2.1. Expletive Negation

EN has been extensively studied, particularly in Romance languages (Espinal Reference Espinal1992; Portner & Zanuttini Reference Portner, Zanuttini, Horn and Katō2000; Del Prete Reference Del Prete2008; Yoon Reference Yoon2011; Makri Reference Makri2013; Pană Dindelegan & Maiden Reference Pană Dindelegan and Maiden2013; Greco Reference Greco2019, Reference Greco2020; Jin & Koenig Reference Jin and Koenig2021). This phenomenon is evident in clauses where the negative marker does not alter the sentence’s polarity, which remains affirmative:

Focusing on English, Jin & Koenig (Reference Jin and Koenig2021) note that comprehensive descriptive grammars of SE, such as Huddleston & Pullum et al. (Reference Huddleston and Pullum2002) and Quirk (Reference Quirk1985) make no mention of EN, nor is its use permitted by prescriptive grammars. However, Horn (Reference Horn2010: 125) crucially observes that EN does appear in informal English, describing it as a feature of ‘English parole’, referencing de Saussure’s well-known classification. Jin & Koenig (Reference Jin and Koenig2021) further support this claim through their corpus analysis of social media and networking websites:

In these instances, EN requires a c-commanding trigger, such as keep from, miss, avoid, among others. Additional verbs associated with EN include fear, regret and complain (Horn Reference Horn2010; Jin & Koenig Reference Jin and Koenig2021). In spoken English, EN can also be triggered by comparatives, as noted by Jin & Koenig (Reference Jin and Koenig2021: 69):

Thus, EN is indeed attested in spoken English. However, this raises a theoretical challenge: EN can only occur when the negative marker acts as a syntactic head, while it is unavailable when the negative marker is a maximal projection (Greco Reference Greco2022).Footnote ³ English negation is generally classified as a maximal projection, which should preclude its ability to function as EN. Yet empirical evidence from Horn (Horn Reference Horn2010: 125) and Jin & Koenig (Reference Jin and Koenig2021), as well as not-ACC-ing examples themselves, contradict this expectation. This classification of English negation as a maximal projection stems from its lack of characteristics typically associated with head-like negation, such as licensing Negative Concord (NC) (Zanuttini Reference Zanuttini1991; Haegeman Reference Haegeman1995; Zeijlstra Reference Zeijlstra2004; Giannakidou Reference Giannakidou, von Heusinger, Maienborn and Portner2011; Chierchia Reference Chierchia2013), (25a) and enabling why-not questions (25b):

Let us examine the latter phenomenon. According to Merchant (Reference Merchant2001), the English negative marker can adjoin to why because why is an XP. For this to occur, the negative marker must also be an XP. In contrast, when negation functions as a head, such as the Italian non, it cannot combine with why, rendering the Italian equivalent, perché non?, ungrammatical. Since English not is classified as an XP, occurrences of expletive EN in English are unexpected given the head requirement discussed earlier. This same reasoning extends to the absence of the NC phenomenon in (formal) SE.

Still, empirical evidence, such as the not-ACC-ing constructions discussed in this work, challenges the expectation that English cannot host EN. We (tentatively) suggest that this empirical puzzle could be better understood in light of variation across English varieties. In particular, the availability of EN might reflect influence from varieties such as African American English (AAE), where negation arguably behaves more like a syntactic head and licenses NC (Green Reference Green2009) (26). AAE is also individuated by Morris (Reference Morris2021) as the language variety where the first attestations of not-ACC-ing constructions emerged (see (27) for a relevant example from a social-media post on X):

While a thorough sociolinguistic or contact-based analysis is beyond the scope of this article, we note that addressing this theoretical puzzle will require further research in that direction. Specifically, we speculate that the presence of EN in informal English may reflect the influence of varieties in which negation has a different syntactic status. If this is correct, then the syntactic behavior of not in English may be more variable than previously assumed, functioning as a head in certain constructions, despite its canonical analysis as a maximal projection.

This variation may also be shaped by broader diachronic dynamics, including the Spec-to-Head and Head Preference Principles (Van Gelderen Reference Van Gelderen2004).Footnote ⁵ In this respect, work by Blanchette (Reference Blanchette2017) and Blanchette & Lukyanenko (Reference Blanchette and Lukyanenko2019) shows that NC, though ungrammatical in formal Standard English, is efficiently processed in informal speech, and that social stigma influences acceptability judgments. These findings suggest that a more nuanced account of negation across English varieties could help clarify the syntactic status of EN in constructions like not-ACC-ing. We leave this issue open for future research.

Now, we wish to clarify our use of the term expletive in relation to negation. Our focus is specifically on negation and not-ACC-ing constructions (for an overview on expletiveness, see Tsiakmakis & Espinal (Reference Tsiakmakis and Espinal2022), who discuss also other expletive elements, e.g. expletive subjects). Nonetheless, in line with previous research (Wood Reference Wood, Zanuttini and Horn2014; Delfitto et al. Reference Delfitto, Melloni and Vender2019; Greco Reference Greco2019, Reference Greco2020), we adopt the view that expletiveness does not equate to semantic vacuity. We argue that the negative marker not in not-ACC-ing constructions continues to function as an operator and is not semantically vacuous. Rather than reversing truth conditions on the propositional content, it operates on the presuppositional layer, generating interpretive effects related to irony and unexpectedness (see section 4).

2.2. ACC-ing constructions

As the label suggests, not-ACC-ing constructions appear closely related to ACC-ing constructions. ACC-ing constructions are characterized by a subject in the accusative case followed by a gerund (Stowell Reference Stowell1982; Reuland Reference Reuland1983; Matsuoka Reference Matsuoka1994; Lowe Reference Lowe2019) and an optional adverb. The main superficial difference between an ACC-ing construction and a not-ACC-ing construction is the presence in the latter of an obligatory negative marker preceding the accusative subject:

ACC-ing constructions are argued to be clausal, as they permit adverbial modification and always require a subject (Lowe Reference Lowe2019: 318). That is evident in the following sentences, where there and it must be present in the structure:

The clausal nature of ACC-ing constructions becomes more evident when compared to other gerundive constructions, which are nominal in nature. Consider, for example, Poss-ing constructions (33) and NP-ing constructions (34):

Poss-ing constructions feature subjects in the genitive case, which distinguishes them from ACC-ing constructions. NP-ing constructions, on the other hand, represent the canonical case of a nominal gerund. Additional evidence for the clausal nature of ACC-ing constructions is seen in the following contrast: when two ACC-ing constructions are conjoined, they do not trigger agreement of the main verb (examples (35) and (36) below, in Reuland Reference Reuland1983: 107), unlike Poss-ing constructions (37) and NP-ing constructions (38):

Structurally, it has been argued in the literature that gerunds, including ACC-ing constructions, are reduced structures. Stowell states that gerunds ‘lack the COMP position entirely’ (Stowell Reference Stowell1982: 561; cf. also Abe Reference Abe1986; Yamada Reference Yamada1987; Matsuoka Reference Matsuoka1994). Let us examine the evidence supporting this claim. One key aspect is that gerunds/ACC-ing constructions cannot cooccur with a complementizer like to (Rosenbaum Reference Rosenbaum1967) or that, unlike infinitives:

Additionally, a gerund/ACC-ing construction cannot serve as a [+wh] complement to a c-commanding verb or preposition:

On these grounds, we adopt Stowell’s view that gerunds lack the COMP position. In more recent, cartographic (Rizzi Reference Rizzi and Haegeman1997; Belletti Reference Belletti and Rizzi2004; Cinque & Rizzi Reference Cinque, Rizzi, Heine and Narrog2012) terms, gerunds and ACC-ing constructions are reduced structures lacking the whole LP (Rizzi Reference Rizzi and Haegeman1997; Rizzi & Bocci Reference Rizzi, Bocci, Everaert and van Riemsdijk2017).

Crucially, not-ACC-ing constructions are not simply the negated counterparts of ACC-ing constructions; rather, they represent a distinct syntactic configuration, as we will discuss in more detail in section 4. For now, let us anticipate this point by considering the behavior and distribution of negation in contrasts such as (41) and (42):

In (41), an ACC-ing construction appears as the complement of the preposition of, with negation following the accusative subject. In (42), a not-ACC-ing construction is used in the same syntactic position, but crucially negation precedes the accusative subject, as is typical of not-ACC-ing configurations. This difference in the linear and structural position of negation reveals a deeper distinction between the two constructions. Note also that (42) is ungrammatical: we return to the root nature of not-ACC-ing constructions in section 3.2. The difference in interpretation and acceptability further supports the view that negation occupies distinct syntactic positions in the two structures, and that not-ACC-ing constructions cannot be reduced to mere negated ACC-ing clauses.

3.1. Expletive Negation in not-ACC-ing constructions

Not-ACC-ing constructions respond positively to all the relevant tests for EN. As noticed by Pereira (Reference Pereira2023), not-ACC-ing constructions are incompatible with strong NPIs such as either, at all, yet, anywhere (Huddleston & Pullum et al. Reference Huddleston and Pullum2002). Crucially, they may cooccur with PPIs, hence elements such as too, already and somewhere, which may be licensed only in positive environments (Espinal Reference Espinal1997; Israel Reference Israel1997; Huddleston & Pullum et al. Reference Huddleston and Pullum2002; Giannakidou Reference Giannakidou, von Heusinger, Maienborn and Portner2011):

A further crucial aspect for the purposes of this work is the following. An additional negative marker may appear after the subject, and when that happens, it can license a NPI (Pereira Reference Pereira2023):

This is important because it shows that the first negative marker cannot license NPIs, therefore it does not scope over the rest of the sentence. The second one, instead, licenses NPIs and functions as SN. This is a further distinction which supports the view that the first negative marker (the one which is invariably present in not-ACC-ing constructions) instantiates a EN rather than SN. We will come back to this crucial point in section 4, where we capitalize on this in our structural analysis of not-ACC-ing constructions.

As anticipated in section 2.1 on EN, we use the traditional term expletive to refer to the ‘special’ kind of negation which does not reverse the truth-value conditions of the sentence (Horn Reference Horn1989; Horn & Katō Reference Horn and Katō2000), but rather generates interpretive effects tied to irony and unexpectedness. In this sense, expletiveness is not tantamount to semantic vacuity (Wood Reference Wood, Zanuttini and Horn2014; Zanuttini et al. Reference Zanuttini, Wood, Zentz and Horn2018; Delfitto et al. Reference Delfitto, Melloni and Vender2019; Greco Reference Greco2019, Reference Greco2020). Consider the following sentences, which illustrate that EN is not semantically empty: the distinctive ironic interpretation of a not-ACC-ing construction vanishes when the negative marker is omitted from the corresponding non-negated sentence in (46). Note also that (46b) would be grammatical only if used as a caption of a social-media post (a typical environment for a not-ACC-ing construction):

3.2. Not-ACC-ing constructions convey new information

Not-ACC-ing constructions are used in isolation in social-media contexts, where users post content spontaneously and out of the blue, thereby conveying entirely new information. Crucially, this is why not-ACC-ing constructions are also acceptable as responses to questions like ‘What is happening?’ Indeed, on X (formerly Twitter), the prompt ‘What is happening?’ appears in the box where users compose their posts.

3.3. Not-ACC-ing constructions are a root phenomenon

Not-ACC-ing constructions are a strictly root phenomenon, instantiating a special case of EN, one in which the presence of negation does not require syntactic licensing. This contrasts with other instances of EN in English, where negation typically depends on a c-commanding trigger for proper licensing (section 2.1; cf. Horn (Reference Horn2010) and Jin & Koenig (Reference Jin and Koenig2021)). Their superficial structural make-up is always the following: negative marker + subject in the accusative case + gerund, thus not allowing any licensor.

Importantly, notice also how not-ACC-ing constructions cannot function as sentential subjects (49), nor be the complement of a preposition such as of (50):

3.4. Not-ACC-ing constructions and left-peripheral elements

A distinctive feature of not-ACC-ing constructions is their interaction with left-peripheral elements and discourse-related constituents. They cannot host wh-elements (51a), nor why (51b), nor focalized constituent (51c), while topics are admitted (51d):

Moreover, not-ACC-ing constructions seem to exhibit a distinct prosodic contour, which appears to be similar to that of focalized constructions, though not identical (Pereira Reference Pereira2023). Intonation rises when uttering not me, then the rest of the sentence is prosodically flatter (see section 4.2 and footnote 12).

3.5. Not-ACC-ing constructions and adverbial modification

Not-ACC-ing constructions present an asymmetry when it comes to preposed adverbs. Adverbs like stupidly and sadly cannot precede the negative marker (52a, b), while no problem arises if adverbs follow the subject (52c, d):

The only adverbs which may be preposed and precede the negative marker are definitely and totally, and here lies the asymmetryFootnote ⁸ (ex. (12) is repeated below in (53)):

4.1. Excluding possible analyses: not-ACC-ing constructions are neither pure ACC-ing constructions, nor (reduced) clefts

As we have discussed, it is interesting to notice that English not-ACC-ing constructions are intuitively related to ACC-ing constructions.

Let us consider the parallelism between them. Both share a restriction to the non-finite, gerundive form of the verb with an -ing suffix. Additionally, neither displays a subject in the nominative case, relying instead on an accusative form. At first glance, the primary distinction between them is the presence or absence of the negative marker. However, they differ in significant respects. As discussed in section 3.3, not-ACC-ing constructions are a root phenomenon, rendering them ungrammatical when embedded. They cannot function as sentential subjects or as the complement of a preposition like of. In contrast, these restrictions do not apply to ACC-ing constructions. Unlike not-ACC-ing constructions, ACC-ing constructions cannot occur in isolation; they most commonly appear as sentential subjects or as complements of a higher c-commanding element, such as a preposition:

Therefore, not-ACC-ing constructions cannot be regarded as pure ACC-ing constructions. An alternative analysis might consider not-ACC-ing constructions as instances of reduced clefts. However, as we will show, this possibility must also be excluded.

Intuitively, a sentence like (57a) could resemble a reduced version of the cleft sentence in (57b):

It is important to note, however, that negation serves its usual function in full cleft or relative sentences, reversing the truth-conditions of the sentence. This contrasts with not-ACC-ing constructions, which are obligatorily affirmative despite the presence of negation. In (57a), the negative marker instantiates EN, whereas in (57b), it functions as SN. In (57a), the speaker expresses irony, sarcasm and unexpectedness about taking the train early in the morning, and there is a veridicality inference (Pereira Reference Pereira2023) according to which the speaker indeed takes that train. In other words, in (57b) the speaker does not take the train, and the utterance lacks the ironic flavor.

It is true that, on the one hand, one could mark the sentence in (57b) pragmatically, conveying the opposite of what is uttered in an ironic way:

However, on the other hand, there is no licit way in which a not-ACC-ing construction could be uttered with a different prosody such that its truth conditions are reversed. Crucially, only the affirmative interpretation (with an EN reading) is available.

Further evidence that distinguishes not-ACC-ing constructions from clefts comes from a syntactic contrast, and it regards embedding. Not-ACC-ing constructions are always root (see section 3.3), while clefts may be embedded. Since clefts involve a finite verbal form with progressive morphology, they can be embedded under predicates like believe. This is not the case for gerunds, especially those with an accusative subject (i.e. ACC-ing constructions, which are central to this discussion). As we have discussed, gerunds may function as sentential subjects or complements of prepositions like of. Therefore, in the examples below, we will see that clefts can be embedded, while not-ACC-ing constructions are always root clauses. To illustrate this, the type of higher clause must differ.Footnote ¹⁰ Specifically, (59b) is ungrammatical in English, but this is because the verb believe only takes finite clausal complements. The ungrammaticality here arises from this restriction, not from the root nature of the not-ACC-ing constructions in brackets, which is more clearly visible in (59c):

Based on the contrasts explored in this section, not-ACC-ing constructions cannot be reduced to similar structures. We propose that they should be regarded as a distinct, specific syntactic phenomenon.

4.2. Proposal: focalization of the ACC-ing construction through Expletive Negation

To present our proposal for the structural analysis of not-ACC-ing constructions, we address the following questions, which will guide our argumentation:

1. I. Where and how is EN represented in the structure of not-ACC-ing constructions?

2. II. Where does the non-embeddability of not-ACC-ing constructions come from?

3. III. How does the subject of a not-ACC-ing construction obtain accusative case?

4. IV. How is their peculiar interpretation obtained?

In a nutshell, we argue that not-ACC-ing constructions are instances of negated, focalized ACC-ing constructions (AspPsFootnote ¹¹), with an EN reading arising from negation being merged in the LP (Greco Reference Greco2020) (60)–(61):

Regarding question I, we argue that Greco’s (Reference Greco2019, Reference Greco2020) analysis of EN is a fitting approach. In his account, the negative marker is merged in the LP, rather than in the TP area as it is in standard SN constructions (Belletti Reference Belletti1990; Laka Reference Laka1990; Zanuttini Reference Zanuttini, Adriana and Luigi1996, Reference Zanuttini1997, Reference Zanuttini, Mark and Chris2001). This allows the negative marker in the CP to be ‘blind’ to the lower v*P phase (Chomsky Reference Chomsky and Kenstowicz2001), which is closed and impenetrable to further syntactic operations (Phase Impenetrability Condition). In other words, the negative marker is merged too high and cannot ‘see’ into the lower phase, which is closed. This explains the EN reading and the incompatibility of not-ACC-ing constructions with NPIs such as at all and their cooccurrence with PPIs like already. According to Giannakidou (Reference Giannakidou1997, Reference Giannakidou2000) and Zeijlstra (Reference Zeijlstra2004), PIs are elements that contain a dependent variable requiring licensing by a negative (or nonveridical) operator. This operator must be able to bind all such free variables in the vP domain to license them. If no appropriate operator is available – e.g. because the v*P has already been closed off when negation is merged – then this binding fails. As a result, structures involving unlicensed NPI variables are excluded, while PPIs are accepted (phases are underlined in (62)):

It has already been argued in the literature that the distinction between low and high negation in Modern English plays a crucial role in distinguishing certain independent constructions. In particular, Tubau (Reference Tubau2020), argues that the position of first-merged negation, whether in a TP-internal or TP-external position, causes contrasts in polarity-reversing question tags, neither/so-coordination, either/too adverbs and other constructions. Additionally, this distinction has been proposed as key to understanding the structure of yes-no questions and their response particles (see Holmberg Reference Holmberg2015; Wiltschko Reference Wiltschko, Bailey and Sheehan2017).

We also propose that in not-ACC-ing constructions, an entire ACC-ing construction is moved to the specifier of FocP. This accounts for their non-embeddability and new information interpretation. Let us explain why this is the case. As proposed in Greco (Reference Greco2020) for Italian Surprise Negation Sentences (Snegs; see example where negation is expletive, the entire TP mi è scesa dal treno Maria is focalized and moves to the specifier of FocP, resulting in the new information interpretation and resistance to embedding.

We argue that the same mechanism applies to not-ACC-ing constructions. ACC-ing constructions are reduced structures that lack the LP (Stowell Reference Stowell1982; Reuland Reference Reuland1983; Yamada Reference Yamada1987; Matsuoka Reference Matsuoka1994). As a result, apart from the gerundive verb form, an ACC-ing construction is structurally equivalent to the TP moved in Snegs. Consequently, the same mechanism governing Snegs applies to not-ACC-ing constructions: a TP is moved to the LP of the clause, where it is focalized by landing in the specifier of FocP. The key difference is that, in Snegs, a finite TP is moved, whereas in not-ACC-ing constructions, it is an ACC-ing construction that undergoes this movement. Focalization yields the new information interpretation, as in Snegs. Crucially, the ACC-ing construction moves as a single structural unit, which accounts for the incompatibility with embedding (addressing question II): as shown by Greco’s (Reference Greco2020) for Italian Surprise Negation Sentences, natural languages seems to disallow the embedding of big structural chunks.

Let us examine these claims in detail. Among the benefits of this proposal, consider, the new information interpretation that not-ACC-ing constructions carry. We recall here that not-ACC-ing constructions can be served as answers to propositional questions (64), while they cannot be employed as answers to entity questions (65):

According to Alonso-Ovalle & Guerzoni (Reference Alonso-Ovalle, Guerzoni, Arregi, Chang, Fagyal, Merchant and Takahashi2004) and Brunetti (Reference Brunetti2004), we assume that what carries new information, at least in answers, lays in the focus domain of a sentence and the rest of the sentence is deleted via ellipsis. The fact that not-ACC-ing constructions can only serve as answers to propositional questions suggest that the entire ACC-ing construction is the focalized part, as we predict in our analysis. Moreover, if the ACC-ing construction realizes the focus of the sentence, we should expect it to influence the sentence prosody. This expectation aligns with findings from the literature, which indicate that focus always induces some suprasegmental phonetics changes (Bocci Reference Bocci2007, Reference Bocci2008; Frascarelli Reference Frascarelli2000). Specifically, focus is associated with a marked tonal rise (L+H* in ToBi transcription). Importantly, this appears to be the case in not-ACC-ing constructions (section 4.4), which exhibit a rising intonation at the beginning of the utterance, followed by a flatter prosodic pattern (Pereira Reference Pereira2023). However, a more systematic investigation into the prosody of these sentences is required to substantiate this observation further.Footnote ¹²

As per question IV, we argue that the special interpretation of not-ACC-ing constructions arises from the focalization of a reduced ACC-ing constituent under EN. EN targets the presuppositional layer introduced by focus, rather than the truth-conditional content of the clause (Abels Reference Abels2005, 1996; Delfitto et al. Reference Delfitto, Melloni and Vender2019; Greco Reference Greco2020). As a result, the proposition is interpreted as true but unexpected or undesirable, which explains both the inference of veridicality and the absence of NPI licensing: EN does not create a downward-entailing environment.

Following Rooth (Reference Rooth, Féry and Ishihara2016), the presence of focus introduces a set of alternatives. The role of EN is to assert that p (the focalized proposition) does not belong to the speaker’s expectation set (i.e. the set of likely or assumed-to-be-true propositions). Crucially, this does not mean that ¬p is asserted. Rather, p is assumed to be true (hence the veridicality inference), but the speaker expresses a negative attitude toward that truth; e.g. surprise, annoyance, or undesirability.

Structurally, this is ensured by EN being merged high, after the previous v*P phase is closed, so it does not affect the propositional content directly, but only the attitude the speaker wants to convey toward the proposition. This explains both why EN does not license NPIs (as it does not create a downward-entailing environment) and why the proposition is interpreted as true.

Moreover, the schema in (60) also justifies, at least, two more facts: (i) not-ACC-ing constructions are compatible with topicalized constituents while resisting focalization; (ii) they cannot host wh-elements. We argue that all these restrictions are accounted for by a unique assumption: the focalization of the ACC-ing construction. Starting from the first restriction, if the ACC-ing construction moves to the focalized phrase, it leaves no space for other focalized phrases, since only one functional projection for focalization is available in the LP (Rizzi Reference Rizzi and Haegeman1997) and it is already occupied. The same holds for the incompatibility with wh-elements, since they compete for the same SpecFocP position, according to the original proposal in Rizzi (Reference Rizzi and Haegeman1997). On the other hand, no restriction is found for topicalized items, and, therefore, not-ACC-ing constructions can host them. We have discussed how ACC-ing constructions are reduced structures that lack the LP. Therefore, the Finiteness position, responsible for Nominative Case assignment to the subject, is missing (building on Phase Theory; Richards Reference Richards2007; Mackenzie Reference Mackenzie2018). But then, why is accusative case employed instead (question III)? We argue that not-ACC-ing constructions display accusative subjects since accusative case is the default case in English (Schütze Reference Schütze2001). Since the ACC-ing construction is defective in the LP, where the Finiteness position is found, it follows that there is no case assigner from which the subject can receive Nominative Case. We adopt Schütze’s (Reference Schütze2001) perspective on default case, which arises when the mechanisms of case assignment break down. In such instances, constituents are spelled out in the default case, as occurs, for example, when no case assigner is present or when the constituent has nothing to match with. In English, default case is accusative, as is visible in several examples discussed by Schütze (Reference Schütze2001: 211). We report two of them here:

In (66), Me and I are not arguments and there are no obvious case assigners for them. The same applies to answers such as (67a). Observe that when a finite verb is present, the subject becomes nominative, as in (67b). The same, we argue, applies to ACC-ing constructions. Since they are reduced (Stowell Reference Stowell1982; Reuland Reference Reuland1983; Matsuoka Reference Matsuoka1994; Lowe Reference Lowe2019), Finiteness is absent, and cannot assign nominative case to the subject, which is spelled out as accusative. That is the reason why English not-ACC-ing constructions display accusative subjects. Accusative case being the default case in English is supported by acquisition data as well. In child English, reduced structures like root infinitives (Rizzi Reference Rizzi1993, Reference Rizzi, Deen, Nomura, Schulz and Schwartz2006) often cooccur with accusative subjects (Schütze & Wexler Reference Schütze, Wexler, Stringfellow, Cahana-Amitay, Hughes and Zukowski1996), where the adult target grammar would require nominative case instead.Footnote ¹³

To take stock, we have argued that not-ACC-ing constructions are their own syntactic construction formed by a negated, focalized ACC-ing formation with an EN reading given by negation being merged in the LP.

Now, another issue arises, tied to question IV. Since Belletti (Reference Belletti and Rizzi2004), focalized constituents have been noticed to target both the LP and low IP area (with different interpretive restrictions), and so the question: is the ACC-ing construction focalized in the LP or in the low IP area? We argue for the former on the basis of the expletiveness of the negative marker not. Expletiveness could be not obtained if the negative marker stayed in the low IP area. The v*P phase would still be open and only the SN reading would be accessible. Independent evidence come from the contrast regarding preposed adverbs we mentioned in section 3.5: adverbs can only follow the subject, never precede it.Footnote ¹⁴

Following Rizzi & Bocci’s (Reference Rizzi, Bocci, Everaert and van Riemsdijk2017) account of the LP, preposed adverbs target the Modifier (Mod) position in the LP. Observe the sequence of positions in the LP as shown by these authors:

Notice how the Mod position is lower than the Focus position. Recall that in the present analysis, the whole ACC-ing construction is focalized in SpecFocP, i.e. in a structurally higher position than the Mod position. From our proposal, the contrast in (68) follows straightforwardly: when an adverb is placed inside the ACC-ing construction, after the subject, the sentence is grammatical; on the contrary, when an adverb is placed before the ACC-ing construction, it cannot find a suitable slot, as the Mod position is lower than the focalized ACC-ing formation (see, however, footnote 9 on definitely and totally).

This analysis also has the advantage of accounting for all syntactic and interpretive properties of not-ACC-ing constructions, deriving them solely from two mechanisms: high negation in the LP and the focalization of the ACC-ing construction.