‘NEC’ DISCOURSE STRUCTURING PARTICLE (LATIN) ENDURED IN CONTEMPORARY PORTUGUESE ‘NEM’

ABSTRACT

Contemporary Portuguese does not have a direct negator for D-quantifiers (*‘Não cada cachorro late’ ‘Not every dog barks’) (Souza, 2017); but a combination of ‘todo’ with ‘nem’, derived from Latin’s ‘nec’ (‘Nem todo cachorro late’/ ‘Not all dogs bark’), apparently yields a universal negation reading (¬∀). However, ‘nem’ is restrained to the very beginning of the sentence (*‘Eu não vi nem todo cachorro correr’ ‘I didn’t see every dog run’), or follows a topic (‘À minha aula, nem todo aluno veio’ ‘To my class, not every student came’) (Peres 2013). Following Gianollo (2020), ‘nem todo’ is analyzed as a discourse structuring particle. ‘Nem’ contains an additive layer over a negation component. It denies the maximization presupposition triggered by the universal quantifier, linking the new information given by the pronounced sentence to previously established contextual information. ‘Nem todo’ instructs the conversation participants to correct their common ground expectations about the participation of the entire nominal domain in the situation. By ‘nem’s distribution in Brazilian Portuguese today and its presence from the XV century to the XXI century, ‘nec’ as a discourse structuring particle did not disappear in romance languages, contrary to Gianollo (2020).

KEYWORDS
nominal quantifier negation; universal quantifiers; diachronic evolution from Latin to Portuguese; nec (Latin); nem (Portuguese)

RESUMO

O português hodierno não apresenta partícula negadora especializada em quantificadores nominais (D-quantificadores) (*‘Não cada cachorro late’) (Souza, 2017). Entretanto, é frequente a combinação do universal ‘todo’ com a partícula ‘nem’, derivada do Latim ‘nec’ (‘Nem todo cachorro late’). Aparentemente, ‘nem todo’ gera a negação do quantificador (¬∀). Porém, ‘nem todo’ está restrito ao fronteamento da sentença (*‘Eu não vi nem todo cachorro correr’) ou após constituintes topicalizados (‘À minha aula, nem todo aluno veio’) (Peres, 2013). Neste trabalho, conforme Gianollo (2020), ‘nem’ seguido de ‘todo’ será analisado como uma partícula estruturante do discurso. ‘Nem’ contém uma camada aditiva, sobre outra negativa. A camada aditiva liga a informação nova trazida pela sentença iniciada por ‘nem todo’ a uma informação do contexto previamente estabelecido. A camada negativa rejeita a pressuposição de maximalidade disparada pelo universal, instruindo os falantes a corrigirem suas expectativas sobre a aplicação da predicação sentencial ao domínio nominal completo quantificado por ‘todo’. O exame da distribuição de ‘nem todo’ em português brasileiro contemporâneo e sua presença do século XV ao XXI indicam que a função de ‘nec’ como partícula discursiva estruturante não se extinguiu em seus sucessores nas românicas, contra Gianollo (2020).

PALAVRAS-CHAVE
negação de quantificadores nominais; quantificador universal; diacronia do latim ao português; nec (latim); nem (português)

Introducing the functions of NEC in Latin and weighing its resilience in modern languages

Gianollo (2020) keeps track of the Latin particle ‘nec’ (‘ne’+ ‘que’ = ‘neque’) in its evolution to its Romance continuations. The historical journey of ‘nec’ from Latin to Romance languages is described as starting with the function of a discursive connector and ending, nowadays, as a non-interpretable feature. According to the author, ‘nec’ originates from adding the Indo-European negative morpheme *nĕ to the enclitic additive conjunction -que. In Latin, ‘nec’ performed many distinct functions, of which the diachronic study cited highlights three:

1. A discourse structuring connector that appears at the very beginning of new discursive units, meaning something akin to the English expressions ‘and not’; ‘furthermore’, ‘it is not the case that’;

2. A negative correlative particle with a meaning close to the English expressions ‘neither’, ‘nor’;

3. An isolated focus particle that can be used with additive interpretation (similar to the English expression ‘also not’) or scalar interpretation (a meaning close to ‘not even’).

Among the many functions of ‘nec’, Gianollo picks precisely those three because the last two have been preserved to several degrees in practically all Romance languages, being very productive up to now; the first, on the contrary, is relevant because, having been primary, has completely disappeared and is no longer found today.

Indeed, one does not need to look far to see that functions II and III are still attested in Brazilian Portuguese and that they are quite frequent, without any restrictions on textual genre or discursive formality. We will offer a few examples in order to illustrate how productive the function still is:

(1)Nem o Fluminense nem o Volta Redonda foram tão dependentes dos atacantes como o Bangu foi de Luis Felipe (Nem Cano, 2023)

“Neither Fluminense nor Volta Redonda was as dependent on strikers as Bangu was on Luis Felipe”

(2)Nem tirar foto, nem filmar; confira regras do aniversário de Anitta (Silva, 2023)

“Neither take photos nor film; check out Anitta’s birthday rules”

Both (1) e (2) exemplify ‘nem’ in function II, as a negative correlative particle. The first example says of two football teams, Fluminense and Volta Redonda, that each of them depends on a single team player, respectively the top scorers Germán Cano, with 14 goals, and Lelê, with 13, to score well in the Championship Tournament. However, neither team depends as much on a single player as Bangu, who only scores because of Luis Felipe. We then have a connection that makes Fluminense and Volta Redonda similar, in addition to a denial that affects both and says that the level of dependence of neither of them reaches that of a third team. As for the second example, it brings about a ban made by the popular singer Anitta on guests for her birthday party, which prohibits two things altogether: taking photos and filming. The two occurrences of ‘nem’ ‘neither’ in (2) are the only tip for us to understand that both things are ruled out, the only source for the negation which equally affects photography and filming, both connected by the absence of permission.

Let us now illustrate the scalar interpretation of ‘nem’ as an isolated particle of focus (the function III):

(3) Nem o pai da criança veio ao chá de bebê! (Quase Ninguém, 2023) “Not even the child’s father came to the baby shower!”

(4)Um buraco negro é uma região do espaço com um campo gravitacional tão intenso que nem mesmo a luz consegue escapar de dentro dele. (O que são, 2023) “A black hole is a region of space with a gravitational field so intense that not even light can escape from it.”

Sentence (3) (from Revista Crescer, 03/18/2023) reports the disappointment of a pregnant woman due to the fact that no guests came to her baby shower, not even the baby´s father. There is no sentential negation marker in the sentence. Due to scalar ‘nem’, we understand from (3) that no one attended. The noun phrase “o pai da criança” “the child’s father” provides the end of a scale with ordered degrees of expectations about attendance, denoting that individual from whom the speaker thought he was the most likely to come. If the person at the extreme end of the scale did not come, the negative extends additively to the other members of the list, whose chances of coming were lower, according to the speaker’s judgment. Sentence (4) was taken from a school website. It states that nothing, absolutely nothing, escapes a black hole. The embedded sentence “nem mesmo a luz consegue escapar de dentro dele” “not even the light can escape from within him” has no other source of negation besides ‘nem’. “A luz” “the light” is the end of a scale of all the things a black hole could suck in, ordered by the chance of each escaping it. Light is the item on the scale that has the greatest chance of escaping because it has no mass and because it is so fast. Therefore, it appears in the sentence after ‘nem’, as the realization of the top of the pragmatic scale. If not even the item with the greatest chance can leave the black hole, additively negation reaches the items with the lowest chances, and we understand that a black hole is inescapable.

To exemplify function I, we repeat here a Latin example offered in Gianollo (2020). The author argues that this function, although historically primitive and attested since the beginnings of the textual tradition, with some signs of retention until the Ancient Romance, was later discontinued, and altogether ceased to occur in modern Romance languages.

(5) Accessum est ad Britanniam ominibus navibus meridiano nearby:PTCP be:3SG of Britannia:ACC all:ABL ships: ABL noon:ABL fere tempore, neque in eo loco hostis around-of time:ABL and-not in that place enemy: NOM est visus be:3SG see: PTCP

(5)	Accessum	est	ad	Britanniam	ominibus	navibus	meridiano
	nearby:PTCP	be:3SG	of	Britannia:ACC	all:ABL	ships: ABL	noon:ABL
	fere	tempore,	neque	in	eo	loco	hostis
	around-of	time:ABL	and-not	in	that	place	enemy: NOM
	est	visus
	be:3SG	see: PTCP

“All the ships arrived in Britain around noon, and not a single enemy was seen there.” (Caes. Gall.5.8)

According to Gianollo, in this Latin example, ‘neque’, as an additive conjunction, connects the first clause of the period, a previous discursive unit, to a new clause, the one it heads. The clause headed by ‘neque’ adds a new discursive unit. Furthermore, the only source of denial of the presence of enemies on British lands upon the arrival of the ships is ‘neque´: notice that the second sentence does not have any other negative particle or word.

After translating (5) into BP (6), in fact, as Gianollo says, it becomes clear that this combo function of adding a new discursive unit and, jointly, inserting a negation that acts on the content of the sentence in which the particle is located, does not endure. Using ‘nem’ in place of ‘neque’, in a translation of example (5), we get:

(6)Nem um inimigo foi avistado ali.

“Not a single enemy was seen there.”

Sentence (6) sounds natural nowadays, but it is not interpreted as an instantiation of function I: it rather exemplifies function III. Sentence (6) has a clear scalar effect: we come to the understanding that no enemy was seen in that location from the denial of the visualization of the extreme element of the scale, which is the smallest number of enemies that could be seen. When one does not see at least one enemy, one also does not see larger quantities of that enemy. The expression “um inimigo” “an enemy” after ‘nem’ realizes the bottommost end of a quantity scale, and the negation spreads to all the members of such a pragmatic scale, as outlined here:

(6’)bottom most alternative: exactly one enemy was spotted there

stronger alternative than the previous one: exactly three enemies were spotted there

stronger alternative than the previous ones: exactly four enemies were spotted there

the strongest alternative of all: all enemies were spotted there

By denying the first alternative, which represents the minimum amount that could represent some visualization of the enemy in (6), ‘nem’ ‘not even’ also denies all the other alternatives, which are stronger, since it is not logically feasible to see a larger amount without also seeing the smaller amounts included in the larger one. Additively, all the alternatives on the scale are computed under negation, resulting in the interpretation that no one who is an enemy was seen on that occasion.

We can make a generalization about the data containing ‘nem’ followed by a nominal phrase such as the one exemplified in (6) in modern Portuguese. It will always be interpreted in function III. For examples of the type in (6), the insertion of ‘sequer’ (‘Nem um inimigo sequer foi avistado ali’) / ‘mesmo’ ‘even’ (‘Nem mesmo um inimigo foi avistado ali’) after ‘nem’ or the nominal phrase introduced by ‘nem’ is always licensed, preserving the typical reading of function III. This data supports Gianollo’s statement about function I having reached its end.

However, we will argue that, although this has not been noticed in the literature so far, or at least not to the best of our knowledge, there is a special context in which ‘nem’ retained the I function of ‘nec’ until nowadays: the one in which ‘nem’ precedes the universal quantifiers ‘todo’ ‘all’ — a nominal quantifier, a D-quantifier, in the sense of Partee, Bach e Kratzer (1987) — and ‘sempre’ ‘always’ — a sentential quantifier, or A-quantifier,¹ in the sense of Partee, Bach e Kratzer (1987).

(7)Nem todos os remédios curam doenças. (“Not all medicines cure diseases”)

(8)Nem todas as saladas são saudáveis. (“Not all salads are healthy”)

(9)Nem sempre chove no Dia de Finados! (“It doesn’t always rain on All Souls’ Day!”)

(10)Cliente nem sempre tem razão. (“Customers are not always right.”)

Examples from (7) to (10) are widespread in current BP. It is impossible to interpret them assuming that ‘nem’ performs function III, in which an entire scale of options ordered by the degree of expectation or probability is reached by the negation, with the nominal preceded by “nem,” realizing the extreme of the relevant scale. Nominal phrases quantified by universals, such as ‘todos os remédios’ “all the medicines” (7) e ‘todas as saladas’ “all salads” (8), cannot count on several distinct alternatives originating from the same domain and ranked on a scale in order of expectation since universals denote the complete set, the whole nominal denotation. Precisely for this reason, phrases quantified by the universal ‘todo’ ‘all’ cannot denote the weakest alternative on a scale since, from a quantitative point of view, there are no quantities greater than the complete nominal domain. What could be more than everything together? Nothing. Perhaps it is precisely due to the impossibility of having a scalar interpretation that these special contexts have been resisted with ‘nem’ in function I.

It is worth making an observation here about ‘sempre’ ‘always’: adverbs of frequency or quantificational adverbs (A-quantifiers) mark the regularity of an event repetition or, in other words, quantify on the number of episodes of the same nature that take place over a given period of time. For instance, in (9), a holiday All Souls’ Day, comes again every year; and we understand the version of (10), taking out ‘nem’ (‘The customer is always right’), as stating that on all occasions when there is a dispute between the customer and the company, the reason must be given to the customer. We can list adverbs of frequency by degree, ranging from the smallest to the largest number of episodes in the temporal interval under examination. A possible list would be: (i) ‘nunca’ ‘never’, (ii) ‘quase nunca’ ‘almost never’/ ‘raramente’ ‘rarely’; (iii) ‘regularmente’ ‘regularly’; (iv) ‘frequentemente’ ‘frequently’; (v) ‘quase sempre’ ‘almost always’; (v) ‘sempre’ ‘always’. A list with this kind of ordering constitutes a scale, in which the items mentioned on the right are stronger than those mentioned on the left, as the smaller frequencies are subsets of the larger ones. If something always happens (for example, it rains every year on All Souls Day for a decade), this logically includes the fact that such a thing also happens frequently (it rains on All Souls Day a good number of years during a decade). However, there is no stronger alternative than ‘sempre’ ‘always’. Therefore, ‘nem sempre’ ‘not always’ could not be associated with a scale with stronger alternatives covered by negation, since no alternatives will show a greater degree. Here too there is a blockage of function III; i.e., ‘nem sempre’ shares with ‘nem todo’ the same conditions that provide the continuation of function I of ‘nem’. Frequency markers with a lower value, which admit stronger alternatives, do license the reading of ‘nem’ in function III, as exemplified in (11):

(11)João não veio à minha aula nem uma vez este mês. (“João hasn’t come to my class even once this month”)

Sentence (11) says that João was absent for the whole relevant month. Since there are stronger frequency alternatives than ‘uma vez’, ‘once’, such as ‘duas vezes’ ‘twice’, ‘três vezes’ ‘three times’, etc., the construction of the pragmatic scale is allowed, and the denial of this student’s presence in my class is additively distributed by all the members of a scale, thus enabling function III reading.

Closing the comment on ‘nem sempre’, it is worth remembering that Lewis (1975) treat ‘every’ and ‘always’ as “synonyms”, as tripartite quantifiers (quantifiers with restrictor and quantificational domain²), one acting in the nominal domain and the other, non-selective, especially in the adverbial domain. Both generate the same interpretations. So, the sentences ‘Todo galo canta ao alvorecer’ ‘Every rooster crows at dawn’ and ‘Um galo sempre canta ao alvorecer’ ‘A rooster always crows at dawn’ will both show the same truth conditions. Since there are no stronger quantifiers than ‘todo’ ‘all’ and ‘sempre’ ‘always’ in their respective domains, function III of ‘nec,’ with a scalar focus, could not be applied to ‘nem todo’ nor to ‘nem sempre.’ In our proposal, that is the reason why function I, as a discursive connector, did not disappear over time in exactly these contexts. In fact, function III replaced function I in all instances in which ‘nem’ comes before nominal items that can participate in pragmatic scales with stronger, higher-degree alternatives.

Although, in our view, the same reason counts for preserving the I function of ‘nem’ with the two universals, in this article, we will leave ‘nem sempre’ aside and deal exclusively with ‘nem todo’ ‘not all’. The decision relies on the fact that ‘nem todo’ has received many distinct analyses in the grammatical tradition and in academic literature, analyzes that we must discuss here.

Nominal quantifier negation in Portuguese

In Portuguese, nominal negation cannot be produced by the prefixation of the same word that performs propositional negation to a noun, as in, for example, *‘uma não-cadeira’ ‘a non-chair’. ‘Não’ can only be freely composed with predicational nouns, adjectives, and participles, such as ‘a não-possibilidade de socorro’ ‘the non-possibility of help’ and ‘uma ponte não-terminada’ ‘an unfinished bridge’ (Mioto, 1998). In Portuguese it is not grammatical to say *“Não todos os alunos vieram” “Not all the students came”, but you can say ‘Nem todos os alunos vieram hoje’ ‘Not all the students came today’. Traditional grammar (TG) compendiums generally do not record the fact that ‘nem’ is licensed exclusively with ‘todo’ and ‘sempre.’

The most famous Traditional Portuguese grammars (Bechara, 2009; Lima, 2011; Cunha; Cintra, 2016) focus on ‘nem’ in function II; in Cegalla, for example, we read “The series nem... nem acquires a negative additive meaning: Nem estudou nem tirou boas notas — “Neither studied nor got good grades” (i.e., he didn’t study, and he didn’t get good grades)” (Cegalla, 2008, p. 202). ‘Nem’ is mainly associated by the grammatical tradition to function II, as a correlative coordinating conjunction. Function III generally gains much less space in these works, which do not even list ‘nem’ as a negation adverb, with the exception of Boregana’s grammar (1996). However, several teaching materials and dictionaries include ‘nem’ among negation adverbs, thus addressing its use in function III. Neves (2000) provides a clear example of ‘nem’ performing Gianollo´s (2020) function III; she classifies ‘nem’ as a negation adverb in the following occurrence: “A patroa quer dar umas voltinhas, nem quer saber do jogo.” “The madam wants to go for a walk, she doesn’t even care about the game” (Neves, 2000, p. 287). Finding examples of function I in teaching materials and akin words is much more difficult. Still, there are a few mentions: “before todos, tudo and sempre, it is up to NEM, by itself, to express negation: “Nem sempre teremos essa sorte” “We won’t always be that lucky”; “Nem tudo que reluz é ouro” “Not everything that shines is gold”; “Nem todos podem pagar esse preço” “Not everyone can pay this price””.³ A grammar consultancy provides the following clarification to a consultant: “In fact, in the presented sentence –– ‘Viajar é bom, mas nem todos temos dinheiro para isso’ “Traveling is good, but not all of us have the money for it” —, one cannot replace ‘nem’ by ‘não’. Therefore, “não todos temos dinheiro” (intended: “not all of us have money”) is not acceptable”. The same consultant comments about the fact that the book Gramática do Português (Fundação Calouste Gulbenkian) treats this construction as a case of negation of the universal quantifier, and also adds that, according to the same source, in this function ‘nem’ can only occur in the very initial position of a phrase or clause. Furthermore, the grammar consultant comments that he refers to “a construction whose motivation neither the source consulted fully explains, nor other grammars manage to record”.⁴ In sum, although “nem todo” and “nem sempre” are very productive forms in Portuguese, the grammatical tradition practically ignores them. Only one generative grammar work deals with the topic, describing these constructions as “negation of the universal quantifier,” without offering any explanation, however, for the fact that only universal quantifiers can be negated by ‘nem’ in Portuguese.

Let us now see what the linguistic literature says about the negation of nominal quantifiers. In English, a proper noun cannot be directly modified by ‘not’ (Mateyak, 1997 gives as an example *‘Not Salome came to the party’), but quantifiers can, although this possibility does not extend to all of them (*‘Not some people came to my party’ *‘Not several events are planned for that weekend’). Beghelli and Stowell (1997, p. 31, e.g., 35) show that ‘not’ can constitute a nominal negation phrase with many generalized quantifiers but not with ‘each’:

(12)a Not more than ten boys ate an ice cream cone.

b ?Not ten boys ate an ice cream cone.

c Not many boys ate an ice cream cone.

d Not all the boys ate an ice cream cone.

e Not every boy ate an ice cream cone.

f *Not each boy ate an ice cream cone.

If we provide the translation of the examples in (12), it becomes clear that in Portuguese, nominal negation is much more restrictive:

(13)a Nãomais de 10 meninostomaram sorvete de casquinha.

b *Não10 meninostomaram sorvete de casquinha.

c Nãomuitos meninostomaram sorvete de casquinha.

d ??Nãotodos os meninostomaram sorvete de casquinha.

e Nemtodo meninotomou sorvete de casquinha.

f *Nãocada meninotomou sorvete de casquinha.

In our view, examples (13a,c,d) do not constitute the combination of a nominal negation particle with a generalized quantifier but instead are cases of sentential negation with implicit material (that is, there is an unpronounced copula verb in the structure, which makes it an instantiation of sentential negation). In (14), let us make this elliptical material visible:

(14)a Não (foram) mais de 10 meninos (os que) tomaram sorvete de casquinha.( = (13a))

(There were boys having ice cream cones, but did not reach ten in number).

b Não (foram) muitos meninos (que) tomaram sorvete de casquinha. ( = (13c))

(There were boys having ice cream cones, but they were not so many).

c Não (foram) todos os meninos (que) tomaram sorvete de casquinha. ( = (13d))

(There were boys having ice cream cones, but they were not the whole group of boys present).

We maintain that these negations are sentential, with elliptical material, given the fact that they can appear in coordination and as short answers to questions, typical contexts of deletion:

(15)a Meninos tomaram sorvete de casquinha, mas (eles) não (eram) mais que dez. ( = (13a)) (The boys had ice cream cones, but (they were) not more than ten)

(16) Quantos meninos tomaram sorvete de casquinha? (How many boys ate ice cream cones?)

Regarding (13e), the very oddness of (16c) indicates that we have a unique characteristic. Although the answer (16d) is imprecise in expressing the quantity of boys who had ice cream in a cone, (16d) is a more acceptable response to (16) than (16c). It is easy to verify that you cannot use ‘nor’ before other nominal quantifiers.⁵

(17)a *Nemmais de 10 meninostomaram sorvete de casquinha. (intended: No more than ten boys ate ice cream cones)

b * Nemqualquer meninotomou sorvete de casquinha. (intended: Not just any boy ate ice cream cones)

c * Nemmuitos meninostomaram sorvete de casquinha. (intended: Not many boys ate ice cream cones)

d Nem todos os meninos tomaram sorvete de casquinha. (Not all the boys ate ice cream cones)

e Nem todo menino tomou sorvete de casquinha. (Not all boys ate ice cream cones)

f *Nemcada meninotomou sorvete de casquinha. (intended: Not every boy had ice cream cones)

g *Nemalgum meninotomou sorvete de casquinha. (intended: Not a single boy ate an ice cream cone.)

Apparently, judging by (17), ‘todo’ is the only Portuguese nominal quantifier with a nominal form of negation. Considering ‘todo’ as the linguistic realization of the logical universal quantifier ∀, one will take ‘nem todo’, in (17d, e) to express universal negation, ¬∀. How can one explain this exceptionality of ‘todo’ in terms of the combination with a particle of negation? To pursue an answer to this question, one must first look at the proposals of the articles already mentioned, which sought to explain why nominal negation is not available for all quantifiers in the English language.

For Mateyak (1997), negation is licensed exclusively with defined quantified noun phrases. So, for the author, a quantifier can be negated as long as it projects a continuation of possible values for the size of the negated domain and as long as the quantifier offers no other semantic contribution besides the quantitative dimension to be negated. This view provides a plausible explanation for the grammaticality contrast between ‘nem todo’ and *‘nem qualquer’ as expressions of ¬∀, given that ‘todos’ (as an isolated word) makes an appropriate answer to a quantity question, but ‘qualquer’ does not. ‘Qualquer’ ‘any’ is a universal quantifier with qualitative content in its semantics. However, this proposal encompasses the existence of more negated quantifiers in English, in addition to ‘not every’ e ‘not all’, such as ‘not a’, ‘not one’, ‘not many’ ‘no more than’ e ‘not less than’. So, even if we consider ‘nem’ in ‘nem todo’ ‘not all’ as a form of nominal negation for the quantificational value of universal ‘todo’ (¬∀), we will not find ‘nem’ combined to the Portuguese correspondents to these English quantifiers, at least not with this same type of interpretation, as seen in (17).

Beghelli and Stowell (1997) employ nominal negation to distinguish between strong distributivity (‘each’ ‘cada’) and weak distributivity / universality (‘all’, ‘every’ ‘todo’), stating that only weak distributivity licenses it, which would also describe the facts in PB (*‘nem cada’/*“not each” x ‘nem todo’/ “not all”). However, regarding the acceptance or not of nominal negation by the different quantifiers, exemplified in (12), the authors say that “Although it is a test capable of putting ‘every’ and ‘all’ together on the one hand, and ‘each’ on the other, it is not clear what the semantic property diagnosed by it would be.’ (Beghelli; Stowell, 1997, p. 31).

Collins (2020), working with English, distinguishes two types of quantifier negation: external negation ([NEG DP]) and internal negation ([[NEG D] NP]). In this proposal, universal quantifiers can only have external negation, while existential quantifiers only have internal negation. The evidence offered is the licensing of negative polarity items. That analysis also cannot be transported directly to Portuguese, given that no existential accepts being preceded by ‘nem’ with a nominal negation interpretation. ( ¬∃) in BP. Furthermore, there are universals in Portuguese, such as ‘qualquer’, that do not accept nominal modification by ‘nem’( ¬).

Perhaps the literature already examined in this section has not enlightened the analysis because the grammars of Portuguese and English are too far apart. Perhaps it would be better to look at analysis done on data from other Romance languages. Well, Polakof (2021) analyses ‘cualquier’ as a universal quantifier involving Free Choice, like originally proposed by Vendler (1967), offering examples of nominal negation for this universal of the Spanish spoken in the Rio Plata area:

(18)No cualquier jugador puede jugar en Boca.

‘Not any player can play in Boca.’

In Portuguese, the translation for (18) would be ‘Não é qualquer jogador que pode jogar no Boca’ or ‘Jogar no Boca não é para qualquer um’, because the language does not count with a negation nominal particle to go with ‘qualquer’: *‘nem qualquer’, as shown in (17b), is not grammatical. The material which in Portuguese can only be constructed with predicative, sentential negation, in Spanish (see (18)), as well as in English, is built as a nominal negation. In Spanish, apparently, a particle of nominal negation, similar in form to sentential negation, can be used both before ‘cualquier’ ‘any’ and ‘todo’ ‘all’:

(19)Han venido no todos los dirigentes ni simpatizantes, sino solo algunos.

‘Not all the directives nor sympathizers have come, only some.’

Now, example (19), reproduced from Etxepare e Uribe-Etxebarria (2011, p. 236, apudPolakof, 2021), on the other hand, has both ‘no’ and ‘ni’, in Spanish, as versions for Portugues ‘nem’. In the variety of Spanish spoken in Rio de la Plata, both universals (‘cualquier’ ‘any’ and ‘todo’ ‘all’) accept nominal negation. In PB only one of them, ‘todo’, counts with something of the kind. Before weighing the arguments for and against using this attested difference to decide whether or not ‘qualquer’ in Portuguese is a universal quantifier, we need to take a look at the behavior of the Spanish universals.

Although ‘todo’ and ‘cualquier’ are not distinguished in terms of their combination with ‘no’ ‘nem’, as we saw in (18) and (19), the linguists who examined the interaction between the semantics of the universal quantifier and the negation particle in Spanish concluded that even in these examples it is an external negation. According to them, no nominal constituent is formed by a negation operating directly on a quantifier. Regardless of presenting itself as NEG+QP, as illustrated in (19), ‘no todo’ is accepted in contexts where negative polarity items such as ‘no pocos’ (NEG+ a few) are not. Etxepare and Uribe-Etxebarria (2018) give some examples:

(20)*Han venido no pocos dirigentes ni simpatizantes.

*‘Not a few leaders nor sympathizers have come.’

The interaction between case-marking prepositions and negated quantifiers is also scrutinized by the authors (apudPolakof, 2021):

(21)a A no pocos les gusta el vino.

(22)a *A no todos les gusta el vino.

The fact that the Spanish case marker preposition “a” cannot appear in a higher position is taken as indicating that the negation has scope over the universal in (22) instead of forming a constituent with it. A second argument presented to support the analysis of ‘no’ in (22) as external negation is the fact that this negation preceding the universal can come in sentences coordinated with another quantifier:

(23)No [_IP a todo el mundo le gusta el vino] o [_IP a cualquiera — el café].

Not everyone likes wine, nor does everyone like coffee. (The Spanish sentence is coordinating ‘todo’ ‘all’ and ‘cualquiera’ ‘any’ under negation)

The final argument given to support that NEG+∀ is not a syntactic constituent is the fact that it does not support clause ellipsis:

(24)a. Ha ido donde no todos *(han ido).

* He went where not everyone (has ever gone). (The verb “go” in Spanish receives the very same form both times)

Resuming the tests used by Etxepare and Uribe-Etxebarria (2018), we conclude that ‘no todos’ ( ¬∀) is not a syntactic constituent in Spanish, but a sentential negation. Polakof (2021) investigates ‘cualquier’ ‘any’, furthermore looking deeply into the interaction between negation and the universal when it is in the subject position of generic sentences and also when in the position of verb complement in episodic sentences, as well as in modal sentences. Given the complexity of the argumentation, we refer the interested reader to the cited article. What matters is that, with data support, she concludes that ‘cualquier’ is a Free Choice Item that may combine with ∀ or ∃ and an indeterminate universal quantifier. Therefore ‘cualquier’ cannot cooccur with negation, despite the fact that, in non-argumental sequences, ‘no cualquier’ constitutes a legitimate case of negated constituent.

Although there is some debate about the status of ¬ in that variety of Spanish, the literature converges in recognizing the existence of (some) negation of constituents with quantifiers in Spanish. Even though we are talking about two Romance languages, Spanish and Portuguese, there are differences in this area of grammar that do not authorize a unified approach. In Spanish, the constituents with negated quantifiers accepted begin with ‘no’, a particle homonymous to the sentential negation (Nocualquier jugador puede jugar en Boca / Ronaldonopuede jugar en Boca). Spanish also has the particle ‘ni’, a continuation of ‘nec’ (Latin) (Es importante mencionar que no todo videojuego puede ser considerado un E-Sportnitodo jugador puede ser considerado un “ciberatleta”).⁷ On the other hand, in Portuguese, unlike in Spanish and English, quantifiers cannot come preceded by a form homonymous with the sentential negation ‘no’, as exemplified in (16c).

We conclude that there is no negation of constituents with quantifiers in Portuguese. The impression that ‘nem todo’ corresponds to a syntactic constituent is illusory, regardless of the analysis adopted for ‘not all’ (English) or ‘no todos’ (Spanish). One strong impediment for ‘nem todos’ to be analyzed as a nominal negation, a single syntactic constituent, is the fact that it is restricted to the border region of the sentence:

(25)a Nem todos os meus filhos comeram as verduras. (‘Not all of my children ate the vegetables’)

b Os meus filhos (não) comeram todas as verduras. (My children ate all the vegetables./ My children didn’t eat all the vegetables.)

c *Os meus filhos (não) comeram nem todas as verduras. (Intended: ‘My children didn’t eat all of the vegetables’)

The fact that the negation of the universal quantifier ‘todo’ is only allowed to occur in the sentential initial position had already been observed by Grammar of Portuguese, the book from the Calouste Gulbenkian Foundation, as we have already said; in this book, it is also observed the following:

[A] constituent where a universal is negated by ‘nem’ [‘nem todos’ ‘not all’, ‘nem sempre’ ‘not always’] can only appear in a non-initial sentence position in case another constituent is topicalized, as in “o livro, nem todos os estudantes o tinham lido” “the book, not all students had read it”. (in this case, the initial constituent “o livro” ‘the book’ was moved to the initial position). (Peres, 2013, v. I. p. 482).

However, no explanation for this restriction to sentential fronting is provided. We offer some explanations. Assume that ‘qualquer’ ‘any’ and ‘todo’ ‘all’ are universals in Portuguese, although occurrences of ‘nem todo’ are attested but occurrences of *‘nem qualquer’ are not. ‘Nem’ cannot be seen as a formant in nominal quantifier constituent restricted to universals contributing negation because ‘nem’ simply does not form a syntactic constituent with ‘todo’. ‘Nem’ is a discourse structuring particle (Gianollo, 2020); that function only survived in Portuguese in the face of universal quantifiers due to the fact that nominal universals do not allow for stronger alternatives, and, therefore, do not generate the reading of scalar focus particles. ‘Qualquer’ is also a universal nominal quantifier, it is indefinite and a FCI (Free Choice Item) that expands its quantification domain into alternatives, each containing an element of the nominal domain and situated in a different possible world. The pragmatic scale formation requires its bottommost element (the nominal realized after ‘nem’) to be a specific noun phrase, like ‘João’, in ‘Nem mesmo João faltou hoje’ ‘Not even João was absent today’. The non-pronounced alternatives will be the individuals ordinated in expectation degrees, according to their chances of being absent, in the speaker evaluation. One cannot say *‘Nem qualquer aluno faltou’ (indented: ‘Not every student is absent’) because ‘qualquer aluno’ ‘any student’ is not a specific nominal. Also, for the reading of a structuring discursive particle, ‘nem’ requires to be followed by a defined nominal denotation. As we have already said, ‘nem todos’ ‘not all’ does not admit stronger alternatives because it denotes a maximum, specific quantity. That is why ‘nem todo’ ‘not all’ resisted in function I: it is not fit for function III. On the other hand, ‘qualquer’ ‘any’, although universal, cannot have a specific reading, having an epistemic or non-identification of the referent component, which is part of its indefinite character. Therefore, ‘qualquer’ ‘any’ does not present a presupposition of maximality, on which the I function of ‘nem’ depends in modern Portuguese. Furthermore, despite the fact that Free Choice Items like ‘qualquer’ are universals (Dayal, 2004; Szabolcsi, 2019; Oliveira, 2005), in Brazilian Portuguese, they also operate like negative polarity items (IPN), as in “João não comprou qualquer jornal”, “John didn’t buy any newspaper”, which receives two interpretations: (i) “John bought no newspaper; and (ii) “John bought a newspaper, but not the one of his choice” (Oliveira, 2005). Recall that in Spanish, where there are the forms ‘no todo’ and ‘no cualquiera’, ‘cualquier’ is exclusively a universal quantifier of free choice, never operating as an IPN. In this section, we argued that it would be an error to analyze ‘nem’ as a negative nominal syntactic constituent, and we explained why Portuguese does not have ‘nem qualquer’. ‘Nem’ is not a particle that forms nominal negations with universals in general but is rather a structuring particle of discourse, which introduces the sentence corresponding to a new informational unit, which is joined to an older, contextual informational unit. This new information unit repairs the presupposition of maximality triggered by ‘todo’ and is assumed by the speaker to be the old information unit, part of the conversational background. In the next section, we will present in detail our proposed analysis for ‘nem todo’.

‘Nem’ and ‘nem todo’ as a discourse structuring particle

The grammaticalization process (functional enrichment) undergone diachronically by ‘nec’ presented a unique, well-defined path, according to Gianollo (2020). On the one hand, the mechanism that governed the recovery of alternatives associated with the original operator (from Latin) evolved towards a scalar interpretation (which constitutes a process of pragmatic enrichment), as is typical of additive particles. The author points out that the disappearance of the competitor “ne...quidem” accelerated this process. On the other hand, she says, the increase in connections and the decrease in syntagmatic variability over time are in direct correlation with a change in the form of alternatives, which go from discursive units to individuals ordered on a scale. The researcher sees a pattern over time: the decline or loss of discursive dependence. This general trend is palpable in today’s Portuguese data. Function II is fully productive nowadays (‘Melatonina não é remédio para dormir nem suplemento’ ‘Melatonin is neither a sleeping pill nor a supplement’) and, in the vast majority of times, there is a single appearance of the particle in the sentence (‘Hoje nem enfrentei fila’ ‘Today I didn’t even stand in line’), even before a noun phrase, we have function III (‘Hoje, nem turista vem mais ao Pantanal, porque não tem peixe’ ‘Nowadays, not even tourists come anymore to Pantanal, because there are no fish’). However, as already stated, ‘nem todo’ ‘not all’ cannot name an individual participating in a pragmatic scale, as it simply does not denote an individual among others in the same domain. We propose that, accordingly, ‘nem todo’ has become one of the last strongholds of resistance to the function I of ‘nec’ in modern Portuguese (‘Nem todos os heróis usam capas’ ‘Not all heroes wear capes’).

To support our proposal that ‘nem todo’ is a remnant in BP of the function of ‘nec’ structuring the discourse, it is important to rule out the possibility that it is something recent. For Gianollo, function I of ‘nec’ (Latin) receded or ended in the ancient Romance languages. Could it be that, after a period of non-existence, this ‘nem todo’ function would have resurfaced with greater force as a modern innovation? A survey in Corpus do Português generated the following results:

Chart 1
– The presence of ‘nem todo’ in Portuguese over the centuries

century	XII	XIII	XIV	XV	XVI	XVII	XVIII	XIX	XX
nem tudo	0	0	0	5	7	8	61	106	6.478
nem todo	0	0	2	0	5	7	4	24	5.076
nem todos	0	0	1	19	29	35	93	189	10.296
nem toda	0	0	3	1	4	3	17	27	1.422
nem todas	0	0	2	4	7	7	43	70	4.037
TOTAL	0	0	8	29	52	60	218	416	27.309

Source: Author’s elaboration with Ana Clara Polakoff.

The lack of attested occurrences in the Galician-Portuguese phase of the archaic period (in the first two columns, from the left) does not necessarily prove that the function did not exist then, as well as the magnitude of the number of occurrences attested in the 20th century by itself does not prove a real increase in the use of ‘nem todo’, considering that the absolute number of sources available grew exponentially lately. Whether the continuation of function I in the 12th and 13th centuries was not documented, but it was used; whether the function I had disappeared completely for some time and returned later, in the so-called period of common Portuguese or historical prose, there is no factual basis in favor of one of these two options. Either way, it is undeniable that, in the last six centuries of Portuguese, there has been an uninterrupted great strength in the use of ‘nem todo`. No way this function is dead!

Gianollo (2020) gives a basic semantics to ‘nec’, subsumed in the functions performed by both this particle and its substitutes, still used in Romance languages today. Focusing her analysis on the bimorphemic nature of this Latin particle (ne-c, ne-que), the author proposes that all its functions share a syntactic structure with two components: one, more external, corresponding to additivity, and another, more internal, corresponding to negation. This is the mirrored order of the morphological formation of the Latin particle, which first brings the negation morpheme ‘ne’ and to its right ‘que’ / ‘c’, the additive conjunction. This inversion of ordering between morphology and syntax is attributed to phonological formation rules.

As a structuring particle of discourse, ‘nec’ introduces a complete sentence that belongs to a new discursive unit, connected to a previous discursive unit. In Latin, the connected discursive units could present the same polarity or reverse polarity. As a structuring particle of discourse, ‘nec’ introduces a complete sentence that belongs to a new discursive unit connected to a previous discursive unit. In example (26), the polarity is reversed:

(26) [7.4] Qua ex re creverat cum fama tum opibus, which: ABL of things: ABL grow: 3SG and reputation:ABL and power:ABL magnamque amicitiam sibi cum quibusdam regibus Threciae close: ACC friendship: ACC refl:DAT with-certain:ABL king:ABL Thracia:GEN pepererat. [8.1] Neque tamem a caritate patriae potuit recedere. search: 3DG and.not meanwhile of love: ABL homeland: GEN can:#SG recrudify:INF

(26) [7.4] Qua	ex re	creverat	cum fama tum	opibus,
which: ABL	of things: ABL	grow: 3SG	and reputation:ABL and	power:ABL
magnamque	amicitiam	sibi cum quibusdam	regibus	Threciae
close: ACC	friendship: ACC	refl:DAT with-certain:ABL	king:ABL	Thracia:GEN
pepererat.	[8.1] Neque	tamem a caritate	patriae potuit	recedere.
search: 3DG	and.not	meanwhile of love: ABL	homeland: GEN can:#SG	recrudify:INF

‘Having grown in reputation and power, he (Alcibíades) sought to endure strong friendships with certain Thracian kings. Even so, he never managed to abandon his love for his homeland. (Nep. Alc. 7.4–8.1) (Gianollo, 2022, p. 41)

Underlying ‘nec’ in any function is a syntactic-semantic basis in which a focus projection whose nucleus is ‘que’ (phonologically reduced to ‘c’ in ‘nec’). Its complement is the maximum projection nucleated by the negation operator carried out by ‘ne’. This, in turn, takes another maximum projection (of an undetermined category) as a complement. We assume the presuppositional analysis adopted by Gianollo for additivity, in which the predication present in the focused element must be valid for at least one of the alternatives (p or q) projected by Focus, linked by the additivity of ‘nec’ and present in the context (C), as in (27):

(27)presuppositional analysis for additive particles like ‘nec’

a. asserted content: nec p

b. presupposed content: ∃ q ∈ C ∧ q ≠ p

(27b) states that ‘nec’ ‘nem’ introduces a sentence that presents the proposition p, combining it with another proposition, q, which can be carried out by another sentence previously uttered in the speech, as in (26); both sentences (p e q) belong to the same context, that is, they are units of the same discourse.

Let us now see how ‘todo’ fits into this idea. Since Barwise and Cooper (1981) it has been assumed that some generalized quantifiers, but not all, are presuppositional. In the case of ‘todo’ ‘all’, there are two presuppositions: (i) the domain over which ‘todo’ quantifies must be populated, that is, it cannot be empty; and (ii) the predication must apply to all components of the entire quantified domain. The latter is the maximality presupposition. In a contextual dynamic, a sentence denotes a partial function from possible worlds to truth values. Presuppositional triggers need to have their presuppositions satisfied in the possible worlds in which the truth of the sentences of which they are part is verified. Contextual dynamics treats presuppositions as filters, which select, for sentence denotation, only those possible worlds in which the presuppositions are satisfied, discarding the others. The truth value of a sentence containing a presuppositional trigger is undefined in a world in which its presupposition is not satisfied. Presuppositions are part of the conversational background common to speakers (common ground) (Stalnaker, 1973), which can be defined as follows:

The common ground [...] is a set of propositions that the participants of a certain conversation [...] mutually assume they can take for granted [...] the common ground describes a set of worlds, the context formed by those possible worlds in which all common ground propositions are true. Context constitutes the set of worlds that, for what is now taken for granted, could be the actual world (Von Fintel, 2008, p.1).

Therefore, the common ground of sentences containing ‘todo’ excludes worlds in which their domain of quantification is empty and further excludes worlds in which the predication applies on less than the entire domain of quantification of ‘todo’, since the truth value of the sentence that contains it remains undefined in a world that does not contain the whole domain as a predication argument. We represent in a basic way the presupposition of a non-empty domain in (28a) and the maximality presupposition in (28b); once both are combined, we have in (28c) the description of the ‘todo’ presuppositional trigger:

(28)a. presupposed content 1(non-empty domain): ∃(x) ∧ p(x)(w) =1

b. presupposed content 2 (maximality): ∀(x)→q(x)(w) =1

c. [[todo]]^g= λp. λq. λx. λw: [∃(x) ∧ p(x)(w) =1]∧[∀(x)→q(x)(w) =1]

d. In (28c), p is the nominal predicate in the quantifier restriction, and q is the sentential predicate in the quantifier matrix

According to (28c), a proposition like ‘Todos os animais que nadam são peixes’ ‘All animals that swim are fish’ presupposes that there are animals that swim and that no element of the domain of swimming animals is outside the group of fish. These presuppositions make up the conversational background, taken as beliefs on which all participants in the conversation agree, i.e., propositions that everyone previously accepted as true in the worlds in which a given assertion will have its truth verified. Let us also suppose that ‘todo’ ‘all’ is within a focus position:

(29)[_FOC Todos] os animais que nadam são peixes. ‘All animals that swim are fish’

Focus is a prosodic-pragmatic phenomenon that limits the discursive contexts in which a sentence can be used. It is treated in a semantics of alternatives, which understands that the focused constituent generates at least two distinct alternatives for its content (Rooth, 1992). The semantic alternatives for a constituent like the quantifier ‘todo’ ‘all’ can only concern the denotation of that quantifier. (29) is understood as a reinforcement of the presuppositions linked to the quantifier ‘todo’, so that (29) could be paraphrased as “There are animals that swim and absolutely all of them, without exception, are fish”. The alternative is to deny this paraphrase, refusing one of the presuppositional contents, either the presupposition of existence or that of maximality. The most vulnerable to contestation is the presupposition of maximality. For example, if I do not believe in (29) because my knowledge of the world tells me that there is at least one animal that swims but is not a fish, the whale, I would respond like this to whoever told me (29):

(30)Nem todos os animais que nadam são peixes. A baleia nada e não é um peixe. (Not all animals that swim are fish. The whale swims and is not a fish.)

‘Nec’, in Latin, was a focus-sensitive particle of negation (Gianollo, 2017, 2020), which could narrow the focus of a sentential constituent. Adopting this proposal for ‘nem’, the focused constituent it combines with is ‘todo’, highlighting two alternatives, namely: (i) both presuppositions are maintained; (ii) both presuppositions are not to be maintained. Applying (27) to (28), we get the following analysis for (30):

(31)a. asserted content: nem p = Nem ([_FOC Todos] os animais que nadam são peixes) (Not ([_FOC all] animals that swim are fish)

b. presupposed content: ∃ q ∈ C ∧ q ≠ p

c. q =[[∃(x) ∧ r(x)(w) = 1] ∧ [∀(x) → s(x)(w) = 1]]

d. [[nem todo]] = [^¬ [[∃(x) ∧ r(x)(w) = 1] ∧ [∀(x) → s(x)(w) = 1]] = There are swimming animals, but not all of them are fish.

Following Gianollo (2017, 2020), we claim that, in its function as a structuring particle of discourse, ‘nem’ has the meaning of and-not (∧ ¬), linking two discourse units, one old and one new. When ‘nem’ is preceding ‘todo’ ‘all’, the discursive unit to which the proposition containing ‘nem todo’ is linked is not another assertion, but, instead, the content of the common ground relative to the presuppositions associated with the universal quantifier ‘todo’ ‘all’. However, it is not necessary for someone to first say (29) for someone to say (30). A ‘nem todo’ sentence can be pronounced ‘out of the blue’, without this meaning that its content is discourse-initial. The effect of using ‘nem’ before ‘todo’ is the partial refutation of the presuppositions of this quantifier, that is, the refusal of the maximality presupposition, which, prior to the assertion of ‘nem todo’ is already part of the common ground. The beliefs assumed to be common to the participants in the conversation constitute the initial discursive unit, inseparable from the use of the quantifier ‘todo’. In other words, asserting (30) presupposes that all speakers have in mind (29), even when the content of (29) is not realized in the conversation. The presuppositions evoked by ‘todo’ are taken as old information. We then have a contrastive focus, in which the interlocutor is invited by the speaker of ‘nem todo’ to carry out a correction in part of the beliefs taken as common until then by the participants in the conversation. The new information is that the maximality presupposition should not filter the possible worlds that make up the conversational background. This is a pragmatic repair, which is one of the types of presupposition accommodation. Whoever uses ‘nem todo’ believes their interlocutor believes that the presupposition of maximality linked to ‘todo’ applies and also believes that their speaker is wrong about this. The use of ‘nem todo’ instructs conversation participants to correct the common conversational background accordingly. By using ‘nem todo’, the speaker instructs the participants in the conversation to admit into the common conversational background worlds in which the predication applies to less than the totality of the components of the quantified nominal domain.

Final words

We argue in this article that ‘nem’ in PB does not behave as a true negation of a constituent (True Constituent Negation, TCN) nor as a standard nominal negation for universal quantifiers. Neither can ‘nem’ be described as an item of Illusory Constituent Negation (ICN, à la Etxepare e Uribe-Etxebarria (2018) since ‘nem’ is not available for all quantifiers that can be fronted in the sentence.

We proposed an analysis for ‘nem’ in apparent constituency with ‘todo’ or ‘sempre’ aligned with a continuity in Portuguese of the function I described by Gianollo for ‘nec’ (from Latin), a function that was thought to have been interrupted and no longer attested in modern Romance languages: the function of discursive structuring connector. The opening of a new discursive unit is the key to explaining why the distribution of ‘nem sempre’ and ‘nem todo’ is limited to the very beginning of the sentence or, at most, to the position immediately after a topicalized constituent. We adopted the semantics proposed by Gianollo for the base of ‘nec’ (Latin), common to all its functions: a two-layer complex in which an additivity operator (a focus restrictor) dominates a negation operator. In the uses of ‘nec’ as a discursive structuring particle, a contrastive character is always present. Expanding Gianollo’s analysis, we proposed that sentences beginning with ‘nem todo’ introduce a new discursive unit. The position of ‘nem’ is confined to the beginning of the sentence so that the new discursive unit is coordinated with an old discursive unit, which is not materialized in words but is part of the discursive context or, more precisely, the conversational background. It is the set of beliefs assumed by the speaker to be shared among the participants in the conversation. Therefore, it has a pragmatic status. However, the content of the conversational background to which the new information is linked is dependent on a semantic presupposition triggered by the universal quantifier ‘todo’ ‘all’. Since Barwise and Cooper (1981) it has been known that not all quantifiers are presuppositional. Universals such as ‘all’ and ‘always’ presuppose that their domain has individuals, that it is not empty. ‘All’ and ‘always’ are quantificational, requiring certain properties, respectively, of the quantity of the nominal domain participating in the situation or the number of repetitions of the situation in a given temporal interval. ‘Qualquer’ ‘any’ is qualitative. ‘Qualquer’ (in BP, a Free Choice Item sometimes, and sometimes a Negative Polarity Item), even being a universal quantifier, does not trigger the same presuppositions as ‘todo’. Hence the non-existence of *‘nem qualquer’ in Portuguese.

‘Todo’ triggers two assumptions: (i) that its domain is populated (non-empty) and that of maximality, since the use of ‘todo’ triggers the incidence of sentential predication on each component of the populated nominal domain, resulting in the insertion of the entire nominal domain as a participant in the situation denoted by the predicator. In a Dynamic Theory fashion, these presuppositions act as filters, reducing the number of possible worlds that make the proposition true. By identifying both presuppositions prior to verifying the truth of the proposition, the speaker would eliminate from the selected possible worlds all those in which the sentential predication reached less than the entire set of components of the nominal domain quantified by ‘todo’. By using ‘nem’ before ‘todo’, the speaker brings a new instruction, which says that his interlocutor(s) must reconfigure their beliefs that form the conversational background, discarding the presupposition of maximality, in order to admit, for the verification of the truth of the proposition, the possible worlds in which the predication applies to less than the complete nominal domain. This is a strategy to repair the conversational background. The presupposition (i), that the nominal domain is not empty, is maintained, but the speaker is instructed to “drop” (ii).

The maintenance of one of the presuppositions explains why sentences with ‘nem todo’ are not interpreted as ‘for all of the individuals, it is false that …’ (= ‘nothing’/ ‘nobody’). ‘Nem todos os alunos faltaram’ ‘Not all the students were absent’, said in front of an empty classroom, is interpreted as some students are absent and others are not, and does not fit with a scenario in which no one was expected to be there (??‘Nem todos os alunos faltaram, pois não existe nenhuma turma que use esta sala.’ ‘Not all students were absent, as there is no class that uses this room’). On the other hand, ‘Não é verdade que todos os alunos faltaram’ ‘It’s not true that all the students were absent’, said in front of an empty classroom, can be naturally falsified both by the absence of some students and by the non-existence of the class (‘Não é verdade que todos os alunos faltaram, já que não há alunos que pudessem comparecer ou faltar, pois não existe nenhuma turma que use esta sala’ ‘It is not true that all students were absent, as there are no students who could attend or be absent, as there is no class that uses this room’). The contrast shows that ‘nem’, in combination with ‘todo’‘all’, results in the denial of maximality, not in denying the existence of any participants in the situation; and that the presupposition that the nominal domain (‘os alunos’) is not empty is maintained.

Through the connection operated by ‘nem’ between two discursive units, one old and one new, ‘nem todo’ sentences are always issued in a context in which there is a general adherence to the belief that the predication is valid for all components of the nominal domain (a belief subsumed in the old unity of the discourse), and the novelty, the new information introduced by the express sentence that contains ‘nem todo’, is the need to adjust this belief (originated on the presupposition of maximality) to the fact that, contrary to the scenarios in which the presupposition of maximality is valid, there are exceptions to the rule. It is a pragmatic strategy of repair, of adjusting the conversational background.

Syntactically, we proposed that ‘todo’ is in a focus position in sentences with ‘nem todo’. Focus creates alternatives. Alternatives are at the basis of the semantics of ‘nec’ ‘nem’ proposed by Gianollo. In the case of ‘nem todo’, there will be only two logical alternatives: the one in which the maximality assumption is active (e.g., in the interpretation of ‘Todos os alunos faltaram’ ‘All students were absent’) and the one in which it does not apply (e.g., as in the interpretation of ‘Nem todos os alunos faltaram’ ‘Not all students were absent’). Evidently, the focus discards the option that would be naturally associated with ‘todo’ and leads to assuming the remaining alternative, contrary to maintaining the assumption of maximality. Given the semantics of ‘todo’, logical alternatives ordered from the weakest to strongest are not available, which makes unfeasible pragmatic scales such as those perceived when ‘nem’ precedes a proper name, as in ‘Nem João veio hoje’ ‘Not even João came today’ (meaning that nobody came today).

The semantics of the universal ‘todo’ barred the conversion of ‘nem’ before it into a scalar focus particle, whose reading requires alternatives with distinct individuals ordered by some criterion. ‘Nem todo’ ‘not all’ (and ‘nem sempre’ ‘not allways’) are strongholds of the function of ‘nem’ as a discursive structuring particle, which was preserved only there in PB. Before proper names and defined descriptions, as defended by Gianollo, function I may have been reinterpreted and replaced by III, over the centuries.

In our view, it is mistaken to consider function I of ‘nec’ ‘nem’ completely disappeared after the ancient romance phase. The I function of ‘nec’ remains fully active in modern Portuguese.

We conclude about BP (and perhaps about modern Portuguese, in general) that the language doesn’t count with negative quantifier nominal negation, be it of type True Constituent Negation (TCN) or of type Illusory Constituent Negation (ICN), be it internal or external negation; the idea of a syntactic composition, of constituency, between ‘nem’ and ‘todo’, implicitly assumed by grammatical tradition and school support materials, is an illusion created by its adjacency, which in turn is actually the product of a demand for ‘nec’ ‘nem’ that of coming adjacent to the focused item, which gives you the alternatives in play in both functions I and III.

Acknowledgments

I thank Ana Clara Polakoff, my partner in the research that resulted in this article, which was conducted during my visit (“capacitação”) to the Universidad de la República Uruguay, UdelaR, in Uruguay. I am also grateful to the CAPES PRINT Program, which, through the Postgraduate Program in Linguistics at the Federal University of Rio de Janeiro, made possible my stay in Uruguay from January to March of 2023.

REFERÊNCIAS

• BARWISE, J.; COOPER, R. Generalized quantifiers and natural language. Linguistics and philosophy, Dordrecht, v. 4, n. 2, p. 159–219, 1981.
• BECHARA, E. Moderna gramática portuguesa. 36. ed. Rio de Janeiro: Nova Fronteira; Lucerna, 2009.
• BEGHELLI, F.; STOWELL, T. Distributivity and negation: The syntax of each and every. In: SZABOLCSI, A. (ed.). Ways of Scope Taking: Studies in Linguistics and Philosophy. Dordrecht: Springer, 1997. v. 65. p. 71–107.
• BORREGANA, A. A. Gramática universal: língua portuguesa. Lisboa: Texto Editores, 1996.
• CEGALLA, D. P. Novíssima gramática da língua portuguesa 48. ed. São Paulo: Companhia Editora Nacional, 2008.
• COLLINS, C. Outer negation of universal quantifier phrases. Linguistics and Philosophy, Dordrecht, v. 43, n. 3, p. 233–246, 2020.
• CUNHA, C.; CINTRA, L. F. L. Nova Gramática do Português Contemporâneo. 6 ed. Rio de Janeiro: Lexikon, 2016.
• DAYAL, V. The universal force of free choice. Linguistic variation yearbook, Amsterdam, v. 4, n. 1, p. 5–40, 2004.
• ETXEPARE, R.; URIBE-ETXEBARRIA, M. Context-sensitive aspects of constituent negation. In: MARTIN, R.; GALLEGO, A. J. (org.) Language, syntax, and the natural sciences. Cambridge: Cambridge University Press, 2018. p. 77–98.
• GIANOLLO, C. Grammaticalization parameters and the retrieval of alternatives: Latin nec from discourse connector to uninterpretable feature. In: GERGEL, R.; WATKINS, J. (ed.). Quantification and scales in change. Berlin: Language Science Press, 2020. p. 33–65.
• GIANOLLO, C. Focus-sensitive negation in Latin. Catalan Journal of Linguistics, Bellaterra, v. 16, p. 51–77, 2017.
• LEWIS, D. K. Adverbs of quantification. In: KEENAN, E. L. (ed.). Formal semantics of natural language. Cambridge: Cambridge University Press, 1975. p. 3–15.
• LIMA, C. H. da R. Gramática normativa da língua portuguesa Rio de Janeiro: José Olympio, 2011.
• MATEYAK, H. Negation of noun phrases with not. IRCS Technical Reports Series, 89, 1997. Disponível em: http://repository.upenn.edu/ircs_reports/89 Acesso em: 27 Mar. 2023
  » http://repository.upenn.edu/ircs_reports/89
• MIOTO, C. Tipos de negação. Cadernos de Estudos Linguísticos, Campinas, v. 34, 1998.
• NEM CANO, nem Lelê: Bangu foi o time que mais dependeu de um jogador para marcar no Carioca. Portal R7, 2023. Disponível em: https://esportes.r7.com/futebol/nem-cano-nem-lele-bangu-foi-o-time-que-mais-dependeu-de-um-jogador-para-marcar-no-carioca-22032023 Acesso em: 03 Apr. 2023.
  » https://esportes.r7.com/futebol/nem-cano-nem-lele-bangu-foi-o-time-que-mais-dependeu-de-um-jogador-para-marcar-no-carioca-22032023
• NEVES, M. H. M. Gramática de usos do português. São Paulo: Ed. da UNESP, 2000.
• O QUE SÃO buracos negros? Brasil Escola UOL, 2023. Disponível em: https://brasilescola.uol.com.br/o-que-e/fisica/o-que-sao-buracos-negros.htm Acesso em: 03 Apr. 2023
  » https://brasilescola.uol.com.br/o-que-e/fisica/o-que-sao-buracos-negros.htm
• OLIVEIRA, R. P. de. Qualquer e o conceito de livre-escolha. DELTA: Documentação de Estudos em Linguística Teórica e Aplicada, São Paulo, v. 21, p. 251–277, 2005.
• PARTEE, B. H.; BACH, E.; KRATZER, A. Quantification: a cross-linguistic perspective [NSF proposal]. Amherst: University of Massachusetts, 1987.
• PERES, J. A. Negação. In: PAIVA RAPOSO, E. et al. (org.). Gramática do Português. Lisboa: Gulbenkian, 2013. v. I. p. 461-498.
• POLAKOF, A. C. The negation of cualquier NP. Isogloss: Open Journal of Romance Linguistics, Bellaterra, v. 7, p. 1–24, 2021.
• QUASE NINGUÉM apareceu em meu chá de bebê, nem mesmo meu namorado, desabafa grávida. Crescer Online, 2023. Disponível em: https://revistacrescer.globo.com/fique-por-dentro/noticia/2023/03/quase-ninguem-apareceu-em-meu-cha-de-bebe-nem-mesmo-meu-namorado-desabafa-gravida.ghtml Acesso em: 03 Apr. 2023.
  » https://revistacrescer.globo.com/fique-por-dentro/noticia/2023/03/quase-ninguem-apareceu-em-meu-cha-de-bebe-nem-mesmo-meu-namorado-desabafa-gravida.ghtml
• ROOTH, M. A theory of focus interpretation. Natural language semantics, Dordrecht, v. 1, n. 1, p. 75–116, 1992.
• SILVA, V. Nem tirar foto, nem filmar; confira regras do aniversário de Anitta. Portal Bahia.ba, 2023. Disponível em: https://bahia.ba/entretenimento/nem-tirar-foto-nem-filmar-confira-regras-do-aniversario-de-anitta/ Acesso em: 03 Apr. 2023.
  » https://bahia.ba/entretenimento/nem-tirar-foto-nem-filmar-confira-regras-do-aniversario-de-anitta/
• SOUZA, P. da C. A dupla negação pré-verbal no catalão e no português brasileiro: história, variação e uso. 2017. Tese (Doutorado em Filologia e Língua Portuguesa) - Universidade de São Paulo, São Paulo, 2017.
• STALNAKER, R. C. Presupposition. Journal of Philosophical Logic, Dordrecht, v.2, p.77-96, 1973.
• SZABOLCSI, A. Unconditionals and free choice unified. Semantics and linguistic theory, Washington, p. 320–340, 2019.
• VENDLER, Z. Each and every, any and all. In: VENDLER, Z. Linguistics in Philosophy. Cornell: University Press, 1967. p. 70-96.
• VON FINTEL, K. What is presupposition accommodation, again? Philosophical perspectives, Hoboken, v. 22, p. 137–170, 2008.

• 1
  A D-quantifier is a nominal quantifier (of type D, of determiner), such as “each” and “all”, which can distribute over other nominal phrases (‘each child won 2 chocolates’, considering a group of 10 children, leads to setting the total number of candies distributed as 20). An A-quantifier is a sentential quantifier, of the adverbial type, such as “always” (in “always, when a child goes to the doctor, she gets a lollipop”, we understand that, if a certain child attended 5 appointments in three months, he received a total of 5 lollipops during this period).
• 2
  According to the author, the two sentences bellow mean the same: (i) A rooster always crows at dawn = [_QUANTIFIER always, [_RESTRITOR if it is dawn, [_MATRIX a rooster crows]]] (ii) Every rooster crows at dawn. = [_QUANTIFIER for every x, [_RESTRITOR if x is a rooster, [_MATRIX then x crows at dawn]]]
• 3
  Available at: https://sualingua.com.br/e-nem/. Access on: 13 Dec. 2023.
• 4
  Available at: https://ciberduvidas.iscte-iul.pt/consultorio/perguntas/negacao-nem-todas/36309. Access on: 13 Dec. 2023.
• 5
  ‘Nem 10 meninos tomaram sorvete’ is well-formed, but it will not be discussed here because this ‘nem’ serves another function, the III, with a scalar focus reading. The negation propagates to stronger alternatives, so the sentence expresses the idea that the number of boys who had ice cream is necessarily below the quantity denoted by the noun that ‘nem’ precedes, with the negation affecting not only the fact that there were 10 but also that there were more than ten.
• 6
  It may have occurred to the reader that in Portuguese, both forms would be grammatical: ‘A nem todos o vinho agrada’ / ‘Não (é) a todos (que) o vinho agrada,’ but through direct comparison, it would be hasty to conclude anything about universals, considering that the preposition preceding the complement in Portuguese does not have the same grammatical status as the case-marking preposition in Spanish. Furthermore, the status of *?‘Nem a todos o vinho agrada’ is one of degraded grammaticality.
• 7
  Available at: https://www.docsity.com/es/revision-sistmeatica-psicologia-del-deporte/9334951/. Access on: 13 Dec. 2023.

Publication Dates

• Publication in this collection
  11 Mar 2024
• Date of issue
  2023

History

• Received
  23 May 2023
• Accepted
  30 Sept 2023