Next Article in Journal
Generational Variation in Language Convergence: Lexical and Syntactic Change in Dai Lue Under Chinese Influence
Previous Article in Journal
Phonetic Training and Talker Variability in the Perception of Spanish Stop Consonants
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

On the Evolution of Old Portuguese Indefinite jamais ‘Never’—Syntactic Analyzability and Polarity

Faculty of Arts, University of Lisbon, 1249-058 Lisbon, Portugal
Languages 2026, 11(1), 2; https://doi.org/10.3390/languages11010002
Submission received: 21 October 2025 / Revised: 11 December 2025 / Accepted: 17 December 2025 / Published: 24 December 2025

Abstract

In Contemporary Portuguese, jamais ‘never’ is a negative indefinite that encodes temporal semantics and belongs to the set of strong Negative Polarity Items, being able to express negation on its own, in preverbal position. However, it originates from the merger of two non-negative Latin adverbs—iam ‘now/already’ and magis ‘more’—starting as a construction and later becoming an independent lexical unit, with different features. Data from the 13th century onwards shows that, in early attestations, jamais still preserved some level of internal syntactic analyzability, with the possibility of inverse word order and interpolation. The meaning of the construction could be obtained through the sum of its parts, but its occurrence in negative and modal contexts shows that its interpretation became context-sensitive. This independence was eventually lost, with the emergence of an intrinsic negative reading, favoured in negative contexts through the combination of inchoative and comparative strategies in no-longer expressions.

1. Introduction

When looking at complex linguistic constructions, the notion of internal analyzability arises. Analyzability was first defined by Langacker (1987, p. 457), who considers that “it does not refer to the intrinsic complexity of a structure but rather to a person’s awareness of certain aspects of this complexity”. This roughly means that, in the presence of a new complex construction, speakers may have different degrees of recognition of the independent parts that were involved in the formation of the new item. In this paper, I look at the syntactic development of the negative temporal adverb jamais ‘never’, which starts as two independent adverbs, ‘now/already’ and mais ‘more’, that are reinterpreted as a single lexical unit. Throughout this work, I will assess the level of compositionality of jamais in earlier stages of Portuguese, tracking the progressive loss of internal syntactic analyzability.
The Portuguese item jamais ‘never’ belongs to the series of Romance temporal adverbs with polarity features, which originated in the combination of the Latin adverbs iam magis ‘from now on more’. Spanish jamás (cf. Rueda Rueda, 1997; Yamada, 2023) and French jamais (cf. Hansen, 2012) are two good examples of these outcomes. In some cases, though, the temporal adverb preserves only one of the adverbs, the form magis, as is the case of Italian mai or Catalan mai ‘ever/never’.
The negative temporal adverbs that result from Latin iam magis seem to be an innovation of Romance languages, in the sense that none of the Latin items had a direct association with negation. The Latin adverb iam is described as being related to phasality, focality and polarity (cf. Kroon & Risselada, 2002), therefore occurring in different syntactic configurations and assuming several meanings (‘already, ‘still’, ‘furthermore’, ‘now’, ‘from now on’, ‘soon’). Its occurrence with negation is, however, only associated with phasal expressions and, according to Huitink (2005, p. 563, ft.3), iam does not stress negation, since it always expresses positive polarity.
On the other hand, Latin magis ‘more’ was an adverb of degree/intensity present in comparisons. It gave place to the Portuguese quantity/intensity adverb mais ‘more’, usually used in comparative clauses, and also to the adversative conjunction mas ‘but’.
Portuguese jamais ‘never’ is an interesting case of a construction that became an independent lexical unit but also went from a polarity-neutral element to an intrinsically negative item. Apart from the work by Frade (2010), the history of this item is yet to be investigated.

2. Theoretical Background

Before presenting the data, it is important to clarify a few theoretical concepts and the terminology that will be used throughout the article.
The present work follows a generative approach, therefore having a broader perspective on the notion of construction and seeing it as a syntactic phenomenon, with repercussions in other areas of grammar. I adopt the definition of construction presented in Svenonius (2015, p. 16), who states that “a construction is a characteristic formal pattern of syntactic categories or features, usually associated with some meaning and/or discourse function.”
I will adopt the syntactic distinction between a unit and a construction, but also the semantic notion of compositionality (cf. Bybee, 2010). Although the borders between a unit and a construction can be very thin, I will follow Jiménez Juliá (2017) in considering that the former is completely unanalyzable, while the latter will still have recognizable parts. This means that, in terms of internal structure, I will refer to jamais as a unit only when it behaves as a strong NPI meaning ‘never’. In all other cases, I will consider it as a construction, even though its components may have different degrees of independence. Furthermore, according to Bybee (2010, p. 45), analyzability can be maintained even when the meaning of the new unit can no longer be obtained from the meaning of the component parts. This leads us to the semantic notion of compositionality, which I will make use when necessary. Bybee (2010, p. 45) defines it as “a semantic measure” that “refers to the degree of predictability of the meaning of the whole from the meaning of the component parts.”
On the other hand, the study of jamais also demands the study of specific aspects of negation. Negation and polarity items have been widely debated in the literature, but the approaches and the terms used to refer to similar items differ greatly. For instance, items such as nada (PT) ‘nothing’, res (CAT) ‘nothing’ or rien (FR) ‘nothing’ can be referred to as negative polarity item (NPI), a term originally coined by Baker (1970). Another possibility is to use the terms negative concord item (cf. Espinal & Llop, 2022) or n-word (cf. Laka, 1990). This last term was initially used to account for the fact that most Romance items of this type started with an /n/, but it did not necessarily reflect negative morphology.1
The main challenge when describing and classifying such items is their context of occurrence. Despite the strong association with negative contexts, some of these items may occur in non-negative contexts (nonveridical contexts, as in Giannakidou, 2002), such as interrogatives or comparative clauses.
It is well known that negative polarity items occur in negative concord structures, with two or more of these items establishing an agreement relation and expressing one single instance of negation. In this context, Portuguese has been considered a non-strict negative concord language (cf. Giannakidou, 2006), with a singular behaviour among other Romance languages. Portuguese items nada ‘nothing’ and ninguém ‘no-one’, i.a., can be the single element expressing regular negation in preverbal position and they cannot occur in non-negative environments (that is, nonveridical contexts). Contrary to their equivalents in other Romance languages, Portuguese items are not lexically ambiguous between a strong NPI and a modal polarity item (MPI) (as is the case of Spanish) or a weak NPI (as is the case of Catalan) (cf. Martins, 2000). They occur exclusively as negative items in negative contexts.
In order to escape ambiguity, I will follow the feature system proposed by Martins (2000) for the classification of polarity items in Romance languages. This is the system that has been used to describe the evolution of other negative indefinites in Portuguese but has also been applied to other Romance languages from a diachronic perspective.
I will use the terms strong negative polarity item (strong NPI) to refer to what is frequently classified in the literature as n-words/negative concord items and weak negative polarity items (weak NPI) to refer to negative polarity items.
Martins (2000) imports the notion of underspecification from phonology (cf. Rooryck, 1994) and considers three different features for polarity items—affirmative, negative and modal. These three features may receive one of the three possible values—specified (+), non-variable underspecified (0) and variable underspecified (α). A (+) specified value signals the presence of the property represented by the feature. On the other hand, a non-variable underspecified (0) value is attributed when the property marked by the feature is absent. Finally, a feature marked with a variable underspecified value (α) means that the feature in question does not have an intrinsic value, therefore being context sensitive. This means that the feature is filled with a value from a nearby item with which it establishes an agreement relation.
This system predicts that strong polarity items have, at most, one (+) specified feature and no α-underspecified feature. Weak polarity items, on the other hand, have at least one α-underspecified feature but no (+) specified feature. This results in strong negative polarity items being [+neg, 0 mod, 0 aff], while weak negative polarity items are [α neg, α mod, 0 aff]. In practical terms, this means that strong NPIs can only be licensed in negative contexts, while weak NPIs may occur both in negative and in modal contexts. Nevertheless, none of the items is licensed in affirmative–assertive contexts, that is to say, positive declarative sentences.
Having in mind the three types of licensing contexts, negative contexts (or antiveridical contexts) will be understood to be those in which negation is expressed by means of the standard preverbal negation marker não ‘no’ or a strong NPI (for example nunca ‘never’). Modal contexts (following Bosque, 1996) are to be understood as non-negative contexts, which cannot be assessed in terms of truth or falsity of the proposition (nonveridical contexts). For Old Portuguese, these are mostly “questions, imperatives, conditionals, comparatives, the scope of modal verbs, the scope of words expressing prohibition, generic constructions, subjunctive clauses introduced by temporal connective antes que ‘before’” (Martins, 2000, p. 195).

3. Materials and Methods

Moving on to empirical data, I have extracted a total of 403 occurrences of jamais, distributed between the 13th and the 16th century. In order to collect the occurrences, searches were performed in several textual sources from each period, considering different orthographic forms for jamais (yamais, jamays, iamais, i.a.) and also including examples where the two elements—ja and mais—were written separately.2
The examples were then encoded in a database using FileMaker Pro. For each occurrence of jamais, linguistic information was added. Table 1 below summarizes the textual sources considered for each century, as well as the number of words per text and century and also the number of occurrences of jamais found in each text.
An effort was made to build a balanced sample for each century and also to ensure that the number of words per century was similar, although this was not always possible.3
It should also be stated that some of the texts, especially from the 13th and 14th centuries, may not involve consensus regarding dating issues. For instance, the two texts belonging to the Post-Vulgate cycle, namely Demanda do Santo Graal (DSG, Toledo Neto, 2012–2015) and Jose de Arimateia (JAR, Castro, 1984), are problematic sources since they are late copies. The DSG is a copy from the 15th century, but it seems to reflect the language of the original, probably written in 1245 (cf. de Vasconcellos, 1904; Castro, 1993). Studies by Castro (1993), Toledo Neto (2012), Martins (2013), and Pinto (2021) have shown that the language preserved in this text is closer to 13th century Portuguese. For this reason, I have considered it as representative of the 13th century.
On the other hand, JAR was copied in the 16th century, and some authors have pointed out changes to the original text, allegedly introduced by the scribe(s) (cf. Martins, 2013). Data extracted from this source proved problematic, and I have decided to leave it aside from the main corpus. I will not be considering the 83 examples extracted from JAR and I will only make use of them to illustrate particular features of jamais, if necessary.
Finally, Crónica Geral de Espanha (CGE, Pedrosa, 2012 & Miranda, 2013) corresponds to a late copy of the 14th century original. The edition used here reproduces manuscript L., the older of the two remaining manuscripts. Cintra (1951–1990) considers that manuscript L. was reproduced in the first quarter of the 15th century and is closer to the original text from 1344 (the original manuscript, called manuscript Y by Cintra (1951–1990), was lost). I, once again, support my decision to include it in the 14th century sample, based on the works by Martins (2013) and Pinto (2021).
The information in Table 1 clearly shows that, despite a similar number of words for each century, there is a predominance of occurrences of jamais in the 13th century sample, especially in the two Post-Vulgate texts. This may be related to text typology, with chivalry romance favouring the occurrence of jamais, but it may also be a reflex of French influence, since both texts are translations from the French original.4 In any case, this is not a particularity of jamais, since previous works with other NPIs (weak or strong) have shown that these texts are quite rich in such elements (cf. Pinto, 2021, 2024).

4. Jamais in Contemporary Portuguese

In contemporary European Portuguese, jamais ‘never’ is a temporal adverb that behaves similarly to other negative indefinites such as nada ‘nothing’, ninguém ‘no.one’ or nunca ‘never’.
Jamais is considered equivalent to the strong NPI nunca ‘never’, given their contexts of occurrence and their temporal semantics. Both jamais and nunca are negative temporal adjuncts (cf. Móia, 2024) that convey temporal reading, but they are also strong negative polarity items (in the sense of Martins, 2000). This means that they are only licensed in negative contexts. They are able to express sentential negation on their own, in pre-verbal position, as in (1), but they may also appear in post-verbal position, in negative concord constructions as in (2).
(1)Nunca/ jamaistefariamal!
neveryou.2sg.datwould.do.1sgharm
‘I would never hurt you!’
(2)Nãoregressareinunca/jamaisacasa.
negwill.go.back.1sg.Futnevertohouse
‘I will never go back home!’
In Contemporary European Portuguese, neither nunca nor jamais has an existential interpretation; therefore, they do not occur in modal contexts5 such as interrogatives (3). They cannot have a positive reading in affirmative–assertive sentences either, as illustrated in (4).
(3)* Foste nunca/jamaisaRoma?
go.2sg.PastnevertoRome?
‘Did you ever go to Rome?’
(4)*Voununca/jamaisaRoma.
go.1sg.PresnevertoRome.
The sentence in (3) is considered ungrammatical (the asterisk marks ungrammaticality) since both nunca and jamais cannot have an existential reading equivalent to ever.6 On the other hand, the declarative sentence in (4) cannot license these two items, and a positive reading with exclusive temporal interpretation is ruled out as well.
Despite the ungrammaticality of (3) in contemporary data, there are particular contexts that still allow nunca and jamais to occur without a negative reading. The work by Móia (2024) shows that nunca and jamais are not totally interchangeable, precisely due to the contrast found in constructions involving degree, namely, comparative and superlative clauses, where both items allow an existential reading equivalent to ‘ever’. Móia (2024) concludes that both nunca and jamais can occur in comparative clauses, although nunca is much more frequent. However, nunca is restricted to comparatives without a verbal predicate, while jamais is not accepted in such cases and demands a lexical verbal predicate. See the contrast between (5) and (6), with the asterisk marking ungrammaticality.
(5)Sinto-memelhordoque*nunca/jamaissenti.
feel.1sg.pres.me.1sg.reflx.Presbetterthan(n)everfeel.1sg.past
‘I feel better than ever.’
(6)Sinto-memelhordoquenunca/* jamais.
feel.1sg.pres.me.1sg.reflx.Presbetterthan(n)ever
‘I feel better than ever.’
On the other hand, nunca is not allowed in superlative clauses, while jamais seems to be accepted by some speakers (although marginally). Example (7), taken from Móia (2024), illustrates this contrast, showing that, in superlative constructions, jamais may be equivalent to alguma vez ‘ever’, which is the most frequent and natural alternative.
(7)Esta é a ponte mais longa {OKjamais / OKalguma vez / *nunca} construída entre duas ilhas.
‘This is the longest bridge ever built between two islands.’
(Móia, 2024, p. 248, ex. 87)
The ungrammaticality of nunca in superlative constructions and its occurrence in comparative clauses is an intriguing contrast that may be explained diachronically. In early stages of the language, there was a modal polarity item (MPI) nunca that did not persist in the language. Pinto (2024) shows that the modal item nunca had an existential interpretation equivalent to ‘ever’ and was frequent in comparatives, superlatives and other modal contexts, but was not preserved in the language. The fact that we can still find nunca with an existential reading only in comparative clauses suggests that this specific context still presents a frozen instance of the MPI nunca. The acceptability of jamais in superlative constructions is yet to be explained, though. We hope to be able to contribute to the discussion with data presented in the following sections.

5. Jamais in Old Portuguese

5.1. A Few Considerations on the Uses of já and mais

As was mentioned before, jamais results from the frequent combination of two different adverbs inherited from Latin: ‘now/already’ (<iam Lat.), a temporal–aspectual adverb, and mais ‘more’ (<magis Lat.), an intensity adverb.
In early texts, these two items occurred in multiple contexts (negative, modal or affirmative) and independently from one another. They were both preserved in the language until nowadays, along with the negative temporal adverb jamais.
The adverb ‘now/already’ had a larger spectrum of occurrence in earlier stages. Examples (8) to (11) illustrate its use in association with other adverbs, as a sort of reinforcement strategy, with aspectual reading.
(8)Mas,poisjaassihequeelles
butthennowthis.waybe.3sg.Presthatthey
sabenpervosoquelheeu
know.3pl.Presbyyou.2plthewhathim.3sg.datI
queriadizer […]
want.1sg.Impsay
‘But, they already know from you what I wanted to tell them.’
(Pedrosa, 2012 & Miranda, 2013, Crónica Geral de Espanha, 14th cent.)
(9)[…]equerojasempreservuiraDeospor
[…]andwant.1sg.PresnowalwaysservetoGodby
ellesepormỹ.
theyandbyme
‘and I want to serve God forever, for them and for me.’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
(10)Ca,comoquerqueofilhodeDeus
becausehowwantthatthesonofGod
resurgissedamorteaavidaejanunca
return.3sg.Imp.Subjof.thedeathto.thelifeandnownever
possamorrer
candie
‘Because, however the son of God came back to life and he can never die.’
(A. F. Machado, 2013, Diálogos de São Gregório, 14th cent.)
(11)[…]quandoosnossoschegarãojaa
[…]whentheoursarrive.3pl.Pastnowneg
poderãoallcãçarsenãseysmouros
could.3pl.Pastreachexceptsixmoors
‘[…] when ours arrived, they could no longer reach more than six moors.’
(Brocardo, 1997, Crónica do Conde Dom Pedro de Menezes, 15th cent.)
The examples above illustrate the frequent association of with the adverbs assim ‘this way’, sempre ‘always’ and nunca ‘never’ and with the standard sentential negation marker não ‘no’ (as an inchoative strategy). Apart from these adverbs, also occurred associated with other elements, in the expressions já quanto ‘a little’, já quê ‘a bit/anything’, já que quer ‘something’, among others. These occurrences of with quanto, quê and que quer are no longer possible in contemporary Portuguese.
As for the adverb mais ‘more’, apart from the uses in comparative clauses, it was also registered in association with nunca ‘never’, as in (12), but also as a Verbal Phrase (VP) modifier, in negative sentences with the standard negation marker não ‘no’, as in (13). In both contexts, it worked as an intensifier. Both occurrences remain possible in contemporary Portuguese.
(12)Eassiolevárãoperasuacasa,
andthis.wayhim.3sg.acctake.3pl.Pasttohishouse
quenuncamaisabrioosolhosnem
thatnevermoreopen.3sg.Pasttheeyesnor
moveopeenemmão.
move.3sg.Pastfootnorhand
‘And therefore, they took him home, since he never again opened his eyes nor moved a foot nor a hand.’
(Castro, 1984, José de Arimateia, 13th cent.)
(13)-Nademanda,disseocaualeyro,nomuos
in.thequestsay.3sg.Pasttheknightnegyou.2pl.reflx
metadesmais[…]
put.2pl.Pres-Subjmore
‘In the quest, said the knight, do not involve yourself anymore.’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
In early texts, the co-occurrence of both ja and mais in the same sentence was frequent and seemed to be a reinforcement strategy to express the continuation or end of a state of events, with a defined starting point. By combining the aspectual feature of ja with the intensity/reinforcement conveyed by mais, the interpretation of the sequence ja mais7 would literally be ‘from now on… more’. In sentence (14), for example, sets the moment from which an event (auera honrra ‘to have honour’) will no longer repeat itself.
(14)[…]edissequejamaisauerahonrra
[…]andsay.3sg-Pastthatnow.morenegwill.havehonour
‘and he said that from that moment on he will have no more honour’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
The combination of ja with mais does not seem any different from the combination of with other adverbs. However, this is the only combination that will acquire the status of a construction and, in due course, of an independent lexical unit.8
It should be noted that the two adverbs can still co-occur in the same sentence, but, in such cases, they are associated with negation and to the combination of inchoative and comparative strategies (as in van der Auwera, 1998).

5.2. Jamais and Its Relation with Polarity

In the previous section, it became clear that both adverbs and mais could occur in any context alone, without polarity constraints. However, when we look at the construction ja mais, we see that polarity plays an important role in its behaviour and interpretation.
Let us first start by looking at the distribution of the 321 examples found in our corpus, in terms of the polarity context in which jamais appears.
Table 2 clearly shows that there is a predominance of examples of jamais in the 13th century (as mentioned in the introduction), when compared to the following centuries. In fact, the frequency found in the 13th century drops drastically in the 14th century data and is residual in the next two centuries. This may be due to text typology, but it does not fully explain the scarcity of cases of jamais from the 14th century onwards. We can hypothesize that two other factors may have influenced this outcome: i) the progressive disappearance of jamais from affirmative and modal contexts narrowed down its occurrence to negative sentences only; ii) after the 15th century jamais will become intrinsically negative, facing competition from nunca, which was already a strong NPI in the 13th century.
Another particularity of the data collected is related to the strong association of jamais with negative contexts. These are the most frequent contexts of occurrence of jamais in all centuries, even in the earlier attestations. Despite that, and contrary to contemporary data, jamais was actually found in all polarity contexts, as exemplified from (15) to (17), where we find jamais in modal, affirmative–assertive and negative contexts, respectively.
(15)cujdadesuosquejamaiseupossa
think.2pl.Presyou.2plthatnow.moreIcan
ueeracompanhadamessaterreal
seethecompanyof.thetableearthly
redõdaassiasuadacomouyem
roundthis.wayreunitedassee.1sg.Pastin
diadePitecostes?
dayofPentecost
‘Do you believe that I will ever see again the company of the round table reunited just like I did on Pentecost day?’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
(16)Maissempreseiãfirmeseestauesperasempre
Butalwaysbe.3pl.Pres-Subfirmandstableforalways
iamays.
nowmore
‘But always be firm and stable for ever and ever.’
(Xavier, n.d., Corpus CIPM, Documentos da Chancelaria de Afonso II, 13th cent.)
(17)Vemacaamígaquejamaisnomsofrerás
come.2sg.Impherefriendthatnow.morenegsuffer.2sg.Fut
destapena.
of.thispunishment
‘Come here friend, as you will never suffer such punishment’
(Xavier, n.d., Corpus CIPM, Vidas de Santos, 14th cent.)
I start by looking at example (15), which is not possible in contemporary Portuguese, but is still attested in Spanish (cf. Arboleas et al., 2020, p. 82).
In (15), jamais is in the scope of an interrogative clause,9 which is considered a modal (or nonveridical) context capable of licensing weak NPIs and modal polarity items (MPIs). As the term suggests, MPIs are only licensed in modal contexts.
The meaning conveyed by jamais in such cases is equivalent to ‘ever’. There are only 15 examples of jamais in a modal context with existential interpretation, the majority found in texts from the 13th century.10 They all correspond to contexts that have been identified as licensers of weak NPIs in Old Portuguese (cf. Martins, 2000; Pinto, 2021, 2024), with interrogative, completive and conditional clauses being the most frequent cases.
Contexts as the one in (15) also occur with the MPI nunca ‘never’ in Old Portuguese. According to Pinto (2024), the MPI nunca was licensed in modal contexts and had an ‘ever’ reading, but it disappeared from the language, being replaced by the expression alguma vez ‘any time/ever’. If we compare the use of jamais in (15) with nunca in (18), we see that the interpretation is equivalent. In fact, none of the 15 modal contexts displays jamais co-occurring with the MPI nunca, which suggests that they competed for the same function in such cases (but this is not true in negative contexts until jamais becomes a strong NPI).
(18)Quẽcreryanucaamỹsẽ
whobelieve.3sg.Condnevertomewithout
testemoynhadeJhesuChristo,seeu
witnessofJesusChristifI
quisessedizerqueasespinhaseră
want.1sg.Pres.Subsaythatthefishbonesbe.3pl.Imp
riquezas
riches
‘Who would ever believe me without a Christian witness if I wanted to say that the fishbones were riches?
(A. F. Machado, 2013, Diálogos de São Gregório, 14th cent.)
Against expectations, jamais only appears once in a superlative construction. Given its marginal acceptance in such contexts in Contemporary Portuguese, we would expect to find more occurrences of jamais with an existential reading in a superlative clause. Nevertheless, the overall number of modal contexts is low, so we cannot draw solid conclusions from here.
On the other hand, example (16) presents jamais co-occurring with the adverb sempre ‘always’, in an affirmative–assertive context. In Contemporary Portuguese, the sequence ‘sempre jamais’ is no longer possible since these two items—jamais and sempre—are semantically incompatible, conveying opposite temporal interpretations. There are 34 examples of jamais in an affirmative–assertive context in the corpus. Twenty-five of those examples present jamais in association with sempre ‘always’, usually in the expression para (todo) sempre jamais ‘for ever and ever’. This indicates that, not only was jamais used as a reinforcement strategy of the temporal adverb sempre, but it was also empty from its contemporary negative semantics associated with ‘no point in time’. In fact, in the remaining nine examples, jamais occurs in the Prepositional Phrase (PP) por jamais ‘forever’, displaying the exact same meaning of sempre ‘always’, as illustrated in (19):
(19)[…]etcumJudasotraedordenostroSenor
[…]andwithJudasthe.traitorofourLord
seyadanadoporjamaysnoInfferno
be.3sg.Pres-Subcondemnedofnowmorein.thehell
‘and along with Judas, our Lord’s traitor, may he be condemned forever to hell’
(Barreiro, 2006–2018, Corpus Xelmírez, 14th cent.)
In all of these cases, jamais has a purely temporal–aspectual interpretation, being equivalent to ‘forever’ when it occurs in the PP por jamais, or reinforcing the adverb sempre ‘forever’, in para sempre jamais. It should also be noted that, contrary to modal and negative contexts, where jamais appears as the head of an adverbial phrase, in affirmative–assertive contexts, it never occurs on its own. It is always registered as an adverbial inside a PP.
Similar to what happens with jamais in modal contexts, occurrences of jamais meaning ‘forever’ in similar expressions are also attested for Old Spanish jamás (cf. Rueda Rueda, 1997) and Old French jamais (cf. Hansen, 2012). For Old Spanish we find the PP por jamás, but, contrary to Portuguese data, the use of jamás alone as an adverbial adjunct is also registered as in the example [m]i vida será jamás amarga ‘my life will always be bitter’ (cf. Rueda Rueda, 1997, p. 129, ft. 20). Also, Hansen (2012, p. 82) notes the existence of a frozen expression à/pour jamais with the same meaning of the Portuguese and Spanish counterparts.
The fact that at least these three languages presented a positive item jamais meaning sempre is an argument in favour of the non-negative initial semantics of the construction. It also makes its appearance in a very early stage of Romance languages.
Finally, example (17) shows jamais in a negative context. In such cases, jamais co-occurs with a negative element: the sentential negation marker não ‘no’, the strong NPI nunca ‘never’ or the negative conjunction nem ‘nor’, but in a residual number. There are no occurrences of jamais with a negative indefinite (such as nenhum, ninguém or nada) in negative sentences, which is expected, since these items were not intrinsically negative in early stages and could not convey negation on their own. They were always accompanied by a negative element.
The occurrence of jamais with the sentential negation marker não ‘no’ is the default pattern in negative sentences and it is still possible nowadays, when jamais is postverbal. However, the association of jamais with nunca is not productive in contemporary data and is considered marginal or ungrammatical. In contrast, nunca is present in 24% of the negative sentences with jamais in the corpus. In these cases, nunca is the only negative element in the sentence, responsible for sentential negation. Contrary to what we have seen in the combination of jamais with the adverb sempre, the word order between the two items is flexible. We find both orders, nunca jamais and jamais nunca, and also cases where the two adverbs are not in adjacency. These three scenarios are illustrated in (20) to (22), respectively.
(20)-Se emestalideentrarmos,nuncajamais
-ifinthisfightgo.in.2pl.Fut.Subjnevernow.more
tornaremosaCastella.
go.back.2pl.FuttoCastella
‘- If we go into this fight, we will never get back to Castella again.’
(Pedrosa, 2012 & Miranda, 2013, Crónica Geral de Espanha, 14th cent.)
(21)Jamaisnucas(er)amcorrutosnẽdesenparados
now.moreneverbe.3pl.Futcorruptednorhelpless
‘They will never again be corrupted nor helpless’
(Xavier, n.d., Corpus CIPM, Vidas de Santos, 14th cent.)
(22)[…] eporendejuraronquenuncaa
andfor.thatswear.3pl.Pastthatneverto
crischãos jamaisroubas[s]en,esequitas[s]en
christiansnow.moresteal.3pl.Imp.Subjandse.Reflxquit.3pl.Imp.Subj
daquelafolia
of.thatmadness
‘and for that they swore they would never steal Christians again and they would quit the madness’
(Barreiro, 2006–2018, Corpus Xelmírez, Lírica Profana Galego Portuguesa, 13th cent.)
The frequent combination of jamais with nunca poses a question: was jamais a reinforcement element of the adverb nunca? I argue that it was not. If jamais were a reinforcement particle of nunca, we would expect it to also occur in non-negative contexts. As mentioned before, in Old Portuguese, nunca was also an MPI with existential interpretation. However, there are no attestations of jamais co-occurring with it, in modal contexts. The fact that jamais combined only with the NPI nunca in early texts but stopped occurring with it after becoming a negative item (their combination is ungrammatical nowadays) confirms that the two adverbs had different features. Jamais was still mostly a temporal–aspectual construction that did not compete with the NPI nunca for a similar function yet. In fact, nunca played the role of a sentential negation marker, just like não ‘no’. The apparent reinforcement value of jamais in negative sentences is actually due to the fact that both and mais were (and still are) phasal adverbs. I will come back to this topic.
At least until the 15th century, jamais displayed a ‘from this moment on… no longer’ interpretation and was therefore still temporal–aspectual. However, in the 15th century, we find the first occurrences of jamais with a negative interpretation and as the single negative item in the clause. These contexts are scarce in the corpus, though, with two cases in the 15th and another two in the 16th century. In (23), jamais occurs in preverbal position, without any other negative item and with the temporal negative reading of ‘never’ it has nowadays.
(23)[…] eforamnossoscoraçõestamquebrados,que
[…] andbe.3pl.Pastourheartssobrokenthat
jamaisousamosvolverrrostocomtravos.
neverdare.1pl.Pastturnfaceagainstyou.2pl
‘and our hearts were so broken that we never dare to turn our face against you’
(Brocardo, 1997, Crónica do Conde D. Pedro de Menezes, 15th cent.)
The interpretation of jamais as a negative item meaning never is also suggested by example (24), where it occurs as the second element of the expression nunca, jamais, em tempo algum ‘never, never, in no time’, which presents the three negative temporal items available in Portuguese, ordered by emphatic force. This expression is frequently used to reinforce the idea that an action or event will never take place. In (24), jamais can be understood as an equivalent of nunca ‘never’ and is therefore not interpreted compositionally.
(24)EderomaoditoGonçaloVaaz,conprador,
andgive.3pl.Pastto.thesaidGonçaloVaazbuyer
porqujteliureeassuamolher
byexemptfreeandthehiswife
e/herdeirosperassenprequenuca,jamais
andheirsforalwaysthatnevernever
emtenpoalguu,ssejomdemandadoporello.
intimesomebe.3pl.Pres.Subaskedbyhim
‘And they considered the aforementioned Gonçalo Vaaz, buyer, his wife and heirs, free of debts forever, and that he shall never, ever, in no moment in time, be charged by him.
(Martins, 2001, Documentos Portugueses do Noroeste e da Região de Lisboa (Chelas), 15th cent.)
The appearance of the first cases of jamais apparently as a strong NPI in the 15th century is in line with what is described for Spanish jamás. Yamada (2023, pp. 11–12) also indicates the same chronology for the first attestations of jamás with an entirely negative meaning and equivalent to nunca.11 Anyway, given the scarcity of the examples of jamais as the only negative item in the sentence, we can only state that its status as a strong NPI cannot be prior to the 15th or 16th century.
As we have seen so far, the primary value of jamais is that of a temporal–aspectual adverb, even in negative contexts. Although the primary temporal interpretation is futurity, the use of future tense is more expressive in negative contexts. Table 3 shows the frequency of the future, past and present tenses, according to polarity.
Although the number of occurrences is quite uneven between affirmative and modal contexts compared to negative ones, it is clear that some tenses are favoured in specific polarity contexts. In negative contexts, jamais is highly associated with the expression of futurity, with future and conditional tenses representing more than half of the cases. On the other hand, affirmative–assertive contexts highly favour the use of present tense, due to the nature of declarative clauses, which are usually associated with the expression of facts, generalizations or axioms. Finally, modal contexts seem to favour both present and past tenses, although the low number of examples prevents us from drawing more solid conclusions. In any case, the subjunctive mood is preferred to the indicative mood, which is expected in modal contexts, since the subjunctive mood can license weak NPIs and MPIs, being associated with irrealis interpretations.
In the next section, I will present syntactic evidence to back up the idea that jamais started as two independent adverbs and its independent status as a lexical unit is coincident with its new negative feature.

5.3. The Status of Jamais: A Construction or an Independent Lexical Unit?

Although jamais is an independent lexical item that behaves as a strong NPI in contemporary data, early attestations show that it started as two non-polar adverbs that occurred in any polarity context with a temporal–aspectual interpretation. In this section, we will try to assess its level of internal syntactic analyzability in order to understand when it became a lexical unit.
Our hypothesis is that, in a first stage, and mais were a frequent collocation that evolved into a construction, visible in the first texts, and finally it was reinterpreted as a lexical unit.
According to Jiménez Juliá (2017, p. 100), a string of words that is not yet a unit may be part of a productive model and will have internally recognizable parts. This means that internal word order can be seen as an argument in favour of the initial independence of the two adverbs.
Looking at the examples, we conclude that the predominant pattern was already the one where preceded mais, as in (25), which is found in all the examples, except in three cases from the 13th century. Example (26), however, shows the possibility of having the order mais já.
(25)[…]enuncafolgavasenonquandoafraqueza
[…]andneverrest.3sg.Impexceptwhentheweakness
eratamanhaquenonpodiajamaisandar
be.3sg.Impsuchthatnegcouldalreadymorewalk
‘and he never rested except when the weakness was such that he could no longer walk’
(A. F. Machado, 2013, Diálogos de São Gregório, 14th cent.)
(26)Deffendemosquenenhuuuozeyrononseya
defend.2pl.Presthatnoneattorneynegbe.3sg.Pres.Subj
ousadodeauirsseestaquelde
boldtoconciliatethisthatof
queádeteruoztenha
whohave.3sg.Futvoiceneghave.3sg.Pres.Sub
maysyauozporoutro
morealreadyvoiceforother
‘We defend that no attorney should be bold to conciliate this, that whomever he defends should not be defended by any other henceforth.’
(Xavier, n.d., Corpus CIPM, Foro Real, 13th cent.)
In (26), the two adverbs co-occur in adjacency, but the internal order is reversed. This corroborates the idea that jamais was not a unit yet and both orders já mais ‘now more’ and mais já ‘more now’ were possible in early attestations. Nevertheless, the order mais já is only found in negative contexts. This may be a consequence of the limited number of examples of modal and affirmative–assertive contexts, but it can also suggest that the construction evolved at a different pace in negative contexts.
Another good indicator of some level of syntactic independence of the two items in the construction is the existence of interpolation (cf. Jiménez Juliá, 2017). The possibility of having lexical material between the two adverbs of the construction is an argument in favour of the syntactic independence of the parts.
Examples (27) and (28) show the possibility of the two adverbs being split by other constituents in a sentence. These occurrences are, however, infrequent, pointing to an increasing fixed status of the construction with jamais. From a universe of 321 examples, only 5 exhibited interpolated elements between and mais.
(27)Mayschegadoheodíaquesuayrmãa
butarrivedbe.3sg.Presthedaythathissister
Casandraaujáprophetizado,etjassepodía
Casandrahave.3sg.Imppredictedandnownegit.reflxcould
maysperlongar.
moreextend.
‘But the day his sister Casandra predicted has arrived and it could no longer be extended.’
(Barreiro, 2006–2018, Corpus Xelmírez, Cronica Troyana, 13th cent.)
(28)[…]esemenonvalDeus
[…]andifme.1sg.Reflxneghelp.3sg.PresGod
(quemi-amostre!),janonguarriaeu
thatme.1sg.dat-her.3sg.accshownownegprosper.1sg.ImpI
maisnomundo
morein.theworld
‘and if God does not help me (by showing her to me!), from now on, I will no longer prosper in the world.’
(Barreiro, 2006–2018, Corpus Xelmírez, Lírica Profana Galego Portuguesa, 13th cent.)
In (27) and (28), the two adverbs are separated by lexical material, showing that they were not an independent unit yet. Nevertheless, the scarcity of these examples, which appear only in the 13th and 14th centuries, indicates the progressive loss of independence of the two components. Adjacency and fixed word order were fundamental for the reanalysis of jamais as a unit since adjacency is required for the merger of the two adverbs.12
I introduce here an additional argument to support the idea that, in early occurrences, jamais was still a construction. While contemporary jamais ‘never’ cannot be interpreted compositionally, early occurrences of jamais could be interpreted by the sum of the parts, with each adverb modifying the VP.
In modal and negative contexts, when and mais co-occured, the adverb mais can be seen as an optional pseudoargument “indicating the extent or degree to which the predicate holds”, as proposed by Breitbarth et al. (2020, p. 50) when looking at optionally transitive verbs and alike. These contexts are usually considered ‘bridging contexts’, since they allow ambiguous readings, leading to the emergence of a more grammaticalized form (cf. Lucas, 2007).
Sentences (29) and (30) allow mais to be interpreted as the optional pseudoargument of the verb, expressing extent or degree, despite the adjacency to .
(29)-Padresenhor,seẏsemeueeresiamais.
fatherlordnegknow.1sg.Presifme.1sg.accseenow.more
‘-My father, I don’t know if you will ever see me again.’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
(30)[…] edesentonnonbraadoujamaisoenfermo […]
[…] andsincethennegscream.3sg.Pastnow.morethesick man
‘and since then, from that moment on, the sick man no longer screamed’
(A. F. Machado, 2013, Diálogos de São Gregório, 14th cent.)
In sentence (29) the adverb mais can be interpreted as an extent/degree of the verb ver ‘see’, producing an interpretation equivalent to “ever… again”. Both the modal context and the use of conditional tense contribute to create an irrealis interpretation that eventually translated into ‘ever’. An equivalent scenario can be hypothesized for sentence (30), with mais being the extent/degree of the verb bradar ‘scream’ and translating into an interpretation of the type ‘from this moment on….no longer’.
The possibility of mais being interpreted as a pseudoargument of the verb also shows that the two parts of the construction were still syntactically independent, but ambiguity tends to disappear with jamais becoming a less free construction. The so-called ‘bridging contexts’ are said to be involved in the reanalysis of other non-negative items as negative polarity items across Romance languages (cf. Roberts & Roussou, 2003; Willis et al., 2013).
One final remark that shows the gradual loss of independence of the two items is related to tense features. As we have seen in Table 3, jamais was favoured in future tense contexts, future and conditional being the most frequent tenses in sentences with the construction. However, when we look at the distribution of future tenses across centuries, we see that the association of jamais with a future or conditional verb decreases from the 13th century onwards, while past tenses tend to increase, as well as infinitives and gerunds (considered in the category Other) (cf. Figure 1). The association with present tenses remains steady, though. This same evolution is reported by Hansen (2012) for the French jamais, which points to a more general tendency of these Romance constructions. The widening of its contexts of use may be seen as an indicator of the evolution of the construction into a unit.
The data seen so far seems to corroborate the idea that, in the 13th century, jamais was a construction with syntactic independence, but its parts did not occur freely. Internal word order was already established and interpolation rarely occurred. Also, although the two adverbs could have their own syntactic function, mostly in relation to the VP, this possibility was available only in modal and negative contexts, since affirmative–assertive sentences never displayed jamais at VP level.

5.4. Phasal Adverbs and No-Longer Expressions: The Origin Source of jamais?

In the previous sections, it became clear that jamais was strongly associated with negative contexts, despite the fact that neither of the adverbs in its formation was a NPI.
However, both adverbs and mais share a particularity: they are considered phasal adverbs (cf. van der Auwera, 1998) and integrate phasal polarity expressions (cf. Vaquer, 2021). According to Vaquer (2021, p. 2), these expressions “encode the existence or non-existence of a situation in reference time compared to a preceding time, i.e., the situation’s continuation or discontinuation”.
In this section, I will argue that the association of and mais as phasal adverbs in negative contexts is in the origin of jamais as an NPI.
In his work on phasal adverbs, van der Auwera (1998) draws attention to the Latin inchoative adverb iam associated with non as a way to express discontinuation, that is to say, a no-longer reading. van der Auwera (1998) considers that languages have basic strategies to form no-longer expressions (or ‘inchoative discontinuatives’) with a positive element meaning already (an inchoative element) combined with a negator.13 However, it is also possible to have a combined strategy involving two positive elements combined together with a negation.
van der Auwera (1998) puts forth the idea that the evolution of no-longer expressions is unidirectional and claims that the inchoative strategy inherited from Latin is being replaced by the comparative strategy with forms of magis in Romance languages. Spanish is said to be the only Romance language to still keep the inchoative adverb to express ‘inchoative discontinuative’ values, while languages such as Portuguese are at an intermediate stage, allowing both inchoative and comparative elements. On the other side of the evolution is French, where only the comparative strategy with plus is possible. This hypothesis was rejected by Vaquer (2021, p. 29), who considers that there is no evidence to conclude that Spanish only uses a purely inchoative strategy or that Portuguese, along with Galician and Catalan, is at an intermediate stage towards the adoption of a comparative strategy. Nevertheless, what is interesting in van der Auwera (1998)’s proposal is the fact that he predicts a period of transition from an inchoative to a comparative strategy where the two items combine. This is what is documented for French, Spanish and Portuguese, with and mais being the positive elements co-occurring in negative sentences, with a reinforced no-longer interpretation.
A context like the one in (31), where the inchoative and the comparative strategy coexist in association with a negative element, allowed for the formation of a construction that became negative. Although jamais also occurs in affirmative-assertive and modal contexts, only the presence of negation could have allowed the addition of a negative feature to a purely temporal element.
(31)[…] ediseromquejamaisnomquedariamd
andsay.3pl.Pastthatnowmorenegstop.3pl.Condof
andarataaqueujssematalmessa.
walkuntilthatsee.3pl.Imp.Subjthe.onetable
‘and they said that they, from that moment on, they would not stop walking until they saw the table’
(Toledo Neto, 2012–2015, Demanda do Santo Graal, 13th cent.)
The original meaning of a sentence like (31) is that an action will be discontinued, starting in a certain moment, which is signalled by the presence of . From this idea, there was a widening of the temporal timespan until reaching a ‘no moment in time’ interpretation. On the other hand, being in the scope of a negative element also contributed to the appearance of a [α neg] feature that followed the exact same path of evolution of other Portuguese n-words, becoming intrinsically negative, therefore [+ neg]. This explains the temporary occurrence of jamais in modal contexts and its specialization in negative environments.
Considering Spanish jamás, Llorens (1929, p. 75) stated that it started as a temporal adverb that expressed future temporal negation, with its value sometimes getting mixed up with the perpetuity conveyed by nunca ‘never’. We can assume that Portuguese jamais evolution was also favoured by its association with this value of perpetuity.
Data from Portuguese shows that van der Auwera’s prediction on the disappearance of the inchoative strategy is still not visible in today’s data. This may indicate that the change is still in progress, but there is no way of knowing what the future outcome will be. Contemporary Portuguese still keeps both inchoative and comparative strategies with and mais, and also the possibility to combine the two in the same utterance. These three possibilities are illustrated from (32) to (34).
(32)- Jánãofumo.
nownegsmoke.1sg.Pres
‘I no longer smoke’
(33)- Nãofumomais.
negsmoke.1sg.Presmore
‘I don’t smoke anymore’
(34)- Jánãofumomais.
nownegsmoke.1sg.Presmore
‘I don’t smoke anymore.’
Interestingly, configurations as the one in (34) are rare in diachronic data. We only found five cases of and mais occurring separately. There seems to have been a tendency for and mais to start appearing in adjacency, but that frequent collocation resulted in a construction with different temporal and polarity features, that is, the NPI jamais. The temporal–aspectual reading of as meaning ‘from now on’ is progressively lost and is totally absent from the NPI jamais, as we can see in the contrast between (34) and (35).
(35)- Nãofumojamais.
negsmoke.1sg.Presnever
‘I don’t ever smoke.’
While in (34) it is implied that the person was a smoker before the enunciation moment, and the event of not smoking will only occur from that point onwards, in (35) we assume there was never a time when the person smoked.
At this point, it is clear that jamais as the result of inchoative + comparative strategy and the strong NPI jamais are two different elements. However, at least until the 15th century, all the occurrences found for jamais in negative contexts still correspond to the combination of the inchoative and the comparative elements. The strong NPI jamais emerges from the 15th century onwards, when we find its first occurrences expressing negation on its own. In the corpus, there are no examples of the combination of the inchoative + comparative strategy apart from the ones where ja and mais appear in adjacency (with the five exceptions mentioned before). We hypothesized that the first pattern of combination of the inchoative and the comparative mais started with the two elements being apart. However, they eventually started occurring in adjacency, which allowed for the formation of a new lexical item, the strong NPI jamais. As soon as jamais becomes an independent item with different semantic and syntactic features, the old pattern to combine and mais is restored. From the 16th century onwards, we start finding cases again of negative sentences with the adverbs and mais without adjacency, as in (36).14 This example shows that the strategy to combine inchoative and comparative elements was never abandoned. Both the strong NPI and the combination of and mais to form no-longer expressions coexist nowadays, but the pattern found in the latter structure does not exhibit adjacency of the elements and mais as seen in Old Portuguese. In (36), the word order found is já + neg + verb + mais.
(36)VMqueixadequelhenão
yougive.3sg.Prescomplaintofthatyou.3sg.Datneg
deinotíciasdecoisaalgumadavida
give.1sg.Pastnewsofthingsomeof.thelife
dasenhoradonaJoanaseirdesta
of.theladymissJoanase.Reflxgofrom.this
casa,pornãosepoderaturar
housefornownegse.Imperscanendure
maisassuasdiabruras
moretheherdevilries
‘You complaint that I didn’t give you news of anything about miss Joana’s life and the fact that she left this house because no one could endure her devilries any longer.’
(Marquilhas, 2014, Corpus Post Scriptum, 18th cent.)
It is necessary to investigate more data from the 16th century in order to confirm this idea. I remit this task to future work.

6. Conclusions

In this paper, I focused on the origin and evolution of the strong NPI jamais in Portuguese. This is a negative temporal indefinite that resulted from the Latin expression iam magis. Data from the period between the 13th and the 16th century showed that jamais started as two independent adverbs, and mais, which frequently co-occurred with a temporal–aspectual semantics and had a positive reading.
Nevertheless, they were highly associated with negative contexts, due to the fact that they were both phasal adverbs, used in no-longer expressions. The combination of the inchoative strategy with and the comparative strategy with mais in negative contexts seems to have opened the door for the reinterpretation of the two adverbs as a frequent construction that resulted in a new lexical item with a negative feature and different temporal information.
Although jamais is now considered a strong NPI, I have shown that it only became an independent element after the 15th century, when it first starts occurring as the single negative element in the sentence and with a never reading, cancelling the possibility of being interpreted compositionally. Until then, it kept its internal analyzability, which was still attested with the existence of inversed internal word order, interpolation, frequent association with future tenses and independent syntactic function of the two adverbs. These features are progressively lost as the construction becomes a new lexical unit. Similar outcomes are also reported for Spanish of French equivalents, originated from the Latin iam magis.
The evolution of Old Portuguese jamais can be seen as an example of the relation between syntactic analyzability and the notions of lexicalization and grammaticalization, showing that these processes may be linked. The progressive loss of syntactic independence resulted in the lexicalization of jamais as an independent unit, but there were also changes in meaning and in its grammatical function. Bearing in mind the idea of grammaticalization (in the sense of Heine, 2003), we can consider that jamais exemplifies a successful case of grammaticalization, going from two adverbs to a polarity item capable of expressing sentential negation. It is, however, interesting to note that jamais did not suffer phonological erosion. In fact, in contemporary Portuguese, jamais seems to be pronounced as two prosodic words, with speakers stressing the first vowel [‘Ʒa], which suggests an intuitive awareness of the complexity of the structure, at least on a phonological level.15
The comparison between the evolution and different outcomes of Latin iam magis in several Romance languages is research that I postpone for future work and that can shed some light into the study of these NPIs.

Funding

This research was funded by Fundação para a Ciência e Tecnologia (FCT), grant number UID/214/2025, Centro de Linguística da Universidade de Lisboa.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

The author declares no conflicts of interest.

Abbreviations

Cond conditional
Futfuture tense
Imp imperfect tense
Pres present tense
Subj subjunctive mood
Reflx reflexive pronoun

Notes

1
This was first pointed out by Laka (1990, p. 108), who highlighted that n-words such as nada (Portuguese ‘nothing’) or nadie (Spanish ‘nobody’) came from the past participles nata and nati in the Latin expressions res nata and homines nati (‘born thing’/‘born man’) and did not have negative meaning.
2
In this period, orthography was not established and there was a lot of variation in the graphic forms found for each word, even within the same text. An anonymous reviewer considers that the cases where jamais is written together may constitute an argument in favour of it being a single lexical unit. However, in the first written texts, it was common for words that did not correspond to a lexical unit to be written together. This happened for many reasons (saving time and writing consumables) and may not have linguistic motivations. See example (19), where we find otraedor, corresponding to the definite article o ‘the’ and the noun traedor ‘traitor’ written together.
3
In a few isolated cases, it was not possible to determine the total number of words of the text/source.
4
An anonymous reviewer raises the question of data reliability concerning the examples of jamais taken from the DSG text and argues that these occurrences of jamais may reflect the use of Old French jamais. This hypothesis is ruled out, though. The analysis of Old French jamais (cf. Hansen, 2012) shows that it did not behave like Portuguese jamais. According to Hansen (2012), Old French jamais lexicalized as a negative marker at an early stage, only occurring in non-negative contexts much later. This analysis is not in line with our findings for Old Portuguese jamais, nor did we find the examples of jamais taken from the DSG text to behave any different from the other 13th century textual sources.
5
This is not entirely accurate, since nunca occurs in comparative clauses and jamais is marginally accepted in superlatives. This will be explained in the following paragraphs of this section.
6
A negative reading is not possible either, since postverbal nunca/jamais demands the presence of a regular negation marker.
7
I have translated all occurrences of jamais as two independent elements (‘now more’), making clear the original meaning of the expression (‘from now on…more’). The exceptions are the examples where jamais is unambiguously a strong NPI, in which cases I have translated it as ‘never’.
8
But see the case of French déjà (cf. Hansen, 2014).
9
It is simultaneously within the scope of an interrogative and a completive clause. Both clauses are licensing contexts for modal polarity.
10
There are 12 examples in the 13th century, 2 examples in the 15th and 1 example in the 16th century.
11
It should be mentioned that, as far as example (24) is concerned, we cannot discard the possibility that nunca is licensing both jamais and the expression em tempo algum ‘in no time’ or only the latter. This is so because the expression em tempo algum is only intrinsically negative when involving negative nominal inversion, a phenomenon that begins in the 18th century (cf. Martins, 2015).
12
An anonymous reviewer considers that the examples in (25) to (28) do not cast doubt on the constructional status of jamais. I consider that these examples show a progressive loss of independence and illustrate two important factors for the emergence of the unit jamais: adjacency and fixed word order. Although the two adverbs are still preserved in the language and can be combined (in the inchoative + comparative strategy), there is no longer ambiguity because, in such cases, there is no adjacency between and mais (the word order is usually + neg + verb + mais).
13
For sake of simplicity, I will refer to an inchoative strategy having in mind the use of an inchoative adverb in a negative context as a strategy to express no-longer interpretations.
14
I performed a quick search in the 17th and 18th century letters from the Post Scriptum corpus and found two examples for each. The 17th century sample has 20,000 words, while the 18th century sample has 30,000 words.
15
Despite being an independent word, the first vowel in jamais [Ʒa’majʃ] escapes the Portuguese vowel raising phenomenon and is produced as [a] (instead of the central-mid vowel [ɐ] found in unstressed syllables).

References

  1. Arboleas, S., Batllori Dillet, M., Cama, E., Cartagena, C., Cufí Vilà, N., Molino, N. D., Fuentes, A., Gómez, M., Graboleda Reyes, J., Jordan, S., & Lladó Sedó, A. (2020). La expresión de la polaridade en las lenguas románicas. Universitat de Girona. Available online: http://hdl.handle.net/10256/18095 (accessed on 16 December 2025).
  2. Baker, C. L. (1970). Double negatives. Linguistic Inquiry, 1, 16–189. [Google Scholar] [CrossRef]
  3. Barreiro, X. V. (2006–2018). Corpus xelmírez. Instituto da Lingua Galega, Universidade de Santiago de Compostela. [Google Scholar]
  4. Bosque, I. (1996). La polaridad modal. In Actas del cuarto congreso de hispanistas de Asia (pp. 7–14). Asociacion Asiatica de Hispanistas. [Google Scholar]
  5. Breitbarth, A., Lucas, C., & Willis, D. (Eds.). (2020). The history of negation in the languages of Europe and the Mediterranean, Vol II: Patterns and processes. Oxford University Press. [Google Scholar]
  6. Brocardo, T. (1997). Crónica do conde D. Pedro de Meneses de Gomes Eanes de Zurara/Edição e estudo. FCG/JNICT. [Google Scholar]
  7. Brocardo, T. (2006). Livro de linhagens do Conde D. Pedro. Edição do fragmento manuscrito da Biblioteca da Ajuda (século XIV). Imprensa Nacional—Casa da Moeda. [Google Scholar]
  8. Bybee, J. (2010). Language, usage and cognition. CUP. [Google Scholar]
  9. Camões, J. (2010). Centro de Estudos de Teatro, Teatro de Autores Portugueses do Séc. XVI—Base de dados textual. Available online: http://www.cet-e-quinhentos.com/ (accessed on 16 December 2025).
  10. Castro, I. (1984). Livro de José de Arimateia (Estudo e Edição do COD. ANTT 643) [Tese de Doutoramento em Linguística Portuguesa, Faculdade de Letras]. [Google Scholar]
  11. Castro, I. (1993). “Demanda do Santo Graal”; “Livro de Jose de Arimateia”; “Materia da Bretanha”; “Merlim”. In J. Lanciani, & G. Tavani (Eds.), Dicionario de literatura medieval galega e Portuguesa. Caminho. [Google Scholar]
  12. Cintra, L. F. L. (1951–1990). Crónica Geral de Espanha de 1344, Edição critica do texto português. Academia Portuguesa da Historia, 1951, volume I; 1954, volume II; 1961 volume III; 1990, volume IV. Ed. Fac-similada. Imprensa Nacional-Casa da Moeda. [Google Scholar]
  13. Costa, A. (1979). Os mais antigos documentos documentos escritos em português. Revista Portuguesa de História, 17, 312–321. [Google Scholar]
  14. de Vasconcellos, C. M. (Ed.). (1904). Cancioneiro da Ajuda. Edição crítica e comentada. INCM. [Google Scholar]
  15. Dias, A. (1990). Garcia de resende, cancioneiro geral. Fixação do texto e estudo. Imprensa Nacional—Casa da Moeda. [Google Scholar]
  16. Espinal, T., & Llop, A. (2022). (Negative) Polarity items in Catalan and other Trans-Pyrenean romance languages. Languages, 7(1), 30. [Google Scholar] [CrossRef]
  17. Figueiredo da Costa, S. (2017). Regra de S. Bento em português. Estudo e edição de dois manuscritos. Colibri. [Google Scholar]
  18. Frade, M. (2010). A negação em ‘jamais’. In Cadernos WGT: A negação. CLUNL. Available online: https://clunl.fcsh.unl.pt/wp-content/uploads/sites/12/2017/07/M.-Frade.pdf (accessed on 7 October 2025).
  19. Galves, C., Andrade, A., & Faria, P. (Coords.). (2017). Tycho Brahe parsed corpus of historical Portuguese. Available online: https://www.tycho.iel.unicamp.br/corpus/ (accessed on 16 December 2025).
  20. Giannakidou, A. (2002). Licensing and sensitivity in polarity items: From downward entailment to nonveridicality. In M. Andronis, A. Pycha, & K. Yoshimura (Eds.), Proceedings of the annual meeting of the Chicago Linguistics Society (=CLS) (Vol. 38, pp. 29–53). Chicago Linguistic Society. [Google Scholar]
  21. Giannakidou, A. (2006). N-words and negative concord. In M. Everaert, & H. van Riemsdjik (Eds.), The Blackwell companion to syntax (Vol. III, pp. 327–391). Blackwell. [Google Scholar]
  22. Hansen, M. B. M. (2012). On the evolution of temporal n-words in Medieval French. Language Sciences, 34, 76–91. [Google Scholar] [CrossRef]
  23. Hansen, M. B. M. (2014). Cyclicity in semantic/pragmatic change: The medieval particle ja between Latin iam and Modern déjà. In C. Ghezzi, & P. Molinelli (Eds.), Discourse and pragmatic markers from Latin to Romance languages (pp. 139–165). OUP. [Google Scholar]
  24. Heine, B. (2003). Grammaticalization. In B. Joseph, & J. Richard (Eds.), The handbook of historical linguistics (pp. 575–601). Blackwell Publishing. [Google Scholar]
  25. Huitink, L. (2005). Saying NO to Iam. Journal of Latin Linguistics, 9(2), 561–570. [Google Scholar] [CrossRef]
  26. Jiménez Juliá, T. (2017). Notas sobre locuciones y construcciones sintácticas. Cuadernos AISPI: Estudios de Lenguas y Literaturas Hispánicas, 10, 91–112. [Google Scholar]
  27. Kroon, C., & Risselada, R. (2002). Phasality, polarity, focality: A feature analysis of the Latin particle iam. Belgian Journal of Linguistics, 16, 63–78. [Google Scholar] [CrossRef]
  28. Laka, I. (1990). Negation in syntax: On the nature of functional categories and projections [Ph.D. thesis, MIT]. [Google Scholar]
  29. Langacker, R. W. (1987). Foundation of cognitive grammar. Theoretical prerequisites (Vol. 1). Stanford University Press. [Google Scholar]
  30. Llorens, E. L. (1929). La negación en el Español antíguo com referencia a otros idiomas. Revista de Filología Española, Anejo XI. Junta para ampliación de estúdios—Centro de Estudios Históricos. [Google Scholar]
  31. Lopes, G. V., Ferreira, M. P., & Pedro, M. (2011). Cantigas Medievais Galego Portuguesas [base de dados online]. Instituto de Estudos Medievais, FCSH/NOVA. Available online: http://cantigas.fcsh.unl.pt (accessed on 16 December 2025).
  32. Lucas, C. (2007). Jespersen’s cycle in Arabic and Berber. Transaction of the Philological Society, 105(3), 398–431. [Google Scholar] [CrossRef]
  33. Machado, A. F. (2013). Diálogos de São Gregório. Edição semidiplomática. UFBA. [Digitalized version]. Available online: https://repositorio.ufba.br/ri/bitstream/ri/17922/1/Edi%c3%a7%c3%a3o%20semidiplom%c3%a1tica%20DSGA.pdf (accessed on 20 June 2025).
  34. Machado, J. B. (2003). Tratado de Confissom: Chaves, 1489 (Vol. I: Edição semidiplomática, estudo histórico e informático-linguístico). APPACDM. ISBN 972-8699-29-8. [Google Scholar]
  35. Marquilhas, R. (Ed.). (2014). P.S. Post scriptum. Arquivo digital de escrita quotidiana em portugal e espanha na época moderna. CLUL. Available online: http://ps.clul.ul.pt/index.php?action=home (accessed on 20 July 2025).
  36. Martins, A. M. (2000). Polarity Items in Romance: Underspecification and lexical change. In S. Pintzuk, G. Tsoulas, & A. Warner (Eds.), Diachronic syntax: Models and mechanisms (pp. 191–219). Oxford University Press. [Google Scholar]
  37. Martins, A. M. (Ed.). (2001). Documentos Portugueses do Noroeste e da Região de Lisboa—Da produção primitiva ao século XVI. Imprensa Nacional—Casa da Moeda. [Google Scholar]
  38. Martins, A. M. (2013). Copiar o português duocentista: A Demanda e o José de Arimateia. In R. Álvarez, A. M. Martins, H. Monteagudo, & M. A. Ramos (Eds.), Ao Sabor do Texto. Estudos dedicados a Ivo Castro (pp. 383–402). Universidade de Santiago de Compostela, Servizo de Publicacións e Intercambio Científico. [Google Scholar]
  39. Martins, A. M. (2015). Ordem de palavras e polaridade: Inversão nominal negativa com algum/alguno e nenhum. Diacrítica, 29, 401–428. [Google Scholar]
  40. Miranda, S. (2013). Reconstituição do ms. L da Cronica Geral de Espanha de 1344 (2.ª parte) [Master’s thesis, University of Lisbon]. Available online: http://hdl.handle.net/10451/9403 (accessed on 16 December 2025).
  41. Móia, T. (2024). A distribuição dos adjuntos temporais negativos no português contemporâneo: Negação, concordância negativa e construções de grau. Diacrítica, 38(1), 226–253. [Google Scholar] [CrossRef]
  42. Pedrosa, M. (2012). Reconstituição do ms. L da Cronica Geral de Espanha de 1344 (1.ª parte) [Master’s thesis, University of Lisbon]. Available online: http://hdl.handle.net/10451/9924 (accessed on 16 December 2025).
  43. Pinto, C. (2021). Minimizers and the syntax of negation: A diachronic approach from European Portuguese [Ph.D. thesis, University of Lisbon]. Available online: http://hdl.handle.net/10451/55398 (accessed on 16 December 2025).
  44. Pinto, C. (2024). When ‘never’ meant ‘ever’: The polarity item nunca in Old Portuguese. Journal of Historical Syntax. [online first]. [Google Scholar] [CrossRef]
  45. Roberts, I., & Roussou, A. (2003). Syntactic change: A minimalist approach to grammaticalization. Cambridge University Press. [Google Scholar]
  46. Rooryck, J. (1994). On two types of underspecification: Towards a theory shared by syntax and phonology. Probus, 6, 207–223. [Google Scholar] [CrossRef]
  47. Rueda Rueda, M. (1997). Los terminos negativos en español: Aproximación diacrónica. Centro de Estudos Metodologicos e Interdisciplinares Universidad de Leon. [Google Scholar]
  48. Sobral, C. (Coord.). (2015). Corpus de Textos Antigos em ortuguês até 1525. Available online: http://teitok.clul.ul.pt/cta/ (accessed on 16 December 2025).
  49. Svenonius, P. (2015). Syntactic Constructions. In T. Kiss, & A. Alexiadou (Eds.), Handbooks of linguistics and communication science (Vol 1, pp. 15–24). De Gruyter Mouton. [Google Scholar] [CrossRef]
  50. Toledo Neto, S. (2012). Os testemunhos portugueses do Livro de Jose de Arimateia e o seu lugar na tradição da Estoire del Saint Graal: Colação de exemplos. In L. Mongelli (Ed.), De cavaleiros e cavalarias. Por terras de Europa e America (pp. 579–589). Humanitas. [Google Scholar]
  51. Toledo Neto, S. (2012–2015). Edição da demanda do santo graal [online edition]. Available online: http://alfclul.clul.ul.pt/wochwel/index.html (accessed on 9 June 2025).
  52. van der Auwera, J. (1998). Phasal adverbials in the languages of Europe. In J. van der Auwera, & D. P. Ó Baoill (Eds.), Adverbial constructions in the languages of Europe (pp. 25–145). de Gruyter. [Google Scholar]
  53. Vaquer, M. S. (2021). Phasal adverbials in Romance languages [Bachelor’s thesis, University of Stockholm]. [Google Scholar]
  54. Willis, D., Lucas, C., & Breitbarth, A. (2013). Comparing diachronies of negation. In D. Willis, C. Lucas, & A. Breitbarth (Eds.), The history of negation in the languages of Europe and the Mediterranean. Vol. I. Case studies (pp. 1–50). Oxford University Press. [Google Scholar]
  55. Xavier, M. F. (Coord.). (n.d.). Digital corpus of medieval Portuguese (CIPM—Corpus Informatizado do Português Medieval). Faculdade de Ciências Sociais e Humanas da Universidade Nova de Lisboa. Available online: http://cipm.fcsh.unl.pt/ (accessed on 15 June 2025).
  56. Yamada, A. (2023). An anchoring approach to the diachrony of negative concord in Spanish. Journal of Historical Syntax, 7, 1–34. [Google Scholar]
Figure 1. Frequency of future, past and present tenses in contexts with jamais across centuries.
Figure 1. Frequency of future, past and present tenses in contexts with jamais across centuries.
Languages 11 00002 g001
Table 1. Complete list of consulted texts, by century and number of words.
Table 1. Complete list of consulted texts, by century and number of words.
Cent.DateTextIDSource/EditionN.º wordsN.º occ. ‘já mais’
131214Testamento de Afonso IITACosta (1979) 1.8070
Foros de GarvãoFGXavier (n.d.)6.540
Costumes de SantarémCSXavier (n.d.)5.750
Documentos da Chancelaria de Afonso IIIDCAXavier (n.d.)17.6291
Demanda do Santo GraalDSGToledo Neto (2012–2015)212.145160
Documentos Portugueses do Noroeste e da Região de LisboaDNMartins (2001)29.8470
1280Foro RealFRXavier (n.d.)51.0223
Corpus Xélmirez (sample)XelBarreiro (2006–2018)unknown40
1280Tempos dos PreitosTPXavier (n.d.)1.3350
1269-?Textos Notariais -OxfordTOXXavier (n.d.)8.5830
Cantigas medievais galego-portuguesas (seleção)CMGPLopes et al. (2011)68.20021
1248-1284Vida e Milagres de Santa Senhorinha de Basto (G1)VMSSBSobral (2015)12.4830
415.333225
14 Crónica Geral de EspanhaCGEPedrosa (2012) & Miranda (2013)141.97218
Corpus Xélmirez (sample)XelBarreiro (2006–2018)unknown1
1380Livro de Linhagens do Conde D. PedroLLCPBrocardo (2006)31.7600
Vidas de Santos de um manuscrito alcobacenseVDSXavier (n.d.)31.96016
Costumes de SantarémCSXavier (n.d.)33.6900
ca. 1350Primeira PartidaPPXavier (n.d.)174.7097
1326-1375Diálogos de S. GregórioDGA. F. Machado (2013)106.50710
1337Textos Notariais -OxfordTOXXavier (n.d.)4.2901
?Corte EnperialCEXavier (n.d.)113.6787
638.56660
15 Orto do EsposoOEXavier (n.d.)140.2752
1380-1425Flos SanctorumFSSobral (2015)3.3681
1414-1427Regra de São BentoRSBFigueiredo da Costa (2007)24.1370
1440/1450Crónica de D. FernandoCDFGalves et al. (2017)28.5533
1496História de VespasianoHVSobral (2015)18.9120
1488SacramentalSacrXavier (n.d.)153.0383
1401-1416Vida do Cativo Monge ConfessoVCMCSobral (2015)2.7520
1489Tratado de ConfissonTCJ. B. Machado (2003)27.5981
1463Crónica do Conde D. Pedro de MenesesCDPMBrocardo (1997)132.6995
1483Documentos Portugueses do Noroeste e da Região de Lisboa (CHEL)DNMartins (2001)1.0161
Castelo PerigosoCPXavier (n.d.)unknown3
1445-1517Crónica de D. Afonso HenriquesCDAHGalves et al. (2017)52.4750
531.33219
161504CatecismoCatXavier (n.d.)53.1720
Cartas de D. João IIICDJGalves et al. (2017)63.9470
1510PeregrinaçãoPerGalves et al. (2017)49.5620
A Vida de Bartolameu dos MártiresVFBMGalves et al. (2017)52.3210
1540Gramática da Língua PortuguesaGLPGalves et al. (2017)33.2400
Teatro séc. XVI (Comédia dos Estrangeiros; Auto do Filodemo; Auto da Cananea; Tragicomédia da Serra da Estrela))TeatroCamões (2010)37.9520
1513-1525Memorial Infanta Santa JoanaMISJSobral (2015)48.2877
Cancioneiro de Garcia de Resende (vol I)CGGRDias (1990)66.1729
Cartas Post ScriptumPS-16Marquilhas (2014)116.8721
521.52517
Livro de José de ArimateiaJARCastro (1984)140.13083
Table 2. Distribution of examples by century and polarity.
Table 2. Distribution of examples by century and polarity.
13th Cent.14th Cent.15th Cent.16th Cent.
negaffmodnegaffmodnegaffmodnegaffmod
No. occ.19815124317016121511
%88.06.75.371.728.30.084.25.31188.25.95.9
TOTAL225601917
Rel. frequency0.05%0.009%0.0035%0.0032%
Table 3. Frequency of future, present and past tenses according to polarity.
Table 3. Frequency of future, present and past tenses according to polarity.
Affirmative
(34 occs.)
Modal
(15 occs.)
Negative
(272 occs.)
no. occ%no. occ%no. occ%
Future (future + conditional)617.6%320%15657%
Present2161.8%533.33%4517%
Past (preterit + imperfect + pluperfect)411.8%533.33%6524%
Others (infinitive + gerund)38.8%213.33%62%
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Pinto, C. On the Evolution of Old Portuguese Indefinite jamais ‘Never’—Syntactic Analyzability and Polarity. Languages 2026, 11, 2. https://doi.org/10.3390/languages11010002

AMA Style

Pinto C. On the Evolution of Old Portuguese Indefinite jamais ‘Never’—Syntactic Analyzability and Polarity. Languages. 2026; 11(1):2. https://doi.org/10.3390/languages11010002

Chicago/Turabian Style

Pinto, Clara. 2026. "On the Evolution of Old Portuguese Indefinite jamais ‘Never’—Syntactic Analyzability and Polarity" Languages 11, no. 1: 2. https://doi.org/10.3390/languages11010002

APA Style

Pinto, C. (2026). On the Evolution of Old Portuguese Indefinite jamais ‘Never’—Syntactic Analyzability and Polarity. Languages, 11(1), 2. https://doi.org/10.3390/languages11010002

Article Metrics

Back to TopTop