradigmatization [3], because the constitution of paradigms is in part what grammar is about. We thus consider that a form speakers can freely choose according to their communicative intentions is less grammaticalized than one for which their choice is limited in a systematic way. For this reason, for example, the French personal pronoun may be considered more grammaticalized than its Spanish and Italian counterparts, since its expression in French, the only non-pro-drop Romance language, is no longer optional. We will come back to this point when we discuss the virtual disappearance of the “past simple” and the decrease of the subjunctive in French.

A few remarks are in order here. First and foremost, the cline in (1) is not intended as a tool to measure the distance between the three languages and the mother language, Latin. Second, what we will show is a very robust tendency rather than an absolute principle. For example, in most cases, Spanish will be shown to be the most conservative language. Yet Spanish, just like Portuguese and French, but unlike Italian, developed a future auxiliary out of the motion verb ir ‘go’:

(4) Sp. Va a llover.

go.3sg.ind.prs torain.inf.prs

‘It is going to rain’.

Moreover, as is well known, a grammaticalization process may start without going all the way, and sometimes language seems to evolve in the opposite direction, i.e. there are also cases of degrammaticalization (Norde 2009). However, as cases of degrammaticalization do not seem to amount to more than 10% of all grammatical evolutions (Heine and Narrog 2010), they do not seriously threaten the unidirectionality hypothesis of grammaticalization. Likewise, there is significant evidence that clearly argues in favor of the hypothesis that, all in all, French is more grammaticalized than the other Romance languages under study. A well-known example is that of (late) Latin casa ‘house’, which grammaticalized into the French preposition chez ‘at’, whereas both Spanish and Italian have maintained casa as a noun meaning ‘house’. Another case in point is the Latin noun homo ‘man’, which in French (and Occitan) grammaticalized into the indefinite pronoun on (e.g. Fr. On ne vit qu’une fois ‘You live only once’): this happened in none of the other Romance languages. In addition, in French, the same pronoun on was the source of a second grammaticalization process, so that it now also functions as an equivalent of the first person plural pronoun nous (e.g. on part ‘we are leaving’).

The idea that French, Italian and Spanish are grammaticalized to different degrees, with French being more grammaticalized than Italian and Spanish, and Italian more than Spanish (cf. (1)), will be illustrated here with data concerning central areas of the grammar: auxiliaries, tense and mood, existential sentences and demonstratives. However, the same tendency also holds for other areas, which we will not be able to go into here, such as word order (Marchello-Nizia 2006a: 131; 2009), prepositions (Goyens et al. 2002, Lamiroy 2001), etc.

2 Auxiliaries

Auxiliaries are defined as TAM verbs (Heine 1993), i.e. they express tense, aspect or mood, as in (6):

(6)Fr. Jean commence à manger.

Sp. Juan comienza a comer.

It. Gianni comincia a mangiare.

Johnbegin.3sg.ind.prs to eat.inf.prs

‘John begins to eat.’

In all Romance languages, auxiliaries are the outcome of grammaticalization processes (Lamiroy 1999, Squartini 1998); however, French displays less lexical variation among its TAM verbs than Spanish and Italian do, so its paradigmatic variability is more constrained. Spanish for example has at least six auxiliaries to express inchoative aspect, whereas French has only two, commencer and se mettre:

(7) Sp. Ana empezó/comenzó a reir.

Ann begin.3sg.ind.ps / begin.3sg.ind.ps to laugh.inf.prs

Anasemetió [4] / se puso a reir.

Ann herself put.3sg.ind.ps / herself put.3sg.ind.ps to laugh.inf.prs

Ana se echó a reir.

Ann herself throw.3sg.ind.ps to laugh.inf.prs

Ana rompió a reir.

Ann break.3sg.ind.ps to laugh.inf.prs

‘Ann began to laugh.’

The class of Spanish [5] and, to a lesser extent, of Italian [6] TAM verbs is also larger because certain iterative and habitual auxiliaries which did not survive in French are still in use in these languages. Paradigmaticization is thus more advanced in French:

(8)Sp. Vuelve a llover.

go back.3sg.ind.prs torain.inf.prs

It. ? Torna a piovere.

go back.3sg.ind.prs torain.inf.prs

‘It rains again.’

Sp. Solía verla cada día.

use.3sg.ind.impf see.inf.prs=her.acc.sgevery day

‘He used to see her every day.’

With respect to progressive aspect, Spanish and Italian respectively use estar, stare ‘to be’ and ir, andare ‘to go’ followed by the gerund:

(9a)Sp. Está / va diciendomentiras.

It. Sta / va dicendobugie.

be.3sg.ind.prs / go.3sg.ind.prs say.ger lies

‘He is lying.’

A similar construction used to exist in French, but gradually disappeared from the 17th century on, only surviving in semi-idiomatic expressions such as

(9b)Fr.Le problème va croissant.

the problem go.3sg.ind.prs increase.ger

‘The problem is increasing’.

Several accounts have been proposed in the literature for the disappearance of the gerund with progressive meaning in French, both external factors such as the criticism by normative grammarians, who consider it an Italianism, and internal factors, such as the competition with the imperfect, which can also express progressive meaning (Schoesler 2006). In our view, the emergence of the “extravagant” infinitival structure être en train de and its gradual extension in French, which eventually made the progressive structure with gerund disappear, may be partly[7] due to the fact that the new structure enters the same paradigm as all the other TAM verbs, i.e. a verb necessarily followed by an infinitive. The infinitival structure thus has a “magnetic” effect, attracting all TAM verbs into a similar formal pattern.[8] Whereas être en train de + infinitive only took [+hum] subjects and was followed by agentive verbs at first, its distribution progressively extended to all kinds of contexts (Mortier 2007).

In sum, French TAM verbs not only are more limited than their Spanish and Italian counterparts, the class of auxiliaries is also more homogeneous since they all display the same syntactic property, that of exclusively taking infinitival complements. Therefore paradigmaticization is more advanced in French than in the other languages.

3 The « past simple »

In contemporary spoken French, the passé simple with aoristic meaning has disappeared and has been replaced by the present perfect (Grevisse-Goosse 2007: 882). In written and spoken Spanish in contrast, the equivalent tense, viz pretérito indefinido, is not only common but is obligatory to indicate aoristic meaning [9] and in Italian, the passato remoto, still in use in the South, is disappearing in the North of the country: [10]

(10)a. Fr. ?* Hier il vint.

yesterday he come.3sg.ind.ps

It. ? Ieri venne.

Sp. Ayer vino.

yesterday come.3sg.ind.ps

b. Fr. Hier il est venu.

yesterday he come.3sg.ind.prf

It.Ieri è venuto.

Sp. ?* Ayer ha venido.

yesterday come.3sg.ind.prf

‘He came yesterday.’

The Romance present perfect, as is well known, is the result of a grammaticalization process of a resultative construction with the full verb habere + [past participle of a transitive verb + direct object], which was reanalyzed as [habere + past participle] + direct object, before extending to intransitive verbs. In a seminal paper on the evolution of the Vulgar Latin periphrastic perfect, Harris (1982) convincingly shows that its development consists of four stages, each of which can still be found in some Romance languages or dialects. Whereas the original value of the construction (stage I) is merely aspectual, expressing resultative meaning (which is still the case in Sicilian), in Stage II, the perfect acquires a temporal value, indicating past events whose result lasts until the moment of speech (like in modern Portuguese). In stage III, it marks past events with relevance for the moment of speech (like in modern Spanish) and finally, in Stage IV, the perfect is used to encode narrative events, i.e. it has become a marker of aoristic value, thus replacing the “past simple”. Although all four stages persist in French, of the three languages under analysis here, only French and northern variants of Italian have reached the last stage.

Interestingly, Detges (1999, 2000; 2006) argues that the evolution from stage I to IV should be accounted for in pragmatic terms, i.e. by subjectification.[11] While the resultative aspectual value is the starting point of the evolution of the Romance perfect (as is the case in many languages, cf. Heine and Kuteva 2006:140-182), the present perfect was progressively used to mark past events with current relevance. As pointed out by Detges, there is a metonymic implicature between current knowledge and the previous experience by which it was brought about, which can be exploited for pragmatic reasons: by choosing the present perfect, a speaker coding himself as subject of this construction makes a strong commitment with respect to the current consequences of previous events. In other words, the function of the present perfect is to emphasize the subject’s involvement in the state of affairs expressed by the verb. This metonymic implicature was conventionalized as part of the structure’s temporal meaning. Finally, in stage IV, which French attained in the 18th century, the construction came to refer to the past event itself without necessarily implying its impact at the moment of speech. It is therefore the actual equivalent of what the simple past used to be. Detges (2006: 69) points out that the shift from stage I to II took place much later in Spanish than in French, which probably explains why Spanish still is at stage III and has not reached stage IV, as French has. The above mentioned ongoing evolution in the North of Italy suggests that the Italian perfect might be undergoing a similar grammaticalization process as the one that affected French in the 18th century.

4 Existential sentences

Whereas French and Spanish existential sentences take the form of an impersonal construction with the locative clitic y and avoir / haber respectively (il y a and hay), Italian uses essere in combination with the locative ci. Essere can occur in 3d person singular or plural, c’è or ci sono:

(11)Fr. Il y a beaucoup de monde ici.

it there have.3sg.ind.prs a lot of people here

It.C’ è molta gente quì.

there be.3sg.ind.prsmuch people here

Sp. Hay mucha gente aquí.

have.3sg.ind.prs much people here

‘There are a lot of people here.’

Although Fr. il y a, It. c’è and Sp. hay are all three grammaticalized structures which originated either as an expression of possession (avoir / haber) or of existence (essere) combined with a now totally bleached locative pronoun y / ci, Meulleman (in press: 231 ff.) argues that French il y a is far more grammaticalized than its Spanish and Italian counterparts.

With respect to the four above mentioned parameters of grammaticalization processes, French il y a shows the following results. As far as extension is concerned, three important facts can be pointed out. First, French il y a came to be used as an obligatory tool to restrict a subject in focus, e.g.

(12)Fr. Il n’y a que toi qui sais it there have.3sg.ind.prs only you who know.2sg.ind.prs

où est la clé.

where be.3sg.ind.prs the key

‘Only you know where the key is.’

The Spanish and Italian counterparts here would use the adverb sólo / solo [12] respectively, since hay and c’è do not have a similar function. Secondly, with respect to the following NP, in existential sentences whose main discursive function is to introduce a (hearer-)new referent, the existential N is typically introduced by an indefinite determiner. Whereas this holds as a general rule for the Spanish examples in Meulleman’s corpus, French and Italian allow all kinds of determiners, even proper names (Meulleman, in press: 245), e.g.

(13)Fr. Ainsi à Roncevaux il y a Roland, son cor et

thus in Roncevauxit there have.3sg.ind.prs Roland, his horn and

son épée, ça fait trois.

his sword, that make.3sg.ind.prs three.

‘Thus in Roncevaux there is Roland, his horn and his sword, that makes three.’

(Le Monde, 28/02/1994)

It. Al fianco del regista ci sono Sienna Miller (…), il

next to the director there be.3pl.pres.ind Sienna Miller (…), the

grande Jeremy Irons

great Jeremy Irons, …

‘Next to the director there is Sienna Miller, the great Jeremy Irons, …’

(Corriere della sera, 04/09/2005)

It should be noted, however, as pointed out by Meulleman (in press: 248), that the extension of the existential NP is more relevant for French than for Italian, since only the French construction is really an impersonal structure, whereas the Italian c’è also may appear as a plural and with a preposed subject, as in

(14)It. Però il disagio c’è

e si sente lo stesso.

But the discomfort there be.3sg.ind.prs and …

‘But the discomfort is there and one feels it all the same.’

(Corriere della sera, 04/09/2005)

And finally, French il y a has become extremely common in sentence initial position of a bi-clausal structure in which a non-topic NP appears postverbally in the first clause (Lambrecht 1994: 169):

(15) a.Il y a des garçons qui sont partis.

it there have.3sg.ind.prs boys who go.3pl.ind.prf

‘Some boys left.’

b. Il y a ma mère qui est à l’hôpital.

it there have.3sg.ind.prs my mother who be.3sg.ind.prs at the hospital

‘My mother is in the hospital.’

At the same time, the above examples show that il y a is the most desemanticized of all three existential structures, as it no longer has existential meaning in this case, but merely functions as a discursive tool for thetic sentences. In cases like (15), in Spanish and Italian, subject inversion would occur. In modern French, however, where subject inversion is highly constrained, il y a now functions as a device to obtain the same discursive effect (Béguelin 2000). Interestingly, whereas Spanish hay is never used in this context, specialists of spoken Italian (Aureli 2003 ; Fiorentino 1999) mention a non-standard use of a “weak relative” with c’è that seems similar to the above indicated French structure, e.g.

(16) C’è un uomo che corre sulla spiaggia.

there be.3sg.ind.prs a man who run.3sg.ind.prs on-thebeach

‘A man is running on the beach.’

French il y a is also the most decategorialized of the three structures, since, as we have already mentioned, it no longer has the property of selecting indefinite NPs, which is a characteristic of impersonal existential sentences. Furthermore, il y a is a case of polygrammaticalization, since it also became a temporal preposition meaning ‘ago’. This is neither the case for Sp. hay nor for It. c’è:

(17) Paul est arrivé il y a

trois semaines.

Paul arriver.3sg.ind.prf itthere have.3sg.ind.prs three weeks

‘Paul arrived three weeks ago.’

Note finally that of the three existential structures, only the French one also shows phonetic erosion, il y a being often reduced to “y a” pronounced as [ja]. This form even occurs in written French, as the following press example testifies (Meulleman, in press: 235) :

(18)Rassurez-vous, y a un truc.

calm.2pl.imp.prs you, there have.3sg.ind.prs a trick

‘Don’t worry, there is a trick.’

(Le Monde, 28/01/1994)

5 Mood

That the French subjunctive is more grammaticalized than its Italian and Spanish counterparts was already pointed out by Harris (1978: 172), who considers the French subjunctive to be mainly a marker of subordination, in contrast with the remaining Romance languages, where mood alternation between the indicative and the subjunctive parallels a difference in meaning. The crucial difference between the two modes is based on the speaker's commitment with respect to a state of affairs in the case of the indicative and the avoidance of such a commitment in the case of the subjunctive (Dreer 2007 : 24). Obviously it is not possible to treat all uses of the subjunctive in the three languages in the space allotted here. We will therefore only zoom in upon a certain number of significant facts.

First, although the French subjunctive obviously is still in use, there is no doubt that it is much less frequent than it used to be (Buridant 2000: 337; Dreer 2007: 201 ff.; Lagerqvist 2009: 39). After verbs expressing hope or belief for example, the subjunctive could be used in Old French, but this is no longer the case now:

(19) OFr. Bien quident ce ait faitTristan.

well think.3pl.ind.prs this faire.3sg.sbjv.pstTristan