a-Topics in Italian/Romance and the Cartography of Children’s Inventions

2.1 a-Topics and the Route to DOM

Two specific properties will be highlighted in this section, both from a diachronic and a synchronic dimension, looking at the prepositional marking of the direct object from the perspective of Romance, in which the differential marker is typically the preposition ‘a’.^[2] The two properties are the following:

1. The diachronic development of DOM in Spanish indicates that a-marking was not present in earlier stages on the direct object when it was located in the canonical sentence internal position, yielding the order SVO (von Heusinger 2008, also based on Laka’s 2006 analysis of Cid). In contrast, preposition ‘a’ was substantially present when the direct object was preposed in a clause external Topic position, as in clitic left dislocation/CLLD constructions. In modern Spanish, instead, the direct object is typically a-marked in clause internal position thus illustrating a prototypical instance of DOM in Romance.

2. In some current Romance varieties, a-marking of the object is similarly limited to cases in which the direct object is preposed into the left peripheral Topic position in CLLD. This is straightforwardly the case in Balearic Catalan as described in Escandell-Vidal (2009); to some extent, it is also the case in modern Spanish where preposed objects in CLLD are always a-marked even in those cases in which they may appear without the marker in the clause internal position (e.g. with indefinites in the generic reading, Leonetti 2004); such marking in CLLD has effects on the interpretation (yielding a specific reading of the indefinite left dislocated direct object); to a more moderate extent, this is also the case in standard Italian (Belletti 2018a, 2018b) where preposed pronominal objects in CLLDs are typically a-marked (quasi obligatorily in first and second person, Berretta 1989).

Thus, a-marking appears to be a process primarily affecting the direct object when it is a topic, in the left peripheral Topic position.^[3] Presence of the marker adds some further interpretive feature on the DP to be combined with its topicality. To a first approximation, some affected/involvement interpretation can be identified as the enrichment brought about by the presence of preposition ‘a’ introducing the left peripheral object topic (Belletti 2018a, and Section 2.1.1). Below some relevant contrasts are reproduced, both for the diachronic and the synchronic dimensions indicated in i. and ii, above, in i.′ and ii.′, respectively:

As reported in von Heusinger (2008), 73% of animate definite direct objects in Cid were a-marked and preposed yielding a CLLD construction. In contrast, 80% of animate definite direct objects were not a-marked in sentence internal position, as illustrated in (2) and (3):

Hence, a-marking is limited to occurring in the left periphery in Old Spanish.

As described in Escandell-Vidal (2009), a-marking is exclusively performed on the left peripheral direct object topic, as the contrast in (4) illustrates. Such marking is optional. This is in contrast with Modern Spanish, where a-marking is obligatory on the left dislocated (animate) object in CLLD. This is so also in those cases in which it may be optional in clause internal position in DOM-Spanish, as Leonetti (2004) clearly illustrates through the following contrasts (also Belletti 2018a; Laka 1987):^[5]

As interestingly observed by von Heusinger (2008), modern translations of the sentences from Cid in (2) give rise to a minimal contrast with Old Spanish, as a-marking is quite widespread for direct objects in clause internal position in Modern Spanish, a classical manifestation of DOM. The contrast is illustrated in the examples reproduced in (7), from the indicated translations, in minimal contrast with (2) and (3):

In standard Italian a-marking is instead generally absent, in both clause internal and in the left peripheral topic position. However, whereas in the clause internal position it is completely excluded, in the left peripheral topic position it can be more or less accepted in some limited cases, with some variation among speakers. The best case allowing for a-marking of the left dislocated object is provided by the left dislocated experiencer direct object of psych-verbs of the preoccupare/worry class. The example below quoted from Belletti and Rizzi ‘s (1988) (footnote 27; also Belletti 2018a; Berretta 1989), due to Paola Benincà (Benincà 1986) illustrates the point:

The possibility of (8)a contrasts with the complete unacceptability of (9)a. In (9) there is no difference in acceptability between the object experiencer (9a) and the theme/patient object (9)b, both are excluded, in contrast with (8):

a-Marking of the preposed object is (quasi) obligatory in standard Italian when the object is a pronoun (in particular first and second person, singular, but also more generally). (10) illustrates this point with some relevant examples (from Belletti 2018a; Berretta 1989):

In conclusion, with different constraints and on a variable degree, a-marking of preposed object topics is characteristically a property associated with the topic position in the left periphery. Given the map of the left periphery (Rizzi 1997; Rizzi and Bocci 2017 for detailed updated description) in which a high Topic position is present, we conclude that preposition ‘a’ may thus be associated with the left peripheral Topic head, as illustrated in (11). The diagram in (11) also indicates the steps of the topicalization process, with the direct object moving to Spec/TopP and the preposition raising above it to some higher position, thus reestablishing the pre-positional order proper of Italian:^[6]

(11)

It is often observed that in DOM-languages, the interpretation of the marked direct object in a sentence internal position is presupposed to some extent (Ledgeway forthcoming; Ledgeway et al. 2019 and references cited there). A way to express this property cartographically may be to allow that in such languages also the clause internal low topic position in the periphery of the vP (Belletti 2004) be endowed with preposition ‘a’ (for more on the interpretation it brings about, 2.1.1 below), in a parallel fashion. This is the hypothesis illustrated in (12):

(12)

2.1.1 On the Nature of Preposition ‘a’

A natural question arises: What is the contribution of preposition ‘a’ to the topic interpretation? A precise answer to this question is not easy to formulate as the relevant interpretive feature(s) is(are) not transparent. Nevertheless, a crosslinguistic comparison with English may contribute some relevant ingredients towards it. The usual translation of ‘a’ in English is preposition ‘to’, a correspondence also adopted in (some of) the glosses above (footnote 4). However, at a closer look, preposition ‘a’ and preposition ‘to’ do not completely coincide in the meaning they contribute. In particular, as first pointed out in Kayne (2004), English ‘to’ cannot introduce a benefactive argument, differently from preposition ‘à’ in French, and also, we add, from preposition ‘a’ in Italian. Consider the case of the benefactive argument that can be added to the argument structure of a transitive verb as in the Italian pairs in (13):

(13)

a.	Farò una festa
	I will do a party
b.	Farò una festa ai bambini
	I will do a party *‘to’ the kids/for the kids
c.	Ho comprato un vestito
	I bought a dress
d.	Ho comprato un vestito ai bambini
	I bought a dress *‘to’ the kids/for the kids

The benefactive argument that is introduced by preposition ‘a’ in Italian, can be also added in English, but it is introduced by preposition ‘for’; ‘to’ is excluded in these cases, as indicated by the glosses in the examples in (13). The benefactive interpretation expresses a type of affectedness,^[7] as in the case of the a-marker of left peripheral direct object topics hinted at above. Interestingly, to my knowledge, to-Topics do not seem to exist in English, as their appearance has never been documented.^[8] Indeed, a similar affected status of the object is also recognized in the literature for the a-marked direct object in sentence internal position in Spanish. Consider, in this connection the following examples illustrating both physical and psychological affectedness (from Torrego 1998; von Heusinger 2008):

(14)

a.	Golpearon a un extranjero
	(they) beated ‘to’ a stranger
b.	*Golpearon un extranjero
	(they) beated a stranger
c.	Odia a un vecino
	(she) hates ‘to’ a neighbor
d.	*Odia un vecino
	(she) hates a neighbor

A related component of a-marking may also be one expressing an empathic attitude of the speaker toward the preposed object (in the sense of Kuno and Kaburaki 1977).^[9] So, an involvement component of the interpretation is to be understood as concerning both the object and the speaker. Thus, a feature [+a], suggesting affectedness/direct involvement in the event described by the verb, is proposed here as associated with preposition ‘a’ marking topics.

(11) and (12) can illustrate a diachronic path as the one described for Spanish in the previous section. However, in Spanish also a more grammaticalized step in a-marking may be operating. As Leonetti (2004) observes, a-marking on the direct object may not always be necessarily linked to topicality; ‘a’ can just be a marker of direct objects. Such a marker may thus be seen as a further step in the diachronic path, with ‘a’ being part of the Case-agreement system, as illustrated in (15) (Ledgeway et al. for closely related proposal):

(15)

In all cases ‘a’ contributes the affected/involvement interpretation described and dubbed [+a].

2.1.2 a-Marking as Case Marking: The Case of fare-a(/faire à) Causatives in Italian (and French)

Preposition ‘a’ as part of the Case-agreement system is also found in fare a causatives in Italian and /faire à causatives in French (Kayne 2004). In accord with the third step of a-marking described in (15), such ‘a’ is part of the extended projection of V, as such it is found in the ‘v’ spine. (16) reproduces the structure of a fare-a causative as described and discussed in Belletti (2017, updated in 2021):^[10]

(16)

In (16) ‘a’ is indicated as the expression of a dative Case. Indeed, if the causee argument of a fare-a causative is expressed through a pronoun, such pronoun is realized in the form of a dative clitic, as in (17):

(17)

Maria gli farà mangiare il gelato

Maria to-himcl will make eat the ice cream

This is a most direct indication that ‘a’ is part of the Case-agreement system in the fare-a causative; in turn, this is also a clear indication that the a-marker of the left peripheral topic is not Case related.^[11] In all cases of CLLD discussed above the resumptive clitic present in the clause following the left dislocated a-Topic is expressed in the form of an accusative clitic, both in Italian (in the limited cases in which a-Topics are possible) and in Spanish and in Balearic Catalan.

A last remark should be added at this point. The fact of being part of the Case-agreement system does not void the preposition of its interpretive [+a] feature. Hence, as noted in connection with (15), the a-marked direct object carries the relevant interpretation of affectedness/involvement. It must be noted now that the same holds true for the a-marked external argument of the vP in (16). Since Guasti’s (1993) analysis, the fact that the cause argument introduced by preposition ‘a’ carries an affected interpretation is generally recognized as one of the crucial characteristic properties of this argument in the causative construction.^[12]

Thus, the picture concerning the a-marker looks coherent.

The following further observation completes the relevant overview of the status and role of the prepositional marker ‘a’. It is a well- known fact that a class of psych-verbs, e.g. the piacere (/like) class in Italian, expresses its Experiencer argument with an a-marked-DP. (18) provides one illustrative example:

(18)

A Gianni piace la musica classica

to Gianni likes the music classical

(Gianni likes classical music)

(18) is an instance of a so called quirky-subject, also found in several languages, as widely described in the literature in which the subject properties of the argument introduced by preposition ‘a’ in (18) are described (e.g. Belletti and Rizzi 1988 on Italian, Sigurdosson 2002 on Icelandic). If the experiencer argument of (18) is realized through a pronoun, the pronoun is the dative clitic gli(to-him) in Italian; whence the classical ‘quirky’ status of such an argument as a subject marked with a Case different from nominative (in a nominative-accusative language):

(19)

Gli piace la musica classica

to-himcl likes the music classical

(He likes classical music)

Once again, this Case manifestation is in sharp contrast with the Case properties of a-Topics noted above, which do not correspond to a dative, as the resumptive accusative clitic in CLLD indicates. This property is in turn consistent with the fact that the a-Topics, contrary to quirky-subjects of the type in (19), are not Case related and fill positions which are not related and are external to the Case zone(s) of the clause, i.e. the Topic position either in the left periphery or in the in the vP-periphery of the clause structure.

3.1 Children’s a-Topics in the Left Periphery

As mentioned at various points in the previous sections, standard Italian is not a DOM language, in contrast with other Romance languages such as e.g. Spanish (and southern varieties of Italian as well). Furthermore, in contrast again with other Romance languages, a-Topics also have a very limited distribution (cfr. 8, 10 in 2.1). The previous sections have pointed out that diachronic evidence from Spanish clearly indicates that the process of a-marking started out in the left periphery and only in later stages involved the direct object in clause internal positions. The status of modern Balearic Catalan is especially revealing in this respect, as a-marking involves direct objects only (and optionally) when they are in the left peripheral topic position in CLLD. This may indeed be a characteristic pattern; the three stages illustrated in (11)-(12)-(15) thus schematize a diachronic path.

Evidence from children’s acquisition of CLLD in Italian has recently indicated (Belletti and Manetti 2019) that such pattern may also appear in language development during acquisition. In an experiment eliciting the relevant CLLD construction in which the left dislocated argument is the direct object, the (monolingual) Italian-speaking children investigated have made a wide use of a-Topics in their answers to the eliciting question, as illustrated by the production in (20) (e.g. Che cosa succede al mio amico, il pinguino?/What is happening to my friend, the penguin?):

Children have thus shown the ability to ‘invent’ a construction by overextending it, since a-Topics are very limited and marginal in standard Italian, as pointed out in 2.1. Children’s use of a-Topics in Bellletti & Manetti’s results closely resembles the process of a-marking of direct objects illustrated in Section 2 for Spanish and more specifically for Balearic Catalan: in all of their productions a-marking only concerns the direct object when it fills the left peripheral Topic position. Never is the direct object marked with preposition ‘a’ in the clause internal position, thus clearly indicating that children are not treating Italian as a DOM language.^[14] This systematic property of children’s productions is clearly illustrated by instances as the one in (21), in which in answering the elicitation question which concerned two direct objects involved in two different actions (Che cosa succede ai miei amici, la mucca e il pinguino?/What is happening to my friend the cow and the penguin?), the child realizes the first direct object in clause internal position in a SVO sentence, with no preposition ‘a’ present, and the second as a left dislocated a-marked a-Topic:

These results indicate that there is a grammatical space of ‘invention’, that children are able to exploit. The reasons leading to this overuse of the otherwise very limited a-Topics in the target standard Italian, are discussed in detail in the reference quoted and will not be further developed in detail here. Suffice it to say that such marking with the introduction of the [+a] feature associated with preposition ‘a’ allows children to deal with the computation of a configuration of structural intervention expressed in terms of featural Relativized Minimality/fRM (Rizzi 1990, 2004; Starke 2001), which would be otherwise too hard, in fact impossible, for them to compute at the young ages tested (4–6).^[15] Thus, under the pressure of the complexity of the computation, children exploit a grammatical option which is only rarely present in standard Italian. They show a capacity to somehow invent a construction by overextending it: their input may only contain the construction in a very limited way; in fact, productively, only with first and second person left dislocated personal pronouns; with lexical DPs, only marginally with object experiencers of psych-verbs (cfr. 2.1: 8–10). Children go beyond what they hear: their a-Topics are always lexical DPs of actional verbs. This is a very characteristic feature of children’s behavior during development. Interestingly, in this case children show that they are accessing the grammatical space that allows for this option and realize it in a way that is familiar from other languages. Considering the different status of preposition ‘to’ in English described in 2.1.1, the expectation is that young English-speaking children could not resort to a similar grammatical option, thus leading to the absence of ‘to-Topics’ in their productions in parallel experimental conditions. A new research question is generated by the results and considerations presented here, to be investigated and tested in future work.

Looking at the issue from the perspective of language change, it is possible that this overuse of a-Topics by the young Italian speaking children investigated may have revealed a potential change in progress in current standard Italian, possibly ultimately leading to DOM along the lines of the path illustrated by (11)-(12)-(15). Clearly, children are at the first stage (11), as shown by (21) and all the cases in which in a SVO sentence the direct object has never been marked with preposition ‘a’ by them. This remains as an open question; possibly in a few generations there will be an answer to it.^[16]

3.2 Some Further Considerations on a-Topics and Animacy Based on Children’s Productions

Recent experimental results have added some new elements that may bear on the investigation of the value that preposition ‘a’ contributes for the interpretation of the a-Topic. Belletti and Manetti (2021) have manipulated the ‘animacy’ of the object topic. Among their aims was that of determining whether an inanimate object would qualify for a-marking, thus yielding a felicitous a-Topic for the children’s grammar.^[17] The core of the results is that this is not the case. Some inanimate a-Topics were provided by children in their answer to the elicitation question (e.g.: che cosa succede a questa cosa, la macchina?/what happens to this thing, the car? Alla macchina, la lava/to the car, he washes it- M., 5;8, from Belletti and Manetti 2021: 14). However, this type of production happened to a very limited extent (12.5% of children’s answers). This is in sharp contrast to its overwhelming presence in the results from Belletti and Manetti (2019), where the direct object topic was always animate and a-Topics were produced in 74% of children’s CLLDs. As the authors conclude, these results indicate that animacy appears to play a role in the licensing of a-Topics (or whatever the proper feature is, with the proviso of footnote 15).^[18] This conclusion is not surprising if the status of affected object characterizes the a-Topic, in a way also implying some involvement on the part of the speaker (i.e. the answering child in this case). Possibly, a special interpretation was attributed by the child to the inanimate a-marked object in the few cases in which it was expressed as an a-Topic in the left periphery (as is plausibly the case for the moving object ‘car’ in the example mentioned above). Be it as it may, it seems that a fair conclusion to draw is that the enrichment in the interpretation of the left dislocated direct object topic provided by preposition ‘a’ preferably, in fact almost exclusively, concerns animate objects.