3.1 A Historical Overview of the Ecology of Macanese

What is now known as the Macao Special Administrative Region of the People’s Republic of China is famous as the last European settlement in Asia, having been returned to Chinese sovereignty only in 1999. Portuguese traders were active in Macao probably already in 1553, although it appears that they became a permanent presence only a few years later, between 1556 and 1557 (Pinharanda Nunes, 2014). By the time of the settlement of Macao, Goa and Malacca had already been occupied for almost fifty years, and the Portuguese had already started developing a trade network in Southern China (Ansaldo, 2009). Tomás (2009: 50) pointed out that the Portuguese presence in Asia ‘was based on a network system rather than on the control of a territory for the production of goods’, which led to ‘cross-pollination’ of cultural and linguistic elements among the settlements (see also Ansaldo, 2009: 75). The convergence of features among apc s is easily explained in this context.

According to Pinharanda Nunes’s analysis, the ‘vast majority’ of speakers of non-standard varieties of Portuguese, creoles, and other languages did not have close contacts with the minority speaking European Portuguese, and hence received little input in the metropolitan Portuguese of the time. In this scenario, it is likely that Kristang had a major role in the formation of Macanese, and perhaps, as hinted at earlier (Section 1), the latter might even be seen as the continuation of the former (Pinharanda Nunes, 2014: 30).

What about the languages spoken by the ethnic Chinese population? The Sinitic languages native to the area surrounding Macao belong to the Yue/Cantonese group (especially, the Zhongshan dialect). However, Hokkien-speaking fishermen and traders were active in Macao too (Baxter, 2009). Actually, in the context of the so-called “China trade”, i.e., the trade between China and (some) Western countries between the 16th and the early 20th century, a form of pidginized Portuguese developed as a lingua franca: the main ports involved in the early days of the China trade were Macao and Canton, and the first speakers of this pidgin probably were Chinese traders who developed this variety in contact with speakers of ‘Portuguese’ (used here as an umbrella term for all the non-standard and creolized/pidginized varieties mentioned above) and/or Macanese (Baxter, 2009; Li and Matthews, 2016). We will return to this point below.

However, there is some debate as to the role of Sinitic languages in the formation of the Macanese creole. Differently from Malacca, at least until the mid-19th century the contacts between the Portuguese and the Chinese population were regulated: only few Chinese traders could enter the inner city, which was protected by walls, and they had to leave at nightfall. While a significant number of ethnic Chinese had converted to Christianity by the mid-17th century, their integration into the community was slow, (again) before the 19th century (Pinharanda Nunes, 2012b). Also, while, as said above, mixed Portuguese-Asian families were common in Macao, apparently this did not include Chinese wives at the beginning, and the Chinese presence in Macanese households was mostly limited to the muitsai (maidservants; Pinharanda Nunes, 2012b: 316).

In this scenario, Sinitic varieties should have had a very limited role in the formational period of Macanese. There is however some consensus on the point that Cantonese, the dominant Yue dialect in Guangzhou, Hong Kong and in present-day Macao, had a role in the later development of the Macanese creole, after the relaxation of the rules on the separation between the Portuguese and the Chinese population, and with the diffusion of intermarriage with ethnic Chinese. This is visible, for instance, in the use of the copula sã (< Port. são), under the likely influence of the Cantonese copular verb 係 haih (Baxter, 2009: 286–287), inter alia.

Also, there are reasons to hypothesize that the segregation between foreigners and ethnic Chinese in early Macao was not as rigid as is usually assumed, and some (Hokkien-speaking) Chinese traders were allowed to live in the settlement from very early on (Ansaldo and Matthews, 2004: 2). Moreover, while marriages between the Portuguese and the Chinese were not common, a distinguishing feature of Macao was the presence of many ethnic Chinese concubines (and illegitimate offspring) in Macanese households (Ansaldo, 2009: 77).⁸ As Ansaldo and Matthews (2004: 4) point out, “[a]s mothers and nannies, these Chinese women must have played a significant role in the development of Macanese from an early stage”. Thus, it is probably advisable to keep an open mind as to possible Sinitic influence throughout the entire history of Macanese (see Arcodia, 2017).

In sum, there is little doubt that Yue dialects, particularly Cantonese, have provided an ongoing influence on the later development of Macanese. As to the early phase of development of the Creole, the Malayo-Portuguese blueprint and restructured Malay varieties were obviously dominant; however, there is reason to believe that Hokkien may also have played a role.

In this connection, it is worth citing the observations made at the end of the 19th century by Pereira (1899–1901: 55, my translation; see also Cabreros, 2003), according to whom there were three distinct varieties (sociolects?) of Macanese.

• a. macaista cerrado / puro, ‘pure Macanese’, “spoken by the lower classes”

• b. macaista fallado pelos chins, ‘Macanese as spoken by the Chinese’

• c. macaista “modified to approximate standard (Metropolitan) Portuguese”

Taken at face value, Pereira’s remarks illustrate two more important aspects of the linguistic ecology of Macanese at the end of the 19th century. Firstly, what we call ‘Macanese’ must have had more basilectal and more acrolectal varieties. Secondly, some form of Macanese was spoken also by the ethnic Chinese, and thus the creole might have been subject to ‘double’ language contact – in the speech of the bilinguals of Eurasian ancestry, and in the speech of the bilinguals whose dominant language was a Sinitic variety.

However, an alternative interpretation could be that Pereira’s macaista fallado pelos chins is rather a pidginized variety, perhaps related to the earlier Portuguese pidgin (used in the China Trade) which we mentioned above. It could have been a rudimentary second language with much variation, ongoing influence from Cantonese, and with a set of crystallized, frequent structures.

The issue is far from trivial: given that, as we shall see below (Section 3.2), we have (very limited) data available for a form of pidgin Portuguese, we might want to know whether this is a stage in the development of Macanese, or just a separate, parallel variety. While we cannot elaborate on this here due to space constraints, there are both important differences and similarities between the pidgin and Macanese; given the early model provided by Kristang (as well as, arguably, other apc s), we see as unlikely that the pidgin is a source language for Macanese. We tend to agree with Li and Matthews (2016: 143), who suggest that the three varieties mentioned by Pereira may have represented a continuum “from Portuguese via Macanese spoken natively to pidgin Portuguese spoken by Chinese traders”. Be it as may, the available evidence points toward the existence of a pidginized form of Portuguese/Macanese, still spoken in 19th century Macao; this variety must have thus been yet another player in the linguistic ecology of Macao.

The last phase in the history of Macanese begins in the mid-19th century, as mentioned in the introduction, when schooling in European Portuguese for the Macanese population became more common, and Macanese went through a process of decreolization (i.e., convergence towards European Portuguese). In point of fact, Arana-Ward (1978) refers to the variety spoken before World War ii as the ‘Old Dialect’ or ‘Old Macanese’, as opposed to the more acrolectal variety spoken in her time. This process of decreolization was halted by the end of intergenerational transmission of the creole starting from the 1920s-1930s, when Portuguese (in Macao) and English (in Hong Kong and elsewhere in the Macanese diaspora) came to be seen as more valuable languages by Macanese families (Pinharanda Nunes, 2014: 38). This is mainly because speaking Portuguese was important to obtain good employment in Macao.

To conclude this overview, a summary of the evolution of the linguistic ecology in the history of the Macanese is presented in Table 1 (adapted from Arcodia, 2017: 166).

View Table

3.2 The Corpus Used in the Present Study

While Macanese has never enjoyed official recognition and has never been taught in schools (except for the very recent academic efforts mentioned above, Section 1), we do have at our disposal a significant amount of written documentation from different historical periods.⁹ Moreover, apart from written records, a recent (2007) corpus of spoken Macanese has been collected by Mário Pinharanda Nunes of the Universidade de Macau.

The oldest extant data of a Portuguese variety spoken in Macao, most likely the pidginized variety mentioned above (Section 3.1; see Li and Matthews, 2016), come from two sources: the 澳門紀略 Àomén Jìlüè ‘Monograph of Macao’ (1751), and the 澳門番語雜字全本 Àomén Fānyǔ Zázì Quánběn ‘Compendium of Assorted Phrases in Macau Pidgin’ (printed circa 1870–1890, but most likely older; Li and Matthews, 2016). The Monograph contains geographical, historical and administrative information on Macao, but also a small Chinese-Portuguese glossary (with Chinese characters suggesting the pronunciation of the Portuguese word, following Cantonese phonology); the later Compendium is but an expanded and revised version of the glossary in the Monograph (Li and Matthews, 2016: 149). The Compendium is especially valuable as a source of data because it provides more (short) phrases than the Monograph, and not only individual lexical items.¹⁰ While, as said earlier (Section 3.1), we do not believe that the Compendium language is a direct antecedent of Macanese, we shall anyway take the Compendium data into consideration, as it was arguably part of the linguistic landscape of Macao.

As to the Macanese creole senso strictu, the earliest body of data in our sample comes from the Ta-Ssi-Yang-Kuo (Chin. 大西洋國 Dàxīyáng Guó, ‘The Atlantic Country’, i.e., Portugal). The tsyk was a Portuguese language journal concerning the Portuguese Far East, and the issues from 1899 to 1901 are available as scanned copies on the web.¹¹ The oldest text published in the tsyk dates back to 1824, and while the journal is written, as expected, in Standard Portuguese, it also contains texts written in (some form of) Macanese, presented as “aids for the study of the creole dialects of the Far East” (subsidios para o estudo dos dialectos crioulos do Extremo Oriente). The series contains 8 subsidios on Macanese, which include three poems, five letters from the readers, two descomposturas (‘quarrels’), two collections of riddles and two songs.

However, the larger set of data comes from the œuvre of José dos Santos Ferreira, also known as “Adé” (1919–1993), by far the most prolific writer in the Macanese language. He belonged to the last generation of fluent speakers of Macanese, and his work represents a conscious attempt to record the language before it disappeared (Ansaldo and Matthews, 2004). Within the vast production of this author, we collected a sample of 8 short stories, 3 plays and eleven poems. While this is an extremely valuable set of data, it is important to stress the fact that it might not be representative of the actual language spoken at the time, especially given the broad range of variation that was likely to be present in Macanese society, since the texts all come from a single individual.

Lastly, as mentioned above, there is one corpus of contemporary spoken Macanese, collected by Jean Michel Charpentier (in 1984) and Mário Pinharanda Nunes (in 1999 and 2007). For the present study, we could access (courtesy of the author) the transcriptions of the 2007 recordings, which were conducted in the Macanese community in Vancouver and San Francisco. This is the only significant body of data of the ‘decreolized’, acrolectal spoken variety of Macanese, and is thus of immense value for research on the Macao creole. The transcriptions we used involve seven women and men aged between 78 and 85 at the time of the interview, all born in the 1920s in Hong Kong (except for one informant); also, they had all been in their new countries for at least forty years then.

Besides raw language data, there are also a few grammatical sketches of Macanese available, and they come from different periods in the creole’s history. The oldest references are a 1-page sketch by Coelho (1881), and a brief sketch by Leite de Vasconcellos (1892; 1901). More extensive descriptions are offered by José dos Santos Ferreira himself (1978), and Arana-Ward (1978); note, however, that the variety described by Adé is far more conservative than the one described in Arana-Ward, which is representative of the post-wwii dialect (as spoken in Hong Kong), and reflects the variety of Macanese spoken in Hong Kong at the time. More recent descriptions of the interaction of tam and negation in decreolized Macanese may be found in Pinharanda Nunes and Baxter (2004), Pinharanda Nunes (2011), and Lebel (2018); we will turn to this in the next section.

4.1 The Compendium

As mentioned above (Section 3.2), the relation between the language represented in the Compendium and Macanese proper is open to debate; hence, the data which comes from this source has to be taken with a pinch of salt. In the Compendium we find two preverbal markers, já and logo (only one occurrence of the latter; Li and Matthews, 2016: 169), and two negators, namely nung (transcribed in Cantonese as 噥 nùhng)¹⁷ and nunca (噥加 nùhnggā); they are seen as deriving from Port. não and nunca, respectively (Li and Matthews, 2016: 177). Já and logo are characterized as in Macanese, i.e., as a perfective marker and as a future/irrealis marker; nung is interpreted by Li and Matthews (2016: 177) as a negator for “imperatives” and “modal verbs”, while nunca is analyzed as a general negator “for other cases”.

However, in our perusal of the Compendium material, we found that nung is used also for existential negation (13) and for the negation of possession; and, in one instance, to negate a past/perfective event (14).

View Table

View Table

Nung is used to negate the verbs sabe ‘know’, pôde ‘can’ (referred to by Li and Matthews, 2016 as ‘modal’ verbs; see above) and querê ‘want’ which, as said above (Section 4), are the same verbs which are normally negated by non in Macanese; in one sentence, it is used to negate the verb conhecê ‘to know’. It is also used for locative negation.

Nunca, on the other hand, is used almost only for ascriptive/attributive negation, with adjectival predicates (15) or with the copula (16).

View Table

View Table

While any claims based on such a small amount of data have limited reliability, we may tentatively suggest that, in the Compendium, nung appears to be a general standard negator, with no apparent tam features, and also an existential, possessive and locative negator (just as in later texts in Macanese). Nunca, on the other hand, is used almost exclusively for ascriptive negation in the Compendium: adjectives and the copula are consistently negated by nunca already at this stage.

4.2 The tsyk

In Section 4, we hinted at the fact that in the tsyk we find a brief analysis of the Macanese system of negation: what is said is that nunca is used as the standard negator in past tense contexts, while in the present some verbs require non, while others use nunca. According to the description in the tsyk, the verbs which require non include the above-mentioned querê ‘want’ and sabe ‘know’. Nádi is presented as the negator for the future; interestingly, it is claimed that, when used to negate past events, nunca may combine with the perfective aspect marker já (e.g., nunca já fazé ‘did not do’), something which is not normally possible at later stages of Macanese, as said above (Section 4).

If we look at the data, non is in fact the usual negator not only for querê ‘want’ and sabe ‘know’, but also for pôde and tem, the latter being used for existential negation and for the negation of predicative possession, just as in later texts. See the following example. In (17), we see that the choice between non (nun) and nunca does not appear to be conditioned by tam, but is rather lexically determined.

View Table

As for ascriptive negation, the association between nunca and this subtype of negation is rather strong, but not as in the dos Santos Ferreira corpus (see below, Section 4.3). While the copula sã is most often negated by nunca, we have also many examples in which it is negated by non, as in the following sentence:

View Table

On the other hand, the use of nunca as a negator for (bare) adjectival predicates is almost exceptionless.

In the tsyk, we can find the earliest examples of conditional sentences. As hinted at above (Section 4), nunca is used in the protasis, while nádi is used in the apodosis.

View Table

This is reminiscent of English conditional sentences, in which the more ‘realis-leaning’ (for lack of a better term) present tense is used in the protasis, while the more ‘irrealis-leaning’ will future is used in the apodosis (see de Haan, 2012). However, this does not seem to be the case if the copula sã, and hence ascriptive negation, is involved; see example (3), repeated here for the sake of convenience.

View Table

In the tsyk corpus, the functions of nádi are pretty much the same as in later texts: nádi is the general negator for irrealis, and can be used with any verb except the copula sã and adjectives. Hence, while the use of nunca with the copula does not seem to be the only option, it appears that nádi cannot be used for ascriptive negation. Interestingly, nádi here is apparently used also to negate habituals.

View Table

Example (20) is a positive comment on the Governor of Macao of the time, José Rodrigues Coelho do Amaral. The fact that he was very attentive, and wouldn’t miss a thing, is framed as something stable, which happens regularly.

Thus, in the tsyk we see that, as to standard negation, the ‘division of labour’ among non, nunca and nádi is very close to what we suggested above (Section 4), with nádi covering the semantic space of irrealis, while non and nunca are the default (realis) negators; the choice between them mostly depends on the lexical item. As for ascriptive negation, the difference between non and nunca is perhaps not as sharp as in the later dos Santos Ferreira corpus (see below, Section 4.3): non is used also with the copula, as mentioned above, in quite a few cases. The association between nunca and (bare) adjectival predicates, on the other hand, appears to be very solid, just as seen above for the Compendium data. Besides, in the latter nádi appears to have developed the function of negator for habitual events, something which we find more often in later texts: the overlap of irrealis and habitual has parallels in other apc s, as we shall see below (Section 5.1).

4.3 The dos Santos Ferreira Corpus

In the dos Santos Ferreira corpus, the roles of the three negators are neatly distinguished. Non is used almost exclusively to negate pôde ‘can’, querê ‘want’, sabe ‘know’ and tem ‘have, there be’; it is thus not only the standard negator for the above-mentioned verbs, but it is also the existential (and locative) negator, and the negator of predicative possession. While it can sometimes combine with the copula sã, this is exceedingly rare (only four occurrences in our corpus; see also Lebel’s data mentioned above, Section 4).

Besides the stabilization of nunca as an ascriptive negator, one more significant difference between the dos Santos Ferreira data and the tsyk materials is that, here, nádi can be used also to negate existentials and predicative possession, as in the following examples.

View Table

View Table

In examples (21) and (22), the verb têm is negated by nádi, rather than non. Example (21) appears to be a conditional sentence, and hence nádi is the expected negator in the apodosis, even though it is an existential sentence. In example (22), on the other hand, nádi negates têm in a main clause. Lastly, in the dos Santos Ferreira corpus, we find the only example of ascriptive negation with nádi.

View Table

Note, also, that this apparent exception is found in the apodosis of a conditional sentence (compare example 3 above), rather than in a main clause.

In short, the system of negation in the dos Santos Ferreira corpus appears to be stable and consistent, with each negator having defined functions. Here, nádi appears to be the irrealis negator not only for standard negation, but also for existential and possessive negation; ascriptive negation is the only subtype which is consistently negated by nunca, with one exception (23). However, as pointed out above (Section 3.2), this corpus is based on the language of a single speaker, and hence it is expected to be more coherent, if compared to the tsyk data.

4.4 The Contemporary Spoken Corpus

In the spoken corpus we find the widest degree of variation in the use of negators. At this stage, the data appear to represent different points of a continuum between more conservative/basilectal and more innovative/acrolectal constructions, with variation even in the language of the same speaker.

In the spoken corpus, we see a shift towards superstrate (Portuguese) grammar in the domain of tam: besides the characteristic creole preverbal markers já, ta and logo, verbs appear also in inflected forms which are functionally analogous to the corresponding superstrate forms: mainly the first and third person singular of the present indicative, and the third person singular of the past perfective and past imperfective (Pinharanda Nunes, 2011; 2012b). This applies also to the copula, which does not appear only in the basilectal form sã, but also in inflected forms. Note that verb forms with superstate inflection may cooccur with preverbal tam markers (24); also, they may also have a functional mismatch with the corresponding superstrate forms (25).

View Table

View Table

Pinharanda Nunes (2012b: 302) suggests that the inconsistent use of superstrate inflection (mostly, present tense forms in past contexts, as example 25) indicates the fact that “speakers are not entirely conscious of their superstrate functions” (on the instability of the verb system of post-wwii Macanese, see also Arana-Ward, (1978): 96–101).

As for negation, Pinharanda Nunes (2011) highlights the fact that nunca is widely used with a broad range of verbs, both in present and in past tense contexts, and may even cooccur with the perfective marker já; nang (equivalent to non) is used in present, past and imperative contexts; nádi, again, is the future/irrealis negator. However, Pinharanda Nunes also points out that inflected past perfective and imperfective forms are rarely negated, and the basilectal forms seem to be preferred in negative clauses. Given the above-mentioned broad range of variation in the spoken corpus, we will not present a thorough analysis of the system of negation due to lack of space; we will thus focus on ascriptive negation, while discussing also examples for other types of negation.

The most salient difference between the written corpora (representing the ‘Old Dialect’; see above, Section 3.1) and the spoken corpus is that, in the latter, the copula is most often negated by non. However, the many occurrences of non with the copula are actually combinations with the inflected third person singular form é of the copula; the basilectal form sã is consistently negated by nunca. Interestingly, these two patterns coexist even in the same speaker, as shown by the following examples.

View Table

View Table

While in (26), just as in example (25), we have a present tense inflected form in a past tense context, and hence there is a functional mismatch between superstrate inflection and function, the negator nam (i.e., non) is selected; however, in the very same conversation, the speaker chooses nunca as the negator for the basilectal copula sã. Also, when the speaker in question needs to negate a bare adjectival predicate, she uses again nunca.

View Table

The use of non to negate the inflected third person singular copula, and nunca to negate the basilectal copula and bare adjectives is very consistent throughout the corpus. The only exception is represented by two instances of bare adjectives negated by non. While in Section 4 we pointed out that non might be either a lexically-conditioned allomorph of nunca or a separate negator deriving from Port. não, and that we favoured neither hypothesis, we also believe that, at this late stage, non is more likely to be the creole version of não. This is because, in our opinion, the fact that non, differently from nunca, is consistently associated with the acrolectal form of the copula, suggests that it is but a construction based on acrolectal material.

For standard negation, however, things are not as neat, and the use of nunca with superstrate inflected verb forms is attested.

View Table

As for nádi, it covers much the same functional space as in previous samples of the language, including many instances in which it is used to negate a habitual predicate. In addition to that, it appears that nádi may negate volition, as in the following exchange.

View Table

In example (30), the interviewer asks a question based on the verb querê ‘want’, to which the informant replies with nádi rather than non. Here, it appears that nádi conveys a semantic nuance of volition; curiously, this is one more aspect in which the semantics of nádi overlaps with English will (not).

Thus, while we did not provide an exhaustive analysis of negation in the decreolized variety of Macanese represented in the spoken corpus, we may suggest that it represent a transitional system, as expected. If we focus on ascriptive negation, we see that there appear to be two subsystems which coexist: a basilectal subsystem, based on nunca and the copula sã, and an acrolectal subsystem, based on non and the inflected copula é. Adjectives seem to conform more to the basilectal model, though exceptions are attested.

4.5 Summary

To sum up, the picture which emerges from the analysis of our corpus largely confirms Lebel’s (2018) analysis of standard negation in Macanese. The key differences lie in the specific behaviour of ascriptive negation, and in the apparently broader functional spectrum of nádi, including the negation of habituals and volition. Also, we are now able to highlight the diachronic evolution of negation, showing the differences between the stages we considered. In Table 3, we summarize the evolution of the markers of negation according to the typology introduced earlier (Section 2).

If we look at Table 3, we see that the full range of negators has been used for standard negation, at least since the xix century; while nádi is conspicuously absent from the Compendium data, given the limited extension of the sample and the unclear relation between the language represented therein and Macanese proper, the significance of this is debatable. Existential, locative and possessive negation are remarkably consistent in being negated by non (with the verb tem): starting at least from the xx century, nádi is also sometimes used with those three subtypes of negation, which thus seem to become part of the realis vs. irrealis distinction that permeates the system. In this respect, ascriptive negation stands out as being apparently ‘immune’ from the realis vs. irrealis distinction, given that nunca is virtually always used to negate the copula and adjectives, also in conditional sentences (see example 3). The main changes in the marking patterns of ascriptive negation is that until the xix century non is often, though not dominantly, used to negate the copula, while this does not happen in later texts; in the most recent spoken corpus, it appears that a basilectal (nunca sã) and an acrolectal (non é) marking pattern coexist, indicating increased influence from superstrate grammar (examples 26–27).

The interaction between negation, the lexicon, and tam categories is summarized in Table 4 below.

View Table

As argued earlier (Section 4), the opposition between realis and irrealis appears to be central to the system of negation in Macanese, at least starting from the tsyk materials. This is immediately apparent for standard negation, but later extends to existential/locative/possessive negation, as may be seen in the dos Santos Ferreira subcorpus. Irrealis also encompasses habituals, as is already evident in the tsyk, and subsequently, in the most recent spoken corpus, it also seems to convey volition. The realis/irrealis opposition is particularly evident in conditional sentences (see example 19), as nunca (or non, depending on the verb) appears in the protasis, and nádi in the apodosis. However, reality status does not come into play for ascriptive negation, as said above, since the copula and adjectives are almost never negated by nádi (but cf. example 23).

Thus, in short, there appear to be three macro-types of negation in Macanese:

• a. Standard negation, based on a realis vs. irrealis opposition, with the choice between two realis negation markers (non and nunca) depending, in essence, on the lexical item involved;

• b. ascriptive negation, with a single marker and no realis vs. irrealis distinction;

• c. existential, locative, and possessive negation, which are overlapping domains in Macanese; they are all based on the negator non and the verb tem, later becoming part of the realis vs. irrealis opposition.

As for macro-type c., we may remark that, as said earlier (Section 2), the marking of existential and possessive negation by the same construction is quite common cross-linguistically, and the use of the same construction also for locative negation is not rare either. It is actually the dominant pattern in apc s (see below, Section 5.1).

Lastly, we argue that the system of negation of Macanese is actually not symmetric, both on the constructional (i) and on the paradigmatic (ii) level.

• i. While, as said above, preverbal tam markers are optional, negators are (unsurprisingly) necessary to express negation, and markers of negation and tam markers mostly do not cooccur. Thus, tam values are carried by the negator itself (compare example 10, above) in negative clauses.

• ii. While there seems to be a near-perfect one-to-one correspondence between lôgo and nádi, in that they both mark irrealis status,²⁰ ta and já are primarily aspect markers, not mood markers: the choice between non/nunca and nádi, on the other hand, is based on reality status. The progressive vs. perfective distinction is thus lost in negation: as said above for Swahili, the Macanese affirmative and negative paradigms are based on partly different tam categories. Moreover, we may say that the marking of reality status is apparently obligatory in negative sentences (at least for standard and existential/possessive/locative negation), but not in affirmative sentences (i.e., ‘asymmetry in the marking of reality status’, in Miestamo’s terms).

We now turn to the discussion of some comparative data from other apc s and from adstrates and substrates of Macanese, in order to bring to light the possible influences on the Macanese system of negation.

5.1 Negation in apc s

When discussing possible influences on the development of Macanese, Kristang is the first variety one wants to look at, given the strong historical (and typological) connection between the Malaccan creole and Macanese. In fact, while there are many similarities in the domain of negation between the two creoles, there are also significant differences.

There are four main markers of negation in Kristang: nang, a negator with imperative semantics; ńgka / nungka, cognate to nunca, which is the general standard negator for realis predicates; nadi, the future/irrealis negator, and nenáng, a ‘not-yet’ negator with perfect semantics (see above, fn. 7; Baxter, 1988). These negators are compatible with all types of predicates, including verbs, adjectives, and noun phrases; this entails that, apparently, ascriptive negation is not different from standard negation (Alan N. Baxter, p.c. 2016). See the following examples, in which bare adjectives are negated by ńgka, nenáng and nadi.

View Table

View Table

View Table

Note that nadi, when combined with stative verbs or adjectives, has an inchoative meaning (Baxter, 1988: 141; see the translation of example 33). Apart from this, nádi covers pretty much the same semantic space as in Macanese: it can be used to negate the apodosis of conditional sentences (see e.g., example 78 in Baxter, 1988: 141) and also habituals, including past habituals. However, while we do have examples of nadi as a negator for adjectives (33), we could find no example of nadi with a nominal predicate in Baxter (1988). We then searched the Kristang corpus stored in the Endangered Languages Archive at soas (Pillai, 2011), and found no example there either. In Baxter’s grammar of Kristang (1988: 67, 141), it is stated that nadi is the irrealis negator which occurs before verbal and adjectival predication; he does not mention nominal predication here.

Also, note that the existential verb teng (< Port. ter ‘have, there be’) may be used as a copula, although this seems to be uncommon (Baxter, 1988: 181–185). The negative form of teng is nté, or the less frequent long form nunténg: however, we did not find examples of nté / nunténg used for ascriptive negation in Baxter (1988). In Pillai’s (2011) corpus, we found only one instance nté / nunténg as ascriptive negator (isi nang teng retu ‘this is not right’).²¹ On the other hand, examples of ńgka as the negator for nominal and adjectival predication are very common. We thus suggest that the use of the negative copula for ascriptive negation, while possible, is not the default, unmarked choice.

Thus, to sum up, it appears that in Kristang there is also some sort of ‘split’ between standard and ascriptive negation: irrealis nadi is used (not often) with adjectival predicates, but possibly not with nominal predicates. However, more data is needed (especially, on conditional sentences) to properly assess whether the ban on nadi with nominal predication is the norm in the language.

Two more apc s which are regarded as being closely related to Macanese (and Kristang) are the now-extinct Bidau Portuguese Creole of East Timor (Baxter, 1990; Baxter and Cardoso, 2017), and the Batavia and Tugu Creole, formerly spoken in the Jakarta region. The available data for Bidau is extremely limited, and only nunca and a negator deriving from Port. não have been identified. As to Batavia and Tugu Creole Portuguese, Maurer (2011) describes a system of negation which closely matches that of Kristang. Nungku is the general standard negator for verbal predicates; nada is the irrealis negator; nang is the negative imperative. However, there is a major difference between Batavia/Tugu and Kristang: in the former, the copula teng is normally used (less often in the Tugu variety; Maurer, 2011) for nominal and adjectival predication. The negative form nonteng is used for ascriptive negation in Batavia/Tugu.

View Table

It would be interesting to know whether nada could be used with adjectives, just as in Kristang. This would entail that ascriptive negation in Batavia and Tugu, differently from Macanese, also has a realis vs. irrealis distinction. Unfortunately, the existing corpora both of the Batavia and of the Tugu variety of this creole provide very limited evidence. In the Batavia corpus, there is but one example of nada, with the verb larga ‘leave’. In the Tugu corpus, collected in Schuchardt (1890), we find this example.

View Table

In the Malay version of this text, the adjective basa(h) ‘wet’ is used. Schuchardt (1890: 202) seems to interpret it as an adjective (translated as German nass ‘wet’); moela is seen by Schuchardt either as a verb or as an adjective, depending on the context (see Schuchardt, 1890: 51, Fn. 64). However, we believe that his interpretation is debatable: if we look at the shape of the word, the Portuguese verb molhar ‘wet’ looks like a better match for moela than the adjective molhado ‘wet’. In point of fact, elsewhere in the Schuchardt Tugu corpus we find the forms moelain and moelàdoe for the Malay adjective basah ‘wet’. Incidentally, the form moeladoe for the adjective ‘wet’ is found also in the Batavia corpus. However, Standard Portuguese would have the reflexive form molhar-se ‘to get wet’. Another example provided by Schuchardt (1890: 202) is nada danu, translated as “nicht verdorben”, lit. ‘not damaged’. On the one hand, the Portuguese noun dano ‘damage’ is surely a better match for danu than the adjective danificado. However, danu is found in the Tugu corpus in the inchoative construction with the verb fika ‘become’: fika danu ‘to become damaged’. Given that fika normally combines with adjectives, this use may be seen as an instance of noun>adjective conversion (Schuchardt, 1890: 204–205).²² Be it as may, the use of a noun with nada, instead of nunteng, proves that the realis vs. irrealis distinction applies also to ascriptive negation (compare also nada perdisang ‘no loss’).

To conclude, there appears to be limited evidence that the irrealis marker nada was used, at least in the Tugu variety, also for ascriptive negation, just as in Kristang. However, more data would be needed for a proper assessment.

As for the acp s of the Indian Subcontinent, Diu Creole Portuguese (Cardoso, 2009) has a very different system of negation, based on different parameters, which we shall not discuss here due to space constraints. In Korlai, we have a general negator nu which combines with different auxiliaries to convey a relatively large number of tam values (see Clements, 2018: 219); among them, we find a future negator nu pa(d) (< Port. não pode ‘cannot’), but its functions do not overlap entirely with those of Mac. nádi, and there appears to be no special pattern for ascriptive negation.

In the above-mentioned Malabar Creole Portuguese, we find the same three main negators as Macanese, i.e., nu / no, nuka, and na(d). Nu / no is the general negator with no tam semantics; nuka is the negator for past events; na(d) is the irrealis negator (Krajinović, 2018). Nu / no is the negator also for the present tense copula tæ (also < Port. ter) and for the past tense copula tinha, which are used in adjectival and nominal predication; nuka is thus not used in ascriptive negation. The copula is generally omitted in the present (always for nouns, often for adjectives), and sometimes also in the past tense: thus, nu / no can be the negator also without the copula tæ / tinha in those cases.

However, just as in Kristang and many other apc s, the Malabar copula tæ (/ tinha) covers also the semantic space of existential, possessive and locative (Krajinović, 2019). The irrealis negator na(d) is used also to negate tæ, but only as the existential/possessive negator. It is unclear whether na(d) is ever used for ascriptive negation: the lack of occurrences might be an artifact of the limited data available. In fact, we may remark that the irrealis marker lɔ, with the same source as Macanese lôgo / lô, is used for ascriptive predication in Malabar.

View Table

Given that na(d) is described by Krajinović as having the same functions as lɔ, we may infer that it can be used also for ascriptive negation. In the absence of more data, we must remain agnostic as to whether this use is actually attested. Thus, while there seems to be a marking pattern for ascriptive negation distinct from standard negation in Malabar, this pattern largely overlaps with that of existential, possessive and locative negation. Besides, the categories involved (present, past, irrealis) seem to be the same for standard and ‘special’ negation constructions. A difference between ascriptive negation and the other types of special negation lies in the optionality of the copula tæ, which seems to apply only to the former. Another difference, more relevant for our comparison with Macanese, is that it is unclear whether irrealis na(d) is used also for ascriptive negation: this, however, cannot be confirmed with the available data.

Lastly, in Dalgado’s (1900) grammar of Sri Lanka Portuguese,²³ he describes a system of negation which is a pretty good match for Malabar: he mentions não as the negator for the present, nunca for the past, nada for the future, and nadía for the conditional (Dalgado, 1900: 41).²⁴ While nada derives from não há-de just as Mac. nádi, nadía is said to be the univerbation of não and dia (< devia / deveria ‘should’). This ‘split’ between nada and nadía mirrors the categories for the affirmative: apart from the usual ló (< logo), described as a marker “for the future of the indicative and subjunctive”, we find also lodía for the conditional mood, which is the product of the contraction of ló(go) and dia. Furthermore, according to Dalgado’s description, the verb tem (again, < ter) is negated by nun just as in Macanese: in the Sri Lankan creole, however, besides existential negation, tem covers also the functional space of ascriptive negation, as in the following example (compare example 34 above).

View Table

This is pretty much the same range of functions we see in the Batavia and Tugu Creole and in Malabar Creole Portuguese, but not in Macanese, which makes use of the copula sã. Moreover, according to Dalgado (1900), tem can be negated by nunca in the past (but nuntinha is the form of the negative imperfective past, similarly to Malabar), and by nada in the future, thus behaving like any other verb.

To sum up, while in most of the apc s considered here ascriptive predication does not seem to have the features highlighted for Macanese, there is some evidence of a split between adjectival and nominal predication in Kristang, arguably the closest relative of Macanese. As to Malabar, while we did not find examples of na(d) in ascriptive negation, more data is needed to assess the compatibility of irrealis with this subtype of negation.

5.2 Negation in Malay

As mentioned earlier, there is little doubt that Malay varieties, especially restructured colloquial varieties, had a significant influence on Kristang, and very likely also on Macanese, particularly in the formational period of the creole (see also Smith, 2012). In Modern (Standard) Malay, the main negators are tidak / tak, the (clausal) standard negator, which also negates adjectives (38), and bukan, the negator for np s, often without the copula (39).

View Table

View Table

Additionally, there are the negative imperative jangan, and the ‘not-yet’ negator belum (see above, fn. 7); bukan may be also used as an emphatic negator for adjectives, implying contrast (on the relation between ascriptive negation and contrastive focus, see Mettouchi, 2009).

However, the Malay variety (or, better, varieties) in use in Macao, especially in the early days of the settlement, may not match the above-described system. We do have access to grammars and dictionaries of Malay written between the 17th and the 19th century: among the early sources, we may mention here George Werndly’s Maleysche Spraakkunst (published 1736; see Mahdi, 2018), and an English-Malay (and Malay-English) dictionary with a grammatical description compiled by Thomas Bowrey (1701). These mention the negator bukan, and another negator which is given in various forms: tijâda, tîda (Roman, 1674), tēda (Bowrey, 1701); while some differences may be attributed to differences in the spelling conventions by these authors, there is no doubt that there was allomorphy for this negator, just as tidak / tak in Modern Malay.

Two later reference works are Marsden’s (1812) and Crawfurd’s (1852) grammars of Malay. Marsden mentions only the forms būkan, glossed as ‘it is not’, and tīdak, glossed as ‘no’ (1812: 88): however, in the examples he provides, only tīdak is ever used. In Crawfurd’s grammar, he mentions bukan, glossed as ‘not at all’, and a wide range of (apparent) alternative forms for tiyada, glossed as ‘no’. These include tiyadak, tidak, tida, tada, ta, and trada: he points out that trada is a “corrupt pronunciation” for tiyada, and that trada bayik ‘not good’ is “in vulgar Malay, the most frequent mode of expressing ‘bad’” (Crawfurd, 1852: clxviii).

Unfortunately, the sources cited here do not provide enough data to allow us to learn the details on the functional range of each negator. Nevertheless, from the examples we could gather (as well as from the glosses), we believe it is safe to say that tijâda / tîda / tēda was the standard negator, used also for adjectives, while bukan had a more limited distribution, and likely had an emphatic nuance, similarly to Modern Malay.

However, it is important to stress the point that these reference works may not have had as their main object of inquiry the colloquial varieties of the language used in everyday interactions, also known as Low Malay. Werndly (1736) was essentially based on the High Malay literary language (see Mahdi, 2018); other grammars, while discussing also vernacular conventions, made large use of literary examples. As pointed out by Ansaldo (2009), the linguistic ecology of Malay-speaking territories (at least until the early 20th century) was very complex, with a range of varieties beyond the simple ‘high vs. low’ dichotomy generally described in the literature. Besides the literary language and regional spoken varieties, there were trade languages and other contact varieties: these restructured varieties were widely used as vehicular languages in Monsoon Asia. As hinted at earlier (Section 3.1), there is little doubt that it is these colloquial and restructured ‘Malay’ varieties, rather than literary, formal registers of Malay, which could have played a significant role in the ecology of Macanese.

In this respect, a very interesting source of data is Dennys (1878)’s handbook of colloquial Malay. While this reflects the Malay vernacular as spoken in Singapore, and at a time when the Macanese creole was already past its formative period, we believe that Dennys’s work is still useful as a point of comparison. This is because Dennys provides numerous examples of actual language use, and he stresses the point that the language represented is a compromise between high (“pure”) Malay and the “form of it current amongst the Chinese, Javan, Boyan and Kling residents of Singapore”, i.e., the restructured varieties used as linguae francae (Dennys, 1878: no page number). The system of negation described in Dennys’ handbook, in essence, is largely compatible with that of Modern Malay described above. Tidak and tada are used as standard negators, both for verbs and for adjectives, while bukan (also bukan-nya) negates np s, and is also used as an emphatic negator. Just as in Marsden (1812), bukan is glossed as ‘is not; it is not’, while tidak is glossed as ‘no; not’. See the following example.

View Table

In example (40), bukan is used as an emphatic negator, while tidak negates the adjective baik. However, note that bukan is sometimes used to negate adjectives in interrogative questions: it is suffixed with the interrogative particle -kah in these cases. This usage is analyzed in an early 20th century grammar of Malay (Maxwell, 1914), as “emphatic denial”, “to signify is it not? is it not so?”. Another grammar from that period, namely Winstedt (1913), provides the following minimal pair.

View Table

Thus, it appears that the ‘division of labour’ between tidak and bukan has been in place for the past two centuries, with a seemingly high degree of continuity.

Moreover, it is worth considering here also a (once widespread) vehicular variety of Malay, i.e., (Singapore) Bazaar Malay.²⁵ In Bazaar Malay, bukan is used to negate adjectives without implying contrast.

View Table

In point of fact, the occurrence of the so-called ‘nominal’ negator bukan with adjectives is one of the criteria used by Khin Khin Aye (2005) to argue that adjectives are a word class separate from verbs in Bazaar Malay. On the other hand, adjectives may be negated by the verbal (standard) negator tak as well in the language. Moreover, negators by themselves do not carry tam values (with the exception of the not-yet maker belum).

In his presentation of the Kristang system of negation, Baxter (1988: 138–144) points out that nenáng ‘not yet’ is the only marker which has a clear functional parallel in Malay (i.e., the above-mentioned belum); ńgka and nádi, on the other hand, are seen as independent from the Malay model. Given that no cognate to nenáng is used in Macanese, Malay influence on the Macao creole should not have been inherited from Kristang; if Malay varieties have influenced Macanese, this is likely to have happened in the early days of the settlement, independently from the Kristang blueprint. In this connection, a restructured variety like Bazaar Malay is a plausible contributor, since it appears to have a specific ascriptive negator, bukan. However, there are two important differences between Bazaar Malay and Macanese: firstly, bukan does not overlap with the standard negator, differently from nunca; secondly, adjectives are not consistently negated with bukan, and the negators themselves have no tam implication. We will return to this below (Section 5.4).

5.3 Negation in Sinitic

As for Sinitic languages, we already mentioned that the influence of Cantonese in the later development of Macanese has generally been acknowledged in the recent literature (see above, Section 3.1). For the system of negation, Lebel (2018: 182–183) suggests that the imperative negator ne-bom / nunca bom ‘not good’ (see above, Fn. 14) could be the relexification of Cantonese 唔好 m̀h-hóu ‘don’t’, also lit. ‘not good’. Lebel (2018: 176) also points out that the combination of nunca and the copula sã in Macanese is very often found in contrastive focus constructions (‘it is not the case that…’), and that this specific usage of nunca may be associated to the parallel Cantonese construction 唔係 m-haih ‘neg-cop’.

Apart from the above, another potential area of overlap is that Cantonese does appear to have a ‘split’ between standard negation and ascriptive negation, somehow reminiscent of the Macanese system. Both adjectives and the copula are negated by 唔 m̀h; moreover, as said above, in Cantonese the negative copula 唔係 mhaih is used for contrastive focus, and, also, to negate modified adjectives.

View Table

Cantonese verbs, on the other hand, are also negated by 唔 m̀h for the present and future, while 冇 móuh is used to negate an event which has not occurred.²⁶

View Table

Thus, just as in Macanese, in Cantonese we seem to have tam distinctions for standard (verbal) negation which do not apply to ascriptive negation; also, the marker of ascriptive negation, namely 唔 m̀h (/ 唔係 mhaih), overlaps with one of the markers of standard negation, just as nunca. However, the tam categories at issue are in fact different from the realis/irrealis distinction which is pervasive in Macanese negation.

Cantonese does have a marker of irrealis though, namely 會 wúih, which is also negated by 唔 m̀h. Interestingly, the functions of 會 wúih seem to overlap to some extent with Macanese lôgo: it is used to indicate future events (generally, predictions) in the apodosis of conditional sentences, and to mark an event as habitual (Matthews and Yip, 2011; Chappell and Peyraube, 2016b). This seems to be true also for its negated form; differently from nádi, though, 唔會 m̀h-wúih is used both for standard and for ascriptive negation. See the following Cantonese example.²⁷

View Table

As shown in example (45), the use of 唔會 m̀h-wúih as an irrealis negator applies to the domain of ascriptive negation too: there is hence no difference between standard and ascriptive negation in this respect.

Besides Cantonese, Hokkien could also have influenced the development of Macanese, especially in its formative period, as argued earlier. Hokkien has a very complex system of markers of negation which convey different tam nuances (Chappell and Peyraube, 2016a): due to space constraints, here we shall focus only on the three most common negators, namely 毋 m̄, 無 bô and 袂 bē. For the sake of convenience, our discussion is based on the Taiwanese variety of Hokkien (also known as ‘Taiwanese Southern Min’), as it is arguably the best described variety. In Taiwanese Hokkien, verbs, the copula, and some adjectives may be negated by 毋 m̄. However, 毋 m̄ actually functions as the standard negator only for the copula and for a limited set of verbs and adjectives; for most lexical items, negation with 毋 m̄ implies volitionality (Lin, 2015: 299–300), as in the example below.

View Table

For ‘plain’ negation of (most) adjectives, either 無 bô or 袂 bē are used. According to Lin (2015), the choice depends on whether the quality described by the adjective is perceived as ‘desirable’ (47) or ‘undesirable’ (48).

View Table

View Table

Both 無 bô and 袂 bē, however, have many other functions, besides negating adjectives. Taiwanese Hokkien 無 bô, similarly to Cantonese 冇 móuh (see above, example 44), is an existential and possessive negator, and also negates the occurrence(/completion) of an event. Besides, 無 bô in Hokkien is also used to negate habitual actions (with the progressive aspect marker 咧 teh), and to negate the protasis in conditional sentences.

View Table

袂 bē, on the other hand, has a clear modal function, as it is used to negate possibility, ability and permission (Yang 2012): it is in fact the negator of the irrealis marker 會 ē (see above, example 48; see Chappell and Peyraube, 2016a). In point of fact, Lien (2015: 174) believes that the difference between 無 bô and 袂 bē as negators for adjectival predicates lies in the speaker’s perspective on the evaluation: while 無 bô indicates certainty, 袂 bē indicates a conjecture. This is consistent with the characterization of 袂 bē as an irrealis negator.

Thus, in short, the Hokkien system of negation resembles Macanese in having a specific negator for irrealis, namely 袂 bē, and a negator for other types of predicates, mostly in the domain of realis, namely 無 bô. A third negator, 毋 m̄, incorporates the sense of volitionality, similarly to Macanese non quêro (Lebel, 2018) or even nádi (see above, example 30). Just as in Macanese, the realis-leaning negator 無 bô is used as the main negator for adjectival predicates, and as the negator for the protasis in conditional sentences. Nevertheless, there are also significant differences between Hokkien and Macanese: ascriptive negation in the former is split between 毋 m̄ for the copula and 無 bô for adjectives. Moreover, adjectives too may be negated by the irrealis operator 袂 bē, just as ordinary verbs.

However, we cannot take for granted that the system of negation of whatever Hokkien variety was current in Macao in the past was the same as that of contemporary Taiwanese Hokkien. In fact, in the evolution from Early Southern Min, i.e., the language between the mid-16th century to the late 19th century, into Modern Hokkien, we see some significant changes in the functions of the above-mentioned negators (Yang, 2017). Firstly, 毋 m̄ normally negates both the copula and adjectives in early texts; 無 bô was used in Early Southern Min to negate nouns which, in a sense, ‘act’ as adjectives, i.e., property nouns (Yang, 2017: 113), and then gradually replaced 毋 m̄ as the main negator for adjectives.²⁸ Similarly, 袂 bē used to be a verbal negator which was later extended to adjectives (also due to the verb-like status of Southern Min adjectives; Yang, 2017: 216–217).

While a more thorough discussion of negation in Early Southern Min is obviously beyond the scope of the present paper, in the light of the discussion above we may suggest that the Hokkien variety which was part of the ecology of Macanese in its formative period might have had a system of negation in which (predicative) adjectives were almost always negated by the same negator as the copula and verbs, i.e., 毋 m̄. This negator often had a semantic nuance of volition with verbs, just as in present-day Hokkien: verbs were part of a more articulate subsystem, which included the negator 無 bô and 袂 bē, with their own tam features. Thus, there appears to be evidence of the fact that ascriptive negation was treated differently from standard negation in Early Southern Min, with the latter involving more markers and tam distinctions, similarly to what happens in Macanese.

5.4 Summary

In this section, we briefly discussed the systems of negation of some apc s, and of the main substrates and adstrates of Macanese. In Table 5, we plot the types of negation and tam features of individual negators in the main languages we considered here, comparing them to the situation of Macanese; for the sake of readability, we omit data on existential negation, as well as on other subtypes of negation (imperative, ‘not-yet’, etc.).

Thus, it appears that, among the languages which contribute to the ‘feature pool’ (Mufwene, 2001) of Macanese, Kristang, Cantonese, Early Southern Min and, with a lower degree of consistency in the marking patterns, Bazaar Malay, do seem to have an asymmetry between standard negation and ascriptive negation. However, in none of these varieties the situation is the same as Macanese; the best matches appear to be Early Southern Min and Kristang. In Early Southern Min, standard verbal negation has a three-way distinction between ‘plain’/volitional negation, perfective negation, and irrealis negation; perfective and irrealis are unavailable for ascriptive negation, as expected. In Kristang, only adjectival predicates, but apparently not nominal predicates, may be found in the construction with the irrealis negator nadi.

We may thus hypothesize that the profile of Macanese, in which ascriptive negation, while using the same marker as standard negation (nunca), does not participate in the tam distinctions expressed by standard verbal negators (essentially, nunca vs. nádi), could well have developed from this feature pool, as four languages in its ecology show a comparable split. The combined influence of Kristang, (some form of) Bazaar Malay, Cantonese and (premodern) Hokkien might in fact explain why Macanese seems unique in this respect among apc s, despite the strong convergence with the latter in many (if not most) other domains of grammar. This would be yet another case in which a difference between Macanese and other apc s might be because of the stronger role that Sinitic (and Kristang) has in the feature pool of the Macanese creole (see Arcodia, 2017). Besides, as mentioned above (Section 5.3), other aspects of the Macanese system of negation may have developed following a Sinitic model: specifcially, the imperative negator ne-bom / nunca bom, as well as the use of nunca sã in contrastive focus construction, could be attributed to Cantonese (and Hokkien) influence (Section 5.3).

However, a word of caution is necessary here: since we have no direct evidence of which variety (or, better, varieties) of Malay and Hokkien were actually spoken in Macao, this hypothesis must remain speculative. Moreover, a larger set of data of Kristang and, Especially, Malabar Creole Portuguese, is needed to gain a better understanding of the interaction of reality status with negation in these varieties.

Also, given the limitations of the available data, we cannot but take into consideration the possibility that the specific features of ascriptive negation in Macanese are due to language-internal developments. Generally speaking, when the quality and quantity of the data is limited, a supporting argument for a contact hypothesis would be that the feature at issue is rare, and thus not very likely to arise spontaneously.²⁹ However, the use of the same negator both for standard negation and for ascriptive negation is not uncommon, cross-linguistically. In Veselinova (2015a)’s sample, 24 out of 96 languages (25%) follow this model, which, incidentally, is widespread in Standard Average European languages (including English). Moreover, while Kristang makes use of nadi for the negation of adjectives (see example 33), we found no attestations of the irrealis negator with nominal predicates: given the close relationship between Macanese and Kristang, perhaps the former entirely abandoned nádi for all types of ascriptive negation, and chose instead the more common general negator nunca – a development that, maybe, could already have been under way in Kristang. Even in this scenario, though, contact could still be part of the picture: the combined influence of the varieties discussed above may have reinforced a tendency which already existed in Macanese (if not already in Kristang, as mentioned above).