ON THE SYNTAX OF OUGHT-TO-BE CONSTRUCTIONS WITH AND WITHOUT A SPECIFIC ADDRESSEE

ABSTRACT

This paper focuses on the (ought-to-be) deontic, which is analyzed from the concepts of obligation discussed in Castañeda (1970), Feldman (1986) and Hacquard (2006, 2010). From co-occurrence tests of an ought-to-be modal with other elements in the CP layer – in particular, with the wh-phrase por que ( why ) and with adverbial elements, it was found that ought-to-be is interpreted between the high projections IntP and ModP. It was also shown that an ought-to-be can be used in an utterance with or without a specific addressee and that this difference reflects in the syntax. The proposal is to extend the analysis by Portner, Pak and Zanuttini (2019), developed for Korean speech style particles, to the ought-to-be deontic, postulating that this deontic, when used with a specific addressee, integrates a structure in which the cP category is projected, which allows features that express the speaker-addressee relation, such as status, to be checked. Consequently, this modal is restricted to the matrix domain. When, however, it appears in an utterance without a specific addressee, it integrates a structure without the projection of cP, since it does not refer to the speaker-addressee relationship. In this case, it can appear in both the matrix and embedded domains.

ought-to-be deontic modality; speaker-addressee relationship; cP category

RESUMO

O presente artigo tem como foco o deôntico ought-to-be , analisado a partir dos conceitos de obrigação discutidos em Castañeda (1970), Feldman (1986) e Hacquard (2006, 2010). A partir de testes de co-ocorrência de um modal ought-to-be com elementos que ocorrem na camada CP – em particular, com a expressão-wh por que e com elementos adverbiais –, constatamos que o ought-to-be é interpretado entre as posições altas de IntP e ModP. Mostramos também que um ought-to-be pode figurar em construções com ou sem addressee específico e que essa diferença se reflete na sintaxe. Nossa proposta é estender a análise de Portner, Pak e Zanuttini (2019), desenvolvida para partículas de estilo de fala do coreano, para o deôntico ought-to-be , postulando que esse deôntico, quando usado com um addressee específico, integra uma estrutura em que a categoria cP é projetada, o que permite que traços que expressam a relação falante-interlocutor, tais como o traço de status , sejam checados. Consequentemente, esse modal é restrito ao domínio matriz. Por outro lado, quando ele ocorre em uma sentença sem addressee específico, ele integra uma estrutura sem a projeção de cP, uma vez que ele não remete à relação falante-interlocutor. Nesse caso, o modal aparece em domínios matrizes e encaixados.

deôntico ought-to-be; modalidade deôntica; relação falante-interlocutor; categoria cP

Introduction

According to Rech, Soares e Guesser (2019) , Castañeda’s (1970) and Feldman’s (1986) concept of an ought-to-be obligation differs from Hacquard’s (2006 , 2010 ). The former associate ought-to-be with a type of obligation that conveys the idea of what something is supposed to be like or how it is supposed to happen, without a particular agent responsible for the event. This definition contrasts with the ought-to-do deontic, which links an event to a participant, putting the obligation to perform the event on a specific agent. Hacquard (2006 , 2010 ) also uses this distinction, associating the differences between these two types of obligation to the position of the modal in the sentence: the ought-to-be deontic is interpreted in a high position, above the categories of Tense and Aspect, while ought-to-do is interpreted in a low position, under Tense and almost all the categories that indicate Aspect. Unlike Castañeda and Feldman’s definition, Hacquard’s ought-to-be deontic corresponds to a directive speech act, which puts on the addressee the obligation to perform the event described in the sentence.¹ Considering both definitions, sentence (1a) and the one transcribed in (1b) both describe an ought-to-be obligation:

• (1) a. There ought to be a more equal distribution of wealth in the world.

• ( FELDMAN, 1986 , p. 179)

• b. Kitty has to brush her teeth.

• ( HACQUARD, 2006 , p. 40)

Feldman notes that, in a sentence like (1a), a more equal distribution of wealth in the world is an ought-to-be obligation, as it does not fall on any particular agent. According to Hacquard, an ought-to-be obligation is an addressee-oriented directive speech act. Therefore, (1b) illustrates this type of obligation if it is used in a context in which the speaker (Kitty’s mother, for example) puts the obligation of the event directly on the interlocutor, who may be the babysitter.

Based on previous studies, it can be assumed that the ought-to-be deontic — according not only to Castañeda (1970) and Feldman (1986) , but also to Hacquard (2006) — corresponds to a modal in a high position ( HACQUARD, 2006 , 2010 ; TSAI, 2015 ; RECH; VARASCHIN, 2018a , 2018b ; RECH; SOARES; GUESSER, 2019 ). The difference between these concepts is based on the orientation of this modal, which can be used in contexts with a generic addressee or no addressee at all, as in (1a), or in contexts with a specific addressee, as in (1b).

The next section shows how the ought-to-be deontic which is addressee-specific, and those with a generic or no addressee react to phenomena such as actuality entailments and (non-)occurrence in the embedded domain . After that, it will be argued that the structure of sentences with the deontic auxiliary modal ought-to-be can be different, based on whether or not they are addressee-specific: if so, there is a cP projection, whose head contains features that refer to the interaction between the speaker and their interlocutor ( PORTNER; PAK; ZANUTTINI, 2019 ); if not, this category is not part of the structure.

The ought-to-be deontic: generic or non-existent addressee vs. specific addressee

According to the literature ( BRENNAN, 1993 ; HACQUARD, 2006 , 2010 ; TSAI, 2015 ; RECH; VARASCHIN, 2018a , 2018b ; RECH; SOARES; GUESSER, 2019 ), the ought-to-be deontic is interpreted in a high position, while the ought-to-do deontic is interpreted in a low position. The main reasons for this distinction are the differences in how these types of deontic modals are related to tense and aspect, as well as their orientation: whether the obligation is put on the participant of the event described by vP, preferably the subject ( ought-to-do ) or on a salient participant in the speech event (ought-to-be).

This article focuses on the deontic modal interpreted in a high position: the ought-to-be . We will show how this deontic modal behaves in relation to actuality entailments and (non-)occurrence of the modal in the embedded domain. We will consider the deontic when there is a specific addressee and when the addressee is generic or non-existent.

Actuality entailments in deontic constructions

Actuality entailments are presented by Tsai (2015) as a consequence of the analysis of modals. They refer to the fact that, when in the perfective form, certain modals cause an entailment that the event described under its scope really happened. This phenomenon has been previously discussed by Bhatt (1999) and Hacquard (2006 , 2009 ), among others. These authors point out that actuality entailment occurs with low modals — dynamic modals and the ought-to-do deontic –, but does not occur with high modals, such as epistemic modals and the ought-to-be deontic.

In Hacquard’s (2006) analysis of French, actuality entailments are consequences of when the perfective aspect moves to a position above where dynamic and certain deontic modals are. Chinese does not use the perfective movement, but instead uses the insertion of an aspectual operator above the modal verb as a strategy for actuality entailments. Examples from Tsai (2015 , p. 290), using the deontic modal yao , have been transcribed in (2):

• (2) a. Akiu yao changchang xiuxi. [irrealis: necessity]

• Akiu YAO often rest

• ‘Akiu needs to rest often’

• b. Akiu changchang yao xiuxi. [realis: human need]

• Akiu often YAO rest

• ‘Akiu often needed to rest’

When yao precedes an adverb of frequency/aspectual operator, such as changchang ‘often’ (2a), the sentence presents an irrealis reading and the modal presents an imperfective reading. However, when yao follows the adverb changchang (2b), the interpretation is realis and the modal has a perfective reading. Analyzing the sentences in (2), Tsai notes that the phenomenon of actuality entailment appears in constructions with low modals, such as the ought-to-do deontic (2b), but not in constructions with high modals, such as the ought-to-be deontic (2a).

The difference in behavior between the ought-to-do deontic and ought-to-be regarding phenomena such as the actuality entailment led Tsai to propose the structure below, in which the ought-to-do deontic is inside the inflectional layer, whereas the ought-to-be deontic is placed by the projection of the epistemic modal in the left periphery of the sentence:

Image 1
The positions of the ought-to-do and ought-to-be deontics according to Tsai (2015)

From this perspective, Tsai formulates the following generalization: actuality entailments are restricted to modals which are below TP, that is, ability, volition, dispositional modals and the ought-to-do deontic. Epistemic and ought-to-be deontic modals, as they are above TP, in the left periphery, do not show actuality entailments.

The phenomenon of actuality entailment based on Hacquard (2006) and Tsai (2015) was shown here to argue that the ought-to-be deontic is interpreted in a high position of the structure, above the tense category.

However, it is not possible to test the actuality entailment in BP in constructions with the ought-to-be deontic without an addressee:

• (3) a. Tem que se fazer o controle das vacinas nos Postos de Saúde .

• Have to one control the vaccines in the health centers.

• ‘Vaccination at health centers ought to be controlled.’

• b. Teve que ser feito o controle das vacinas nos Postos de Saúde, #mas o controle não foi feito .

• Had to be done the control of vaccines in the Health Posts, #but the control was not done.

• ‘Vaccination at health centers ought to have been controlled, #but the control was not done.’

The pragmatic inconsistency of the continuation “mas o controle não foi feito/but the control was not done” shows that the actuality entailment emerges from (3b). However, it is important to note that only modals interpreted below tense can move to the TP category to merge with the verb’s inflection marks. The ought-to-be construction in (3a) becomes an ought-to-do construction when the modal ter que (ought to) bears perfective morphology, corresponding to a passive construction without the passive agent, someone like “nurses” ( Teve que ser feito o controle das vacinas pelos enfermeiros nos Postos de Saúde/ Vaccination at health centers ought to have been controlled by nurses). It seems to be the case that it is not possible to test the actuality entailment phenomenon in BP putting the perfective morphology on the modal. Tsai, however, tests this phenomenon by the combination of the ought-to-be deontic with high adverbs, like changchang (often), see the example in (2).

When the ought-to-be deontic denotes a speaker-addressee relationship, corresponding to a directive speech act, it is also impossible to test its reaction to the phenomenon of actuality entailment. Consider the following examples:

• (4) a. As crianças têm que/devem receber as provas corrigidas .

• The children have to/must receive the tests corrected.

• ‘The children ought to receive their tests corrected.’

• b. As crianças tiveram que receber as provas corrigidas .

• The children had to receive the tests corrected.

• ‘The children ought to have received their tests corrected.’

In sentence (4a), the speaker puts on the addressee - who might be the teacher - the obligation to give the children their tests corrected. This sentence, therefore, corresponds to a directive speech act. In (4b), the auxiliary ought to is in the perfective form; consequently, the ought-to-be reading is blocked. In this case, the modal is used for reporting an order or a necessity, so the modal reading is the ought-to-do type, causing an actuality entailment. Therefore, only high modals do not yield actuality entailments. The performative character of the addressee-specific ought-to-be deontic, in (4a), cannot occur in a perfective form since it makes no sense to give an order to someone to perform an event in the past.

In sum, as observed in the literature ( HACQUARD, 2006 , 2009 ; TSAI, 2015 ), modals interpreted in a high position (the ought-to-be deontic and epistemic modals) do not generate actuality entailment, which therefore makes them different from modals interpreted in a low position (the ought-to-do deontic and dynamic modals).

The ought-to-be deontic in the matrix and/or the embedded domain

As noted in the literature, there are elements restricted to the matrix domain (root phenomenon) , while others move between the matrix and the embedded domain (see HAEGEMAN, 2004 ; PORTNER; PAK; ZANUTTINI, 2019 ). Portner, Pak and Zanuttini (2019) relate this distribution for the item’s orientation: elements which express the speaker-addressee social relationship are oriented towards the moment of speech and are not part of the embedded domain; on the other hand, elements that do not capture this relationship may occur both in the matrix and in the embedded domain. As evidence to this proposition, the authors show the syntactic distribution of speech-style particles, interrogative particles, politeness particles, among others. In (5) and (6) below, examples of the use of a speech-style particle in Korean, transcribed by Portner, Pak and Zanuttini (2019 , p. 3), are shown:

• (5) Ecey pi-ka o-ass- supnita .

• yesterday rain-nom come-pst-decl.form

• ‘It rained yesterday’

• (6) *Inho-ka [ecey pi-ka o-ass- supnita -ko] malhayss-supnita.

• Inho-nom [yesterday rain-nom come-pst-decl.form-comp] said-decl.form

• ‘Inho said that it rained yesterday’

In Korean, there are final sentence particles, called speech-style particles , which encode information about the speaker and the interlocutor, as well as about the level of formality of the situation in which the communication occurs. In the examples above, the particle supnita is used, marking the situation as formal, the interlocutor as socially superior to (or older than) the speaker, and also identifies the type of sentence (declarative). The ungrammaticality of (6) results from the use of this particle in the embedded domain, which points in the direction that elements that encode information that is given in the enunciation context – such as the relationship between the interlocutors and the level of formality of the situation – are restricted to the matrix domain.

This phenomenon can also be verified in Japanese, in the syntactic distribution of the particle -mas , as the examples below show ( PORTNER; PAK; ZANUTTINI, 2019 , p. 4):

• (7) Peter-wa sushi-o tabe- mas -i-ta.

• Peter-top sushi-acc eat-mas-decl.pst

• ‘Peter ate sushi’

• (8) Hanako-wa [dare-ga kuru/*ki- mas -u ka] sitte i-mas-u.

• Hanako-top [who-nom come/come-mas-prs q] know mas-prs

• ‘Hanako knows who is coming’

The particle -mas is a politeness marker that the speaker uses when addressing a specific interlocutor in a slightly more formal communication situation; therefore, it encodes information that captures the speaker-addressee relationship. In this sense, the ungrammaticality of its use in (8) is further evidence that elements that capture information of this nature do not occur in the embedded domain.

A last example from Portner, Pak and Zanuttini (2019) is presented in order to illustrate the relationship between particle distribution in the matrix and/or embedded domain and the information they encode. The following examples show the contexts in which two interrogative particles are used in Korean (-nya and -ni):

• (9) a. Onul nalssi-ka way ilehkey coh- nya ? (talking to oneself)

• today weather-nom why like.this good-int.pln

• ‘Why is the weather so good today?’

• b. Changco- nya , cinhwa- nya ? (in writing)

• creation-int.pln evolution-int.pln

• ‘Is it creation or evolution?’

• (10) Choysen-ul ta ha-ass- ni ? (to an interlocutor)

• best-acc all do-pst-int.pln

• ‘Did you do your best?’

According to the authors, the interrogative particle -nya can appear in the embedded domain, while the particle -ni cannot. It is interesting to note, in the examples above, that -nya occurs in sentences without a specific addressee, as in (9a) and (9b), whereas -ni is used in sentences that directly address an interlocutor, such as (10). These examples corroborate the authors’ claim that elements used in sentences with a specific addressee encode information about the speaker-addressee relationship.

As it has been argued throughout this article, the ought-to-be deontic can be used in contexts without an addressee, with a generic addressee, and also with a specific addressee. Based on Portner, Pak and Zanuttini (2019) and the particles used in examples (5) to (10), it is expected that a modal such as the ought-to-be deontic should present a different syntactic distribution across languages, depending on whether the sentence has a specific addressee or not, assuming that this proposition is valid besides Korean and Japanese. We therefore expect this syntactic difference to be observed not only in BP, as we show in this paper, but in all languages, considering, in line with the Cartographic Approach ( CINQUE; RIZZI, 2008 ), that, whenever the existence of a functional head is proposed in a given language, it should occur crosslinguistically. In (11), the syntactic distribution of an ought-to-be without a specific addressee is shown:

• (11) a. Tem que/Deve haver medidas contra a violência doméstica .

• Have to/Must have measures against the domestic violence.

• ‘There ought to be measures against domestic violence’

• b. Tem que/Deve haver medidas que protejam as mulheres em situação de violência doméstica .

• Have to/Must have measures that protect the women in situation of domestic violence.

• ‘There ought to be measures that protect women in a domestic violence situation’

• c. Muitos concordam que tem que/deve haver medidas protetivas à mulher em situação de violência doméstica .

• Many agree that has to/must have protective measures to the woman in domestic violence situations.

• ‘Many agree that there ought to be protective measures for women in domestic violence situations’

The good formation of the sentences above shows that ter que/dever meaning ought-to-be can occur both in the matrix (11b) and the embedded domain (11c). This result was expected, because the modal in these sentences is not directed towards a specific addressee.

The restriction to the embedded domain is related to the use of a particular linguistic element that expresses the speaker-addressee relationship. For this reason, an ought-to-be directed to the addressee does not occur in the embedded domain, as illustrated by the example below:

• (12) a. O traficante tem que/deve morrer nessa operação .

• The dealer has to/must die in this operation

• ‘The dealer/trafficker ought to die in this operation’

• b. O comandante disse que o traficante tem que/deve morrer nessa operação .

• The commander said the dealer has to/must die in this operation.

• ‘The commander said that the dealer ought to die in this operation’

In (12a), the modal ter que/dever can acquire a deontic reading (apart from that of wish); in this case, the sentence corresponds to a directive speech act, and the obligation falls on the addressee. Considering this use, it is assumed that the factors pointed out by Portner, Pak and Zanuttini (2019) , which refer to the communication situation and the status of the interlocutors ([S<A, S≤ A, S=A, S≥A, S>A], in which S means speaker and A means addressee), become relevant. The hypothesis presented here is that an ought-to-be deontic that corresponds to a performative act needs to be checked for the feature status of the interlocutors, like the particles -ni and -supnita from Korean, and the particle -mas from Japanese (see examples (5) to (10) above). As previously noted, these authors state that the features situation and status are in the head of the cP category, which encodes contextual factors. They also note that the feature status expresses a performative meaning, reason why the cP category is not projected in the higher position of an embedded predicate (p. 30); consequently, an element that encodes this information does not occur in the embedded domain. In (12b), the deontic reading is maintained, but as the report of an order, not a performative act. The obligation is not being placed on the addressee or any other specific participant; this is an ought-to-be with a generic referent. Its occurrence in the embedded domain, therefore, is according to Portner, Pak and Zanuttini’s (2019) hypothesis. Also, the feature status , to be checked in the cP head, only refers to the speaker-addressee relationship, and the modal element used in (12b) relates the subject of the matrix sentence (the commander) to the participant on whom the obligation falls, who in this case is not necessarily the interlocutor.

Another example of the ought-to-be deontic in the embedded domain is shown in (13b) below:

• (13) a. O bebê tem que/deve ser vacinado para Hepatite B.

• The baby has to/must be vaccinated for Hepatitis B

• ‘The baby ought to be vaccinated against Hepatitis B’

• b. O médico disse que o bebê tem que/deve ser vacinado para Hepatite B.

• The doctor said that the baby has to/must be vaccinated for Hepatitis B.

• ‘The doctor said that the baby ought to be vaccinated against Hepatitis B’

(13a) may correspond to a directive speech act, in which the obligation expressed by the modal ter que/dever – an ought-to-be deontic – falls on the interlocutor. When this same sentence occurs in the embedded domain (13b), it loses its performative character, thus it does not express the relationship between the speaker (who gives the order) and the interlocutor (to whom the order is directed) anymore. As it has been argued, in (12a) and (13a) the cP category is projected, given that the modal is an ought-to-be deontic with a specific addressee .

(14) is an example in which the subject of the sentence on whom the modal operates presents the same referent as the interlocutor:

• (14) a. Você tem que/deve trabalhar aos sábados .

• You have to/must work on Saturdays

• ‘You ought to work on Saturdays’

• b. O chefe disse que você tem que/deve trabalhar aos sábados.

• The boss said that you have to/must work on Saturdays

• ‘The boss said you ought to work on Saturdays’

Both in (14a) and (14b), the obligation expressed by the modal falls on the interlocutor, as the pronoun ‘você’ (you), which refers to this participant, corresponds to the subject of the sentences with the modal. Note that, in (14b), ter que/dever is in the embedded domain, even when the obligation falls on the interlocutor. This is possible because the modal creates a relationship between the interlocutor (subject of the embedded domain) and the subject of the matrix sentence, not the participants of the speech event: the speaker and the interlocutor. It is this last relationship that requires checking for the feature status in the head of the cP category, an operation that cannot be performed if the modal occurs in the embedded domain.

The different behavior between a ought-to-be deontic with and without a specific addressee in relation to the occurrence in the embedded domain reveals that the concepts presented by Castañeda and Feldman, on the one hand, and Hacquard, on the other, have their origin in the orientation of the modal, which can occur in sentences with a specific addressee, or in sentences with a generic or non-existent addressee. In the next section, we will argue that this difference is reflected in the structure projected for the ought-to-be interpretation.

On the position of the ought-to-be deontic in the CP system

The phenomenon of actuality entailment, addressed in subsection 2.1, shows that the ought-to-be deontic, similarly to the epistemic modal, occupies a higher position than TP. Thus, according to Tsai (2015) , the natural space for these modals seems to be the left periphery of the sentence, the interface zone between the propositional content expressed by IP and the discursive context ( RIZZI, 1997 ). This perspective is reinforced by the semantic-discursive characteristics of these modals. As noted by Tsai, the epistemic is oriented towards the discourse. This modal in particular considers the speaker’s knowledge of the discursive context. The semantics of the ought-to-be deontic is also related to the perspective of the agents in the communication: the speaker and the addressee.

Progress in the analysis of the ought-to-be deontic and the epistemic modal would be identifying with precision the position of these heads in the left periphery, which has the following configuration ( RIZZI; BOCCI, 2017 ):

• (15) [Force [Top* [Int [Top* [Foc [Top* [Mod [Top* [Qemb [Fin [IP ...]]]]]]]]]]]

• ( RIZZI; BOCCI, 2017 , p. 9)

Relevant tests to detect the position of the projection of high modals refer to situations of their interaction with elements that occur in the left periphery, such as wh-elements, focus, topic and certain fronted adverbials, such as quickly ( RIZZI, 1997 , 2001 ; RIZZI; BOCCI, 2017 ; on BP, see MIOTO, 2001 , 2003 ).

The case of interrogatives with the wh-element why will now be considered. These interrogatives show a different syntactic behavior from the one verified in questions with argumental wh-elements, such as what , and with wh-elements that correspond to low adverbials, such as where and how . The following example shows that why has no sensitivity to negation in English, unlike how: ²

• (16) a. Why didn’t Geraldine fix her bike?

• b. *How didn’t Geraldine fix her bike? ( SHLONSKY; SOARE, 2011 , p. 657)

According to Shlonsky and Soare (2011) , if the ungrammaticality of (16b) results from a violation of Relativized Minimality ( RIZZI, 1990 ), caused by the intervention of negation, as illustrated in (17), the chain formed by why has its tail above NegP, as outlined in (18).

Image 2
How and why interacting with negation

This shows that the base position of why needs to be above the NegP projection.

Using data from Italian, Rizzi (2001) shows that interrogatives with argumental wh-elements, such as che cosa (what) and low adverbials, such as dove (where) and come (how), have to show I to C movement (see (19)). However, this movement is not mandatory for interrogatives with wh-elements that correspond to higher adverbials, such as perché (why) (see (20)):

• (19) a. * Che cosa Gianni ha fatto ?

• a’. Che cosa ha fatto Gianni ?

• ‘What did Gianni do?’

• b. * Dove Gianni è andato?

• b’. Dove è andato Gianni?

• ‘Where did Gianni go?’

• c. * Come Gianni è partito?

• c’. Come è partito Gianni?

• ‘How did Gianni leave?’( RIZZI, 2001 , p. 5)

• (20) Perché Gianni è venuto ?

• ‘Why did Gianni come?’( RIZZI, 2001 , p. 7)

Besides, the author notes that argumental wh-elements do not occur with a contrastive focus (see (21)). However, this restriction does not apply to the element perché in the order perché -Focus (see (22)).

• (21) a. * A GIANNI che cosa hai detto (, non a Piero) ?

• ‘TO GIANNI what did you say (, not to Piero)’

• b. * Che cosa A GIANNI hai detto (, non a Piero) ? ( RIZZI, 1997 , p. 291)

• (22) Perché QUESTO avremmo dovuto dirgli, non qualcos’altro?

• ‘Why THAT we should have told him, not something else?’( RIZZI, 2001 , p. 7)

Rizzi (2001) hypothesizes that elements such as why and perché have their external merge in the CP system, more precisely in Spec of IntP. Besides, he also proposes that the [+wh] feature is intrinsic to the Int° head. Therefore, in questions with perché , the Wh-Criterion ( RIZZI, 1996 ) is met in IntP, between perché and Int° [+wh]. Consequently, I to C becomes unnecessary. This does not occur in argumental wh-phrases or in those in which wh-elements correlate to low adverbs, as (19a’), (19b’) and (19c’), because the generation in Spec of IntP does not extend to these elements. As these elements are not sentential operators, their external merge needs to be internal in relation to IP in order to have their scopes established. In this case, the Wh-Criterion is met via another strategy: the movement of these elements to Spec of FocP and, as a consequence, the rise of the [+wh] inflected verb to Foc°.

Perché merges externally in IntP, this is why it can occur with a contrastive focus, as illustrated in (22): it occupies the position of Spec of IntP, above the FocP projection, and therefore, co-occurs with a focus in the perché- Focus order. The ungrammaticality of the sentences in (21) results from the co-occurrence of argumental and low adverbial wh-elements with a contrastive focus. These elements are inserted in an IP internal position, and move to Spec of FocusP, thus competing with the contrastive focus that occupies this position.

The same analysis can be performed in BP for por que in the left periphery. The well-formedness of (23a) shows that this element has an external merge position above NegP, according to (23b). Example (24), in turn, shows that por que does not move to Spec of FocP, as it co-occurs with a contrastive focus in the por que -Focus order.

• (23) a. Por que o Paulo não viajou? (Why didn’t Paulo travel?)

  b.

• (24) Por que UMA MOTO você comprou? ( Why did A MOTORBIKE you buy ?) (não um carro) (not a car )

Considering that the element por que is originated in a left peripheral position, which can be assumed to be the Spec of IntP, it is possible to use interrogatives with this element as a test to check the highest limit of a modal verb. (25) and (26) below show how an ought-to-be deontic with and without a specific addressee, respectively, interact with the wh-element por que:

• (25) A: A filha da protagonista tem que/deve ter olhos azuis .

• The daughter of the protagonist has to/must have blue eyes.

• ‘The main character’s daughter ought to have blue eyes’

• B: Por que a filha da protagonista tem que/deve ter olhos azuis ?

• Why the daughter of the protagonist have to/must have blue eyes?

• ‘Why does the main character’s daughter ought to have blue eyes?’

• C: Sinceramente, eu não sei a razão dessa ordem .

• Honestly, I not know the reason for that order.

• ‘Honestly, I do not know the reason for that order.’

• (26) A: No Brasil, tem que/deve haver políticas de incentivo aos estudantes .

• In Brazil, have to/must have policies of incentive for-the students.

• ‘In Brazil, there ought to be incentive policies for students’

• B: Por que tem que/deve haver políticas de incentivo aos estudantes no Brasil?

• Why has to/must have policies of incentive for-the students in Brazil?

• ‘Why ought there to be incentive policies to students in Brazil?’

• C: Honestamente, eu não sei a razão dessa obrigação .

• Honestly, I not know the reason of-this obligation.

• ‘Honestly, I do not know the reason for that obligation.’

The fact that (25B) and (26B) can be answered with sentences (25C) and (26C), which highlight causes for the order and the obligation expressed by the modal, shows that the wh-element ( por que ) has scope over the ought-to-be deontic (ter que/dever) . The conclusion, therefore, is that the auxiliary modal ought-to-be is interpreted in a position below the IntP projection.

From this point, we will verify the behavior of ought-to-be in relation to a lower position in the CP system. Before that, consider the adverbial fronting penomenon in (27b), discussed in Rizzi and Bocci (2017):

• (27) a. Gianni ha trovato rapidamente la soluzione .

• ‘Gianni has found quickly a solution’

• b. Rapidamente, Gianni ha trovato la soluzione .

• ‘Quickly, Gianni has found the solution’( RIZZI; BOCCI, 2017 , p. 5)

It is relevant to notice that adverbials such as rapidamente (quickly) can be topicalized and focalized. However, the fronting in (27b) corresponds to a phenomenon that serves to highlight an adverb, without giving it topic or focus status. As observed by Rizzi and Bocci (2017) , even though a sentence like (27b) is similar to a topic structure from an intonational point of view, it has a different interpretation: topics require a connection with the pragmatic-discursive context, which is, necessarily, given information; fronting of an adverb like rapidamente in (27b), on the other hand, does not need to establish a connection with the previous context. If compared to a focalization structure, (27b) is intonationally and interpretatively different (it is not its function to provide new information contrasting with another element previously presented).

Apart from that, the adverbial fronting in (27b) has a syntax that differs from the ones that correspond to the structures of topic and focus. Among the differences identified by Rizzi and Bocci (2017) , an important one is that adverbial fronting is a phenomenon restricted to the sentence in which it occurs, differently from adverbial topicalization or focalization. Thus, in a structure such as (28), rapidamente can act on the matrix sentence (informing that Mario said something quickly), but not on the embedded sentence. This restriction, however, does not apply in topicalization and/or focalization. For example, in a sentence like (29), in which rapidamente is interpreted with contrastive focus, it can have both matrix and embedded readings.

• (28) Rapidamente, Mario ha detto (--) che Gianni ha trovato (* ) la soluzione .

• ‘Quickly, Mario said that Gianni has found the solution’

• ( RIZZI; BOCCI, 2017 , p. 6)

• (29) RAPIDAMENTE Mario ha detto ( ) che Gianni ha trovato ( ) la soluzione, non lentamente .

• ‘QUICKLY Mario said that Gianni has found the solution, not slowly’

• ( RIZZI; BOCCI, 2017 , p. 6)

Rizzi and Bocci (2017) use example (30) to argue that the projection involved in the adverbial fronting, called ModP, occupies a position above the lowest topic projection in the left periphery. On the other hand, ModP must be below IntP, since rapidamente can only follow the complementizer se ( if ), which is the head of IntP, as (31a-b) show.

• (30) Rapidamente, i libri, li hanno rimessi a posto .

• ‘Quickly, the books, they put them back in their place’

• ( RIZZI; BOCCI, 2017 , p. 6)

• (31) a. Mi domando se, rapidamente, Gianni potrà trovare la soluzione .

• ‘I ask myself if, quickly, Gianni will be able to find the solution’

• b. * Mi domando, rapidamente, se Gianni potrà trovare la soluzione.

• ‘I ask myself, quickly, if Gianni will be able to find the solution’

• ( RIZZI; BOCCI, 2017 , p. 7)

ModP can also not precede a Foc° projection, because in sentences such as (32), rapidamente has a topic reading, not a fronted adverb reading.

• (32) Rapidamente, I LIBRI hanno rimesso a posto, non gli articoli.

• ‘Quickly, THE BOOKS they put them back in their places, not the articles’

• ( RIZZI; BOCCI, 2017 , p. 7)

According to the authors, (32) is an appropriate structure as a sequence for (33a):

• (33) a. So che hanno rapidamente rimesso a posto gli articoli...

• ‘I know that they quickly put back in their place the articles…’

• b. No! rapidamente, I LIBRI hanno rimesso a posto, non gli articoli .

• ‘No! quickly, THE BOOKS they put back in their place, not the articles’

• The conclusion that Rizzi and Bocci (2017) come to is that Mod occupies the inferior part of the CP system, only occurring above the lowest topic position.

• Considering the configuration of the CP system proposed by the authors (see (15)), the interaction between the head Mod° and the ought-to-be deontic with and without a specific addressee is tested in (34a-b) and (35a-b), respectively:

• (34) a. (* Rapidamente) Tem que/Deve (rapidamente) denunciar (rapidamente) quaisquer fraudes (rapidamente) na administração pública (rapidamente) .

• (Quickly) have to/Must (quickly) report (quickly) any frauds (quickly) in administration public (quickly).

• ‘(Quickly) Any fraud (*quickly) in public administration (*quickly) ought to be (quickly) reported (quickly).’

• b. (* Rapidamente) Tem que/Deve (rapidamente) haver (rapidamente) uma melhor distribuição de renda (rapidamente) no país (rapidamente) .

• (Quickly) Have to/must (quickly) have (quickly) a better distribution of income (quickly) in the country (quickly).

• ‘(*Quickly) There (quickly) ought to be (*quickly) a better income distribution (quickly) in the country (quickly).’

• (35) a. ( *Rapidamente) O parto (*rapidamente) tem que/deve (rapidamente) ser feito (rapidamente) neste hospital (rapidamente ).

• (Quickly) The delivery (quickly) has to/must (quickly) be done (quickly) in this hospital (quickly).

• ‘(*Quickly) Delivery (*quickly) ought to (*quickly) happen (quickly) in this hospital (*quickly).’

• b. (* Rapidamente) O ginásio (*rapidamente) tem que/deve (rapidamente) ser preparado (rapidamente) para o show (rapidamente) .

• (Quickly) The gym (quickly) has to/ must (quickly) be prepared (quickly) for the show (quickly).

• ‘(Quickly) The gym (quickly) ought to (quickly) be prepared (quickly) for the concert (quickly).’

If the descriptions in (34) and (35) are correct, the position of the interpretation of the ought-to-be deontic occurs in CP, above the Mod category.

Regarding the FocP category, it was noted that the natural order for the interaction between focalized elements and an ought-to-be deontic is Focus > Ought-to-be, as example (36) below shows:

• (36) NO PAÍS tem que/deve (?*NO PAÍS) (rapidamente) haver (rapidamente) uma melhor distribuição de renda (rapidamente).

• IN THE COUNTRY have to/must (IN THE COUNTRY) (quickly) have (quickly) a better distribution of income (quickly).

• ‘IN THE COUNTRY there ought to (*IN THE COUNTRY) (quickly) be (quickly) a better income distribution (quickly)’

Also, considering the order between this deontic and the wh-element por que , shown in (25) and (26) above, the conclusion is that the head ought-to-be is located below the Int and Foc categories and above the Mod category, according to (37):

• (37) [Force[Top*[Int[Top*[Top*[Foc [OughtP [Top*[Mod[Top*[Qemb[Fin[IP... ]]]]]]]]]]]]

The ought-to-be deontic therefore corresponds to a modal head interpreted in a high position in the structure of the sentence. The use of this deontic with a specific addressee presupposes that the speaker assumes that he/she has authority over the interlocutor. For this reason, in this case the modal needs to check features that express the relationship between the interlocutors, such as status [S<A, S≤ A, S=A, S≥A, S>A]. This feature is checked in the head of the cP category – which encodes contextual information ( PORTNER; PAK; ZANUTTINI , 2019 ). The hypothesis presented in this article is based on the behavior of ought-to-be deontics in relation to the phenomenon of actuality entailment and, in a more determinant way, to the non-occurrence of the ought-to-be deontic in the embedded domain (root phenomenon), addressed, respectively, in subsections 2.1 and 2.2. The ought-to-be with a specific addressee is part of a structure in which the functional cP category is projected. On the other hand, an ought-to-be without a specific addressee is part of a structure in which this category is not projected, because the interpretation of this modal does not depend on features that capture the relationship between the interlocutors.

The relevant part of the syntactic structure proposed by Portner, Pak and Zanuttini (2019) for elements that express the relationship between the speaker and the interlocutor has been transcribed below in ( Image 3 ):

Image 3
The speaker-addressee relation at the syntactic structure according to Portner, Pak e Zanuttini (2019)

The ought-to-be with a specific addressee needs to check features in the head of the cP category, the reason why its occurrence is restricted to the matrix domain, which has an interface with the discourse. A linguistic element that is present in the embedded domain does not access the head of the cP category, which cannot be the complement of a higher predicate since it encodes information about the communication situation and of the status of the interlocutors. The representation in ( Image 3 ) explains the restriction of a deontic auxiliary modal to the embedded domain when the obligation is placed directly on the addressee. However, to explain the performative interpretation bound to an ought-to-be with a specific addressee, it is necessary to change, in ( Image 3 ), the type of sentence specified in the SentM category: from declarative (DECL) to imperative (IMP). Therefore, this deontic is relativized to a speech event whose content corresponds to a list of tasks that belongs to the addressee (the addressee’s To-Do List), as proposed by Hacquard (2006) for imperative constructions.

As previously noted, this analysis allows to differentiate an ought-to-be with a specific addressee from an ought-to-be with a generic or non-existent addressee. Thus, the biggest challenge is knowing how to differentiate the structure of an ought-to-be without a specific addressee from an epistemic modal; considering that both are interpreted above the categories of Tense and Aspect ( CINQUE, 1999 ; HACQUARD, 2006 , 2010 ; TSAI, 2015 ) they do not present actuality entailment, happen in both the matrix and the embedded domain and are present in declarative sentences – corresponding to assertions. Regarding this last property, Yanovich (2015) observes that, in Hacquard’s proposal, the content of a speech event in a declarative sentence corresponds to the set of beliefs of the speaker, which is only compatible with the epistemic interpretation. Besides, it is also worth noting that Hacquard illustrates the difference between high and low modals giving representations that correspond to the ought-to-do deontic and to the epistemic modal³ . The position of interpretation of a high deontic modal and its structural differences with respect to an epistemic modal are not specified by the author.

The discussion involving the position of interpretation of an epistemic modal in the syntactic structure is not in the scope of this research, but it is important to observe that it is a very challenging issue that needs to be accurately investigated, as it has a narrow interface with modal semantics.

Concluding remarks

In this article, our aim was to show that the deontic modal ought-to-be can be used without a specific addressee, according to Castañeda (1970) and Feldman (1986) , or with a specific addressee, according to Hacquard (2006 , 2010 ). Evidence was provided to show that, in both cases, the modal is interpreted in the left periphery of the sentence, between the IntP and ModP heads.

The behavior of ought-to-be deontic regarding phenomena such as actuality entailment and (non-)occurrence in the embedded domain was analyzed. The results indicated differences in their syntactic structure. Based on the analysis conducted by Portner, Pak and Zanuttini (2019) for the Korean particles called speech-style particles, it was proposed that an ought-to-be deontic with a specific addressee needs to check features that express the speaker-addressee relationship, such as status ; for this reason, it is part of a structure that has an interface with the functional category cP, whose head allows it to be checked. On the other hand, an ought-to-be deontic without a specific addressee does not refer to the relationship between the interlocutors; consequently, it is part of a structure without the projection of the cP category, thus it can occur both in the matrix and the embedded domain.

Acknowledgements

We thank CNPq for their support (Process 424025/2016-7), conceded by the Edital Universal 01/2016.

REFERÊNCIAS

• BHATT, R. Ability Modals and their Actuality Entailments . Stanford: CSLI, 1999. Disponível em: http://people.umass.edu/bhatt/papers/wccfl-ability-modals.pdf . Acesso em: 7 mar. 2022.
  » http://people.umass.edu/bhatt/papers/wccfl-ability-modals.pdf
• BRENNAN, V. Root and Epistemic modal auxiliary verbs . 1993. Tese (Doutorado em Linguística) - University of Massachusetts, Amherst, 1993.
• CASTAÑEDA, H. On the Semantics of the Ought-to-do . Synthese , Dordrecht, n. 21, p. 449-468, 1970.
• CINQUE, G. Adverbs and Functional Heads: a cross-linguistic perspective. New York: Oxford University Press, 1999.
• CINQUE, G.; RIZZI, L. The cartography of syntactic structures. CISCL Working Papers on Language and Cognition , Siena, v. 2, p. 43-59, 2008.
• FELDMAN, F. Doing the Best We Can: philosophical Studies. Dordrecht: Reidel, 1986. (Series in philosophy, 35).
• HACQUARD, V. On the Event Relativity of Modal Auxiliaries. Natural Language Semantics , Dordrecht, n. 18, p. 79-114, 2010.
• HACQUARD, V. On the interaction of aspect and modal auxiliaries. Linguistics and Philosophy , Dordrecht, n. 32, p. 279-312, 2009.
• HACQUARD, V. Aspects of modality . 2006. Tese (Doutorado em Filosofia da Linguística) - Massaschusetts Institute of Technology, Cambridge, 2006.
• HAEGEMAN, L. Topicalization, CLLD and the left periphery. ZASPapers in Linguistics , Berlin, n. 35, p. 157-192, 2004.
• MIOTO, C. Focalização e quantificação. Revista Letras , Curitiba, v. 61, p. 169-189, 2003.
• MIOTO, C. Sobre o sistema CP no português brasileiro. Revista Letras , Curitiba, v. 56, p. 97-139, 2001.
• PORTNER, P.; PAK, M.; ZANUTTINI, R. The speaker-addressee relation at the syntax-semantics interface. Language , Washington, v. 95, n. 1, p. 1-36, 2019.
• RECH, N.; SOARES, E. C.; GUESSER, S. L. A interpretação deôntica no português brasileiro: um estudo de natureza experimental. Revista Diacrítica , Braga, v. 33, p. 178-195, 2019.
• RECH, N. F.; VARASCHIN, G. Predicados estativos e os tipos de deôntico: ought-to-do e ought-to-be. Cadernos de Estudos Linguísticos , Campinas, v.60, n.1, p.159–177, 2018a.
• RECH, N. F.; VARASCHIN, G. Propriedades do modal deôntico ought-to-be. Alfa , Araraquara, v.62, n.2, p.361–380, 2018b.
• RIZZI, L. On the Position of Interrogative in the Left Periphery of the Clause. In : CINQUE, G.; SALVI, G. (ed.). Current studies in Italian syntax: essays offered to Lorenzo Renzi. Amsterdam: Elsevier North-Holland, 2001. p. 287-296.
• RIZZI, L. The Fine Structure of the Left Periphery. In: HAEGEMAN, L. Elements of Grammar: a handbook of generative syntax. Kluwer: Dordrecht, 1997. p.281–337.
• RIZZI, L. Residual Verb Second and the Wh criterion. In: BELLETTI, A.; RIZZI, L. (ed.). Parameters and Functional Heads . Oxford: New York: Oxford University Press, 1996. p. 63–90.
• RIZZI, L. Relativized minimality . Cambridge, Massachussetts: The MIT Press, 1990.
• RIZZI, L.; BOCCI, G. The left periphery of the clause: primarily illustrated for Italian. In: EVERAERT, M.; VAN RIEMSDIJK, H. C. (ed.). The Wiley Blackwell Companion to Syntax. 2nd ed. Oxford: Blackwell, 2017. p. 589–638.
• SHLONSKY, U.; SOARE, G. Where’s ‘why’?. Linguistic Inquiry , Cambridge, v. 42, n. 4, p. 651-669, 2011.
• TSAI, W. D. On the Topography of Chinese Modals. In : SLHONSKY, U. (ed.). Beyond Functional Sequence: The cartography of syntactic structures. Oxford: Oxford University Press, 2015. v.10. p. 275-294.
• YANOVICH, I. Epistemic Modality. Draft. 2015. Disponível em: http://www.sfs.uni-tuebingen.de/~yanovich/papers/Yanovich_SemCom_epistemic_draft_March2015.pdf . Acesso em: 7 Mar. 2022.
  » http://www.sfs.uni-tuebingen.de/~yanovich/papers/Yanovich_SemCom_epistemic_draft_March2015.pdf

• 1
  “Note that we are using the ought-to-do/ought-to-be labels to refer to the distinction between the class of deontics that puts an obligation on the subject vs. that which puts an obligation on the addressee” ( HACQUARD, 2006 , p. 40).
• 2
  Data adapted from Rizzi (2001) and subsequent works.
• 3
  We transcribe, in (i) and (ii), the representations proposed by Hacquard (2010 , p. 7) for the derivations of an ought-to-do deontic and an epistemic modal, respectively: (i) [_CPSpeech e₀𝛌e₀[_TPT Asp₁𝛌e₁Mod f (e₁) [_VPV e₁]]] (Deôntico ought-to-do) (ii) [_CPSpeech e₀𝛌e₀Mod f (e₀) [_TPT Asp₁𝛌e₁[_VPV e₁]]] (Epistêmico) In (i), the modal acquires a deontic reading of the ought-to-do type; it is interpreted in a low position in the structure and the event variable is related to the event described by the vP (e₁). In (ii), the modal has an epistemic interpretation; it is interpreted in a high position in the structure and the event variable is related to the speech event (e₀).

Publication Dates

• Publication in this collection
  17 Oct 2022
• Date of issue
  2022

History

• Received
  2 Oct 2020
• Accepted
  11 June 2021