Negative structures in neo-standard Italian: non è che (‘it is not that’) + S and mica (‘a crumb’) in comparison

2.1 Negative constructions with mica (< Latin ‘crumb’)

At least since Jespersen (1917), it is well-known that in some Romance languages the original pre-verbal negative particle may be reinforced by adding a post-verbal element with an original lexical meaning, which in the course of time becomes grammaticalized as a post-verbal negative operator. In Italian, the co-occurrence of the post-verbal particle mica (< Latin ‘crumb’) with the pre-verbal particle non fits this interpretation. As Bernini and Ramat explain, discontinuous negation with mica in post-verbal position has long become accepted into the standard to convey “a particular adversative meaning on the pragmatic level; its use […] implies that the speaker presupposes that whatever he is denying is on the contrary considered true or understood as realizable by his interlocutor” (2012 [1996]: 17). In (1), for instance, the speaker presupposes that his/her interlocutor is suggesting that s/he ought to pay attention to what people say:

(Teche Rai data base, 1973)

(KIParla Corpus, BOA1018)

In a similar way, the words uttered by the teacher quoted in (2) imply that the interlocutor (a student) could have quit the degree program if she was disappointed, she had no reason to believe that she would be held back against her will. Note that the latter idea has not been explicitly expressed in the previous interactional turns, it is rather a conversational implicature that is ruled out by resorting to the non … mica discontinuous construction.

An analogous interpretation had already been formulated by Cinque (1991 [1976]: 314), who remarked that “affiancando il mica al semplice non il parlante vuol negare una aspettativa da parte di qualcuno piuttosto che una asserzione. Mica, cioè, ha un contenuto puramente presupposizionale” [by adding mica to pre-verbal non the speaker denies someone else’s expectation rather than a certain assertion. In other words, mica conveys a purely presuppositional content]. Hence, mica is resorted to in order to deny an inference somehow associated to the words uttered in the previous interactional context, either by the interlocutor or by the speaker himself/herself.

In direct questions and requests, this discontinuous negation may function as a politeness, face-saving device “by suggesting that the interlocutor may reply negatively, thus freeing him from the obligation to fulfil the request” (Visconti 2010: 947), as in extract (3a):

Cinque (1991 [1976]: 315) observes that, when occurring in requests similar to (3a), mica could be replaced by a mitigating expression like per caso ‘by chance’, which suggests that the speaker does not really expect a positive answer, but thinks it may be worth a try. On the contrary, in affirmative (3b) and imperative sentences (3c) the insertion of per caso ‘by chance’ would not be acceptable.

Negative imperative utterances containing non … mica presuppose that the addressee would surely carry out a certain action if the speaker did not ask him/her not to do so. This explains why, unlike (4a), (4b) would not be considered a well-formed polite invitation (cf. Cinque 1991 [1976]: 316), despite being grammatically acceptable:

A final interesting point emerging from Cinque’s (1991 [1976]) discussion is that discontinuous non … mica negation may occur in appositive relative clauses (i.e., in relative clauses that convey additional information about a head whose reference has already been established, like 5a) but, as a rule, it is not acceptable in restrictive relative clauses, i.e., in relative clauses that enable the identification of a certain referent by restricting the predication to the (set of) element(s) specified in the clause itself (5b).

Cinque (1991 [1976]: 318) believes that the explanation of this difference lies in the fact that, unlike the propositional content of an appositive relative clause, which is comparable to an assertion and could be omitted without compromising the identification of the head (in 5a, your cousin, whose reference has already been established), the propositional content of a restrictive relative clause (in 5b, ‘there is a girl who does not want to be invited to dance’) needs to be presupposed as true to make the sentence meaningful. The presence of a discontinuous negation would call into question that presupposition, thus making the sentence incoherent.

As for post-verbal mica, the first structure mentioned by Berretta (2002 [1994])^[2], its use is generally restricted to sub-standard varieties of Italian spoken in the Northern regions of the country, where the national language is in close, long-term contact with Italo-Romance dialects that adopt a post-verbal negative particle as the standard, unmarked strategy to express sentential negation. The use of post-verbal mica without pre-verbal non tends to be associated with uneducated speakers whose competency in the national language is heavily interfered with the local dialect, and hence, stigmatised (e.g., Ruffino 2006; Guerini 2011). Other scholars (e.g., Regis 2017: 163–64) documented its presence in literary texts as a stylistic choice aimed at imitating Northern Italo-Romance dialect structures. Its occurrence seems to have recently become more tolerated (cf. Ballarè 2015), especially in direct questions^[3], as suggested by (6), the closing line of the advertising campaign (jokingly) promoting the radio program Forrest (which was run shortly after Mario Draghi was appointed Prime Minister of Italy in February 2021):

The negative construction entailing the use of mica in pre-verbal position, on the other hand, was already mentioned by Rohlfs (1966–69: III, 305) as “gaining more and more ground” in the 1960s, and it “is now regarded as standard in some reference grammars” (Cerruti 2020: 133). As Cerruti illustrates, this structure “seems to specialise in denying Hearer-new content” (2020: 135), that is to say, propositional content that is not known to the hearer, which may (or may not) be Discourse-old, i.e., textually or contextually evoked by discourse elements or inferred on the basis of shared knowledge. In extract (7), for instance, the negated content (‘a certain agreement can be depicted as ‘political’) is not known to the hearer and the speaker blames the choice of the adjective ‘political’ as indicative of not having a complete understanding of the on-going situation:

(Teche Rai data base, 1982)

Another example can be found in (8): as in the previous extract, the negated content (namely, that the students’ house is in Santo Stefano) has not been previously stated, thought the inference may have been triggered by some previous knowledge (e.g., most of the students invited to the party are from Santo Stefano or Santo Stefano is the area where most university students find their accommodation, etc.), and hence be interpreted as Discourse-old.

(KIParla Corpus, BOA3004)

Many scholars (e.g. Dahl 1979: 88; Bernini and Ramat 1996; Van der Auwera and Neuckermans 2006; Visconti 2010; Cerruti 2020: 133) agree that the three aforementioned negative constructions (non … mica; V + mica; mica + V), alongside with certain functions (associated with the development from non-canonical to canonical negation), can be accounted for in terms of a grammaticalization cline, commonly known as Jespersen cycle, whereby “the original negative adverb [in this case, pre-verbal non, FG] is first weakened, then found insufficient and therefore strengthened, generally through some additional word, and this in turn may be felt as a negative proper and may then in course of time be subject to the same development as the original word” (Jespersen 1917: 4). In the last stage of this grammaticalization cline mica ousts non as a pre-verbal negative operator: as Van der Auwera and Neuckermans put it, “we start with a preverbal negator, we end with a new one” (2006: 460), without any significant effect on word order.

Bernini and Ramat (2012 [1996]) point out that negative constructions entailing a pre-verbal negative operator, such as mica + V, appear more frequently in the world’s languages and hence can be considered as the unmarked members of an opposition having as the marked extreme post-verbal negative particles^[4]:

NEG particles in syntactic constructions show a clear tendency to appear in preverbal position (before the finite verb in compound forms), independently of whatever predominant basic order may be. (Bernini and Ramat 2012 [1996]: 23; small caps in the original)

As we shall see (Section 4), cross-linguistic frequency is extremely important in order to establish which structures are more likely to be favoured in language change, and hence, to become part of current usage in neo-standard Italian.

2.2 Non è che + S

The third negative structure mentioned in Berretta’s essay (non è che ‘it is not that’ + S) has been devoted scarcer attention in the literature. One notable exception is Bernini (1992), who identifies three structural features which, in his opinion, deserve special consideration:

1. the tense of the verb form occurring in the cleft sentence: present indicative is certainly the default option, though an imperfect or a future indicative verb form may sometimes occur, as illustrated in (9) and (10) below^[5]:

   (9)	però	poi,	come,	voti	non	era	che	erano …
   	but	then,	as	votes	neg	be-imperf.ind.3sg	that	be-imperf.ind.3pl
   	‘But then, as for the votes, it was not that they were […]’

   (10)	Non	sarà	che	è	più	facile	tollerare?
   	neg	be-fut.ind.3sg	that	be-pres.ind.3sg	more	easy	tolerate-inf
   	‘It will not be that it is easier to tolerate?’

   (Bernini 1992: 196)

2. the mood occurring within the that-clause, which may be either the indicative (11) or the subjunctive (12). The latter choice, which conveys a counterfactual nuance, is inhibited by the alleged disappearance of the subjunctive mood from spoken Italian, its place being taken by the corresponding indicative verb forms (e.g., Berruto 2017a: 41–42); nevertheless, as we will show in the next section, the occurrence of a subjunctive form is not uncommon.

   (11)	Anche	perché	non	è	che	dovevo
   	Also	because	neg	be-pres.ind.3sg	that	have_to-imperf.ind.1sg

   	verificare	quelle	firme
   	verify-inf	those	signatures
   	‘Besides, it is not that I had to verify those signatures’

   (Bernini 1992: 197)

   (12)	Non	è	che	io	mi	preoccupi,
   	neg	be-pres.ind.3sg	that	pro.nom.1sg	refl	worry-pres.subj.1sg

   	ci	mancano	dei	soldi	a	casa
   	pro.dat.1pl	be_missing-pres.1pl	some	money	at	home
   	‘It is not that I am worried, [but] at home some money is missing’

   (Bernini 1992: 196)

3. the interface between non è che and the appearance of negative quantifiers in the that-clause; as Bernini remarks, the latter remain within the scope of the negative particle non, and hence, a sentence like (13) is interpreted as if it contained a single negative operator (‘the treatment is not doing me any good’):

   (13)	e	poi	ho	sospeso	la	cura,
   	and	then	aux.pres.1sg	interrupted	det	treatment,

   	ho	visto	che	non	è	che	mi
   	aux.pres.1sg	realized	that	neg	be-pres. 3sg	that	pro.dat.1sg

   	faccia	niente
   	do-subj.pres.3sg	nothing
   	‘And then I interrupted the treatment, I realized that it is not that it is doing me any good’

   (Bernini 1992: 199)

To be more precise, the polarity of the (either negative or positive) quantifiers occurring within the that-clause tends to be neutralized. The latter are equalled to the former, as illustrated in (14), a semantically equivalent reformulation of (13):

Another structural peculiarity of this construction is that non è che may be followed either by an affirmative or by a negative clause; in fact, in Italian, a negative cleft construction is the only option available in order to deny the propositional content of a negative clause (cf. Bernini 2011). In (15), for instance, what is actually denied is the truth value of the assertion non si è fatto proprio niente per il Mezzogiorno ‘nothing has been done for Southern Italy’ and, at the same time, the contrastive meaning ‘something has already been done …’ is implicitly conveyed:

(Teche Rai data base, 1977)

This leads us to a final distinction formulated by Bernini (1992: 200–203), namely, that between “metalinguistic” and “descriptive” negative cleft constructions. While the former can be paraphrased as “non si può dire x di p” [one cannot assert x about p] and suggest a contrastive interpretation, as extract (15) above, the latter deny a certain propositional content, without implicitly conveying any contrastive meaning, as in (16):

(KIParla Corpus, BOA1018)

Yet, as Bernini (1992: 202) acknowledges, the boundaries between the two categories are blurred: “buona parte delle occorrenze non si lascia attribuire univocamente all’uno o all’altro dei due tipi estremi, ma si colloca in una sorta di continuum tra i due” [most occurrences cannot be univocally interpreted as belonging to one of the two groupings but occupy an intermediate position on a cline between them]. Accordingly, we will leave aside the latter distinction and focus on the three structural features discussed at the beginning of this section.

A few years later, D’Achille et al. (2005: 266) examined the data of the LIP corpus^[6] and found roughly 250 cleft sentences, one third of which were negative cleft constructions. The basic conclusions of their analysis were that i) non è che + S is a mitigating device which may be resorted to in order to explain with greater accuracy the content of a previous utterance^[7], and ii) in spontaneous informal speech, negative cleft constructions are being grammaticalized as negative operators almost equivalent to pre-verbal non.

A similar conclusion was reached by Ramat (2006: 356), who maintained that non è che + S “may undergo semantic bleaching, as it has been the case with many negative strategies in the long run of language history. It may lose, or be on the way of losing, its specific function and simply become equivalent to [unmarked] sentence negation”. This outcome is favoured by the syntactic position of non è che, which precedes the finite verb of the that-sentence, and hence is comparable to a pre-verbal negative operator, the preferred option in cross-linguistic terms (cf. Section 2.1). Hence, a sentence like (13) can easily be rephrased as:

A final point made by D’Achille et al. (2005) was that the analysis of the LIP corpus revealed another use of the negative cleft structure: the occurrence of non è che at the end of the utterance, often with a suspensive intonation, as if the speaker was inviting the addressee to draw his or her own conclusions on the subject under discussion. The same peculiarity had already been noted by Berretta (2002 [1994]: 379), who observed:

In un’altra conversazione del medesimo corpus, non conteggiata per il presente lavoro, ho notato l’uso di non è che sospeso in fine di enunciato, come formula di chiusura generica, che affida la conclusione alla cooperazione dell’interlocutore (o ad una riformulazione del parlante).

[In another interaction of the same corpus, not considered in the present work, I noticed the use of non è che at the end of the utterance, as a generic closing formula, which assigns the conclusion to the interlocutor’s cooperation (or to a reformulation uttered by the speaker herself)].

This use of non è che as a “closing formula” is consistently attested within the KIParla corpus as well – cf. (19) and (20) – and, as we shall see, it can be said to function as a discourse marker in its own right.

(KIParla Corpus, BOA3002)

(KIParla Corpus, BOA3018)

More recently, the structural features and discursive functions of the negative cleft construction have been discussed by Spina (2019), who compares the linguistic materials of the LIP corpus (dating to the 1990s) with those of a similar corpus collected two decades later, the Perugia corpus (cf. Spina 2014). Her analysis reveals that non è che has progressively extended its frequency of occurrence in informal face-to-face interactions and in telephone conversations, while the data concerning television talk shows display no significant difference (see Table 1), probably because they include a higher proportion of careful/planned speech.

Besides, Spina (2019) draws attention to a feature of the negative cleft construction that had not been mentioned in previous research, i.e., the fact that it may function as a double focus-marking device whereby the speaker can direct the attention of his/her hearers to both the negative structure and to a pronominal subject, by extracting the latter from the cleft sentence and placing it before the negative particle non, as io ‘I’ in (21a) or lui ‘he’ in (22a):

In both spoken and written Italian, subject pronouns may be put into focus by placing them in sentence-final position, like in extracts (21b) and (22b), where both pre-verbal non and non è che are equally acceptable. The choice of a negative cleft structure, however, admits two options – the focussed pronoun may be placed either at the beginning or in sentence-final position – unlike standard negation with pre-verbal non, which allows only the latter. This is because by placing the subject pronoun immediately before non we would obtain a pragmatically unmarked SVO sentence, where the pronominal subject (if openly expressed) is interpreted as the topic of the utterance:

Someone might object that the pronominal subjects of (21a) and (22a) are the topic, rather than a secondary focus, of the corresponding utterances^[8]. In fact, while a sentence may contain two (or more) topical elements with different degrees of topicality or givenness (Berretta 2002[1995]: 153), double-focus constructions are comparatively rarer, both in Italian and cross-linguistically. We will leave this question open, though the analysis of the larger conversational context in which extracts (21a) and (22a) are embedded appears to corroborate Spina’s (2019) interpretation by suggesting that the pronominal subjects io and lui function as “second occurrence focus” of the corresponding utterances, displaying “a combination of ‘focusedness’ on the one hand and ‘givenness’ on the other” (Baumann 2016: 503).

To sum up, the studies carried out in the last decades suggest that the higher frequency of occurrence of negative cleft structures may be due to their higher syntactic flexibility in comparison to standard negation with pre-verbal non, as well as to their “pragmatic versatility” (Spina 2019: 111), which enables the speaker to perform a number of discursive functions, including:

1. denying the propositional content of a negative clause (cf. extract 15);

2. when employed as a “closing formula”, inviting the addressee to draw his or her own conclusions on the subject under discussion (cf. extracts 19 and 20);

3. acting as double focus-marking device whereby the speaker can put into focus both the negative structure and a pronominal subject (cf. extracts 21 and 22);

As anticipated, the next section will be devoted to the comparison of the negative constructions occurring within the KIParla corpus with the corresponding occurrences in a thirty-hour sample from the Teche Rai data base. Further empirical evidence will be discussed in order to come to a better understanding of the tendency under way and assess Monica Berretta’s hypothesis concerning sentence negation in neo-standard Italian.

3.1 The data

RAI, the Italian national television company, began its broadcasting in 1954, with a single channel; a second national channel went on air in 1976, while a third channel was launched three years later. A selection of the programs (including documentaries, talk shows, news programs, etc.) broadcast from the 1950s onwards is available at the Teche Rai archive, a digital audio and video data base which can be accessed free of charge^[9].

In the 1950s and 1960s, the national television service fulfilled important social functions, including the promotion of literacy and the spread of Italian to larger sections of the population, who used to speak an Italo-Romance dialect as the most common (or even the only) language of everyday communication. But, most importantly, television was promptly endorsed as one of the language norm authorities that contributed to the emergence and diffusion of the new standard which is the focus of this issue. As De Mauro maintains in one of his well-known essays on the subject (2014: 96),

La televisione si fece scuola di italiano. Pochi anni dopo l’esordio della tv, in aree ancora prevalentemente dialettofone, […] fu possibile stabilire che l’ascolto abituale della televisione valeva, ai fini della padronanza dell’italiano, cinque anni di scuola: gli analfabeti e senza scuola imparavano a capire l’italiano come se avessero la licenza elementare; chi aveva la licenza si trovava proiettato tra gli italofoni all’inizio delle secondarie; con l’ascolto televisivo una licenza media portava all’italiano degli aspiranti universitari.

[Television became an Italian language school. A few years after the introduction of television, in those areas of the country where most people still used to speak an Italo-Romance dialect, it was established that, as far as competency in Italian was concerned, watching tv on a daily basis was comparable to five years of school attendance: illiterate people developed a proficiency equivalent to their counterparts with elementary school graduation; the latter displayed language skills similar to contemporaries entering secondary school; persons holding a middle school diploma attained the same proficiency of graduates aspiring to university studies.]^[10]

Another reason for choosing to analyse linguistic materials from the Teche Rai data base is that the recordings involve speakers from different regions of the country, whose Italian displays various local and regional features, and whose sociolinguistic profile and educational background are accordingly extremely heterogeneous. For the purpose of this study, thirty hours of recordings dating back to the 1960s, the 1970s and 1980s were selected and all the occurrences of sentence negation analysed. An outline of the recordings can be found in the Appendix; all the data were anonymised by removing personal names, place names and other identifying information.

As for the KIParla corpus (Mauri et al. 2019), its size is 70 hours, corresponding to roughly 700 thousand words. The data were collected in two Italian towns, Torino and Bologna, and include several types of interactions recorded in the academic domain, e.g., professor-student interactions in office hours and in oral examinations, academic lessons, semi-structured interviews collected by students within peer-groups, spontaneous conversations recorded by in-group members, involving both students and teaching staff. A noteworthy social feature shared by all the participants is, accordingly, the high level of education. A fully-fledged description of the KIParla corpus is offered in the Preface to this issue, hence we will skip further details and turn to the analysis of the data.

3.2 Results

The sample of recordings from the Teche Rai database contains roughly two thousand occurrences of sentence negation; however, only a marginal percentage of them displays a negation other than unmarked preverbal non. As illustrated in table 2, the total number of non-standard sentence negation occurrences amounts to 40; 29 occurrences out of 40 (amounting to 72.5 %) entail the use of a negative cleft sentence, whereas the remaining 11 occurrences entail the use of mica, either in a discontinuous non V mica structure (9 occurrences; 22.5 %) or in pre-verbal position (2 occurrences; 5 %). There are no examples of mica in post-verbal position: its use as the only negative operator is probably perceived as too regional, too ‘dialectal’ even by Northern Italian speakers, and is accordingly avoided.

In most of the recordings dating to the 1960s, a careful speech style prevails over spontaneous speech and, impressionistically, speech rate appears to be lower if compared to the recordings of the following decades. This however does not justify the apparent lack of variation in sentence negation.

All in all, the occurrences of negative cleft constructions entail less than 1 occurrence per hour of recording; yet this result is consistent with previous investigations (e.g., Spina 2019) attesting a gradual but steady rise in the percent of non è che + S occurrences.

The limited number of occurrences within the Teche Rai sample is largely compensated by the analysis of the KIParla corpus, whose data enable us to draw a clearer picture of the structural and functional features of negative cleft constructions. As illustrated in table 3, we were able to analyse more than 400 occurrences of non è che, and 48 instances of sentence negation involving the use of mica, either in a discontinuous non V mica structure (28 occurrences, amounting to 6.1 % of the total), in pre-verbal position (11 occurrences, 2.5 % of the total), or in post-verbal position (9 occurrences, 1.9 % of the total). It was not possible to determine the corresponding percentages in relation to the total number of occurrences of sentence negation within the corpus^[11]; nevertheless, the raw data show that, within the KIParla corpus, non è che is by far the most frequent non-standard sentence negation strategy – almost fourteen times more frequent than discontinuous non V mica negation – exceeded only by unmarked standard negation with preverbal non (either alone or in combination with another negative quantifier in post-verbal position).

If we turn to the structural features of negative cleft constructions mentioned by Bernini (1992), the first observation is that in both the Teche Rai sample and the KIParla corpus the verb form attested in the cleft sentence is invariably in the present indicative. No occurrences of imperfect (non era che…) or future tense (non sarà che …) are attested.

The analysis of the that-clauses, on the other hand, reveals that, despite the alleged disappearance of the subjunctive mood from spoken Italian, indicative is far from being the default choice: within the Teche Rai sample we notice an almost even split between indicative and subjunctive; as for the KIParla corpus, the indicative mood prevails (279 occurrences, roughly 78 % of the total), but a subjunctive verb form is attested in 77 (21.5 %) occurrences (cf. Table 4).

Quite predictably, the presence of the subjunctive conveys a more tentative and counterfactual value to the content of the that-clause, as in (23), where the speaker is considering the possible explanations for the sudden death of a colleague of his:

(Teche Rai data base, 1982)

(KIParla Corpus, TOC1006)

Both present and imperfect subjunctive (as in extract 24) are attested within the subordinate clause, whose predicate conveys the temporality of the entire sentence: present-time vs. past-time reference, as the last two examples illustrate (cf. Bernini 1992: 197–98).

On the contrary, an indicative verb form prevails in low formality contexts, when the style is closer to spontaneous speech, as in (25), or even because the speaker may be more familiar with the indicative (stava, in extract 26) than with the corresponding imperfect subjunctive form (stesse):

(KIParla Corpus, PTD012)

(Teche Rai data base, 1988)

Occasionally, non è che may introduce a direct question, as in (27), where the interviewer (S2) defiantly asks a politician (S1) whether he is tempted to disavow a controversial law that had recently been approved thanks to his very effort and collaboration:

(Teche Rai data base, 1979)

Within the KIParla corpus, direct questions preceded by non è che entail either a polite offer, as in (28)^[12], or a potentially face-threatening request, as in (29), where the choice of a negative cleft structure functions as a mitigating device which minimizes the impact of the question itself:

(KIParla Corpus, TOD2008)

(KIParla Corpus, PTD016)

In 19 occurrences non è che introduces a negative clause: as we mentioned in Section 2.2, in Italian, a negative cleft structure is the only means available in order to deny the propositional content of a negative utterance. Extracts (30) and (31) are two examples from the KIParla corpus: incidentally, the latter is the only instance of a that-clause containing a conditional verb form (cf. Table 3.4).

(KIParla Corpus, PTD010)

(KIParla Corpus, PTA005)

In this particular context (i.e., when non è che is used to deny the propositional content of the following clause), the indicative mood still prevails over the others: the subordinate clause typically appears in the form of a quotation containing the same mood employed in the original utterance (see Bernini 1992: 202).

A final observation emerging from the analysis of the KIParla corpus is that in 53 occurrences (amounting to 12.9 % of the total) non è che is employed as a “closing formula” (see table 3.3). When used to fulfil this function, non è che occurs in sentence final position (as in 32), but it may occasionally be followed by either an indicative or a subjunctive verb form as in extract (33) and (34), respectively. This suggest that non è che has the potential for becoming grammaticalized as a discourse marker, whereby the speaker invites the addressee to draw his or her own conclusions on the topic under discussion.

(KIParla Corpus, BOA3010)

(Teche Rai data base, 1962)

(KIParla Corpus, PTD016)

1960s

1970s

1980s