Registers of African-derived lexicon in Uruguay: etymologies, demography and semantic change

Abstract The present paper deals with 82 words of possible African origin registered in Uruguay by Ildefonso Pereda Valdés and Rolando Laguarda Trías between 1937 and 1965. Many of the lexical items were probably introduced by enslaved Africans brought to the region during the 18th and 19th centuries. Evidence shows that most of the words are apparently shared with varieties of Spanish outside the Rio de la Plata region, and most of them also appear in neighboring Argentina and Brazil. Furthermore, the African-derived lexicon is often used to denominate the ‘other’ with respect to people and social behaviors, and most of these loanwords are nouns with possible origins in Bantu languages spoken in West-Central Africa, which corresponds to the available demographic data.


Introduction
The present paper deals with 82 words of possible African origin that characterize the variety of Spanish used in Uruguay. Our work is based on two written sources, and we do not focus on a specific Afro-Hispanic variety, but rather deal with loanwords that have been registered and that may or may not be still in use today. According to Thomason's borrowing scale (2001, 69), these loanwords appear to be the result of casual contact between speakers of African languages and Spanish, meaning that it was not necessary for the borrowers to have been fluent in the possible African source languages. We believe that the borrowers were most likely speakers of Spanish in a colonial society that received enslaved Africans during the 18th and 19th centuries. Winford (2005) defines this type of borrowing as 'borrowing under recipient language agentivity'. As far as we know, these words have not been studied currently by sociolinguists or anthropologists, nor have linguists studied their vitality using modern analytical tools. In spite of the lack of sources regarding the speech of Africans and their descendants in Uruguay, we do refer to lexicographical works that enable us to get closer to our object of study. Trías (1969) and Pereda Valdés (1937; Two pioneering Uruguayan authors, Ildefonso Pereda Valdés (1937; 1 and Rolando Laguarda Trías (1969), studied the lexical contribution of Africans and their descendants in the Spanish of the Rio de la Plata in the 20th century. In their publications, they compiled lists of supposed African-derived vocabulary in Uruguay that also served as the point of departure for the volume edited by Álvarez López/Coll (2012). Apart from these works, the African-derived lexicon in Uruguay, which is the topic of this article, has not generated greater interest among linguists. The two authors who produced the original lists, published in Montevideo, had very different backgrounds. Pereda Valdés made a fundamental contribution to documentary and historical sources regarding Africans and their descendants in the Rio de la Plata region, and his vocabularies serve as the basis for the work of Rolando Laguarda Trías (1969) on what he calls afronegrismos ('Afronegrisms') in the same region. These research materials are the starting point of our discussion on the African contribution to Uruguayan Spanish. Our corpus consists of the mentioned word stock. The first challenge facing this study is that the two authors who registered the words did not explain how they collected them or how they selected which lexical items to include on their lists. Thus, it is possible that some terms appear on the lists despite not having actually been used in the Rio de la Plata region. The authors may also have found lexical items in the written sources from Argentina and Brazil that they cite and included them in their lists for reasons that we ignore.

Aims of the study
Our first aim is to confirm the reliability of our corpus by corroborating that the included words existed, and that they were present in the region, as well as to verify their meanings. We also intend to discuss whether these words only appear in Uruguayan Spanish, or if they are shared with dialectal varieties of Spanish apart from those spoken in the Rio de la Plata region. For these purposes, various dictionaries and other available sources that include African-derived lexemes used in Uruguay, as well as in the neighboring countries of Argentina and Brazil, are consulted. In order to deepen our understanding of the analyzed materials, we also consult and discuss written sources cited by Pereda Valdés (1937; and Laguarda Trías (1969). By doing so, we are able to identify semantic changes from a diachronic perspective and to shed light on different (sometimes new) meanings of the lexical items.
The second aim of our study is to identify possible etyma for each word, and to systematize the distribution of African-derived lexemes in word classes and semantic fields according to the various meanings of each of the words in Uruguayan Spanish. We will turn to dictionaries of relevant African languages in order to verify their possible etymologies. Finally, we will discuss the relation between the identified etymologies and the demographic data available on the origins of the slave population in this area and present conclusions that may be drawn about the sociocultural context from the semantic and etymological analyses.
In order to achieve our aims, after this introduction, we will briefly present the written sources used by Laguarda Trías and Pereda Valdés in section 2, before we proceed, in section 3, to an overview of the demographic and linguistic profile of the African population in Uruguay in the 19th century. In section 4, we will present the methodology for the lexical study, and our analysis and discussion of the findings is presented in section 5. The paper closes with some final remarks. Finally, our corpus of 82 words with glosses and possible African etyma is presented as an Appendix.

Sources for the word lists of Laguarda Trías and Pereda Valdés
Ildefonso Pereda Valdés (1899Valdés ( -1996, author of the first primary source, was a lawyer who was known for his poetry and essays. His lexicographic work may be questioned, but it represents a very early time in the region and serves as an unparalleled reference for anyone interested in the subject. 2 In 1937, Pereda Valdés published Los pueblos negros del Uruguay y la influencia africana en el habla rioplatense (The Black People of Uruguay and the African influence in the speech of the Rio de la Plata region), which contains a list of words of African origin used in the Rio de la Plata region. In 1965, the author provides a revised version of this vocabulary in a publication that includes other essays on the social and political situation of African descendants in Uruguay. The terms presented in these two publications do not coincide exactly: for example, bunda, catanga and zambomba appear in 1937 but not in 1965, and benguela, luandas and minas appear in the latter work but not in the former. In any case, it should be noted that Pereda Valdés based his comments on the origins of the words on the work of the Brazilian essayist Renato Mendonça (1973Mendonça ( [1933), who focused on the "African element" in Portuguese, as well as on the vocabulary of Daniel Granada (1957Granada ( [1889), 3 a pioneering lexicographic work on the Spanish of the Río de la Plata. However, none of these authors referred to dictionaries of African languages in their work. Laguarda Trías (1902Trías ( -1998, the second author of concern to us in this paper, was a multifaceted researcher who excelled in areas as diverse as geography, historical cartography, military history, lexicography and etymology. He had received a military education and had published several works, among which we emphasize his contribution to what he calls afronegrismos ('Afronegrisms') in the Rio de la Plata. In his article from 1969, he classifies the words into "false Afronegrismos", "negrismos", "authentic Afronegrismos" and "voices that do not have African origin but that were brought from Africa by black slaves and introduced thanks to them, in the speech of the River Plate" (Laguarda Trías 1969, 100 s.). This article also includes words classified by Laguarda Trías (1969) as "Spanish or Native American deformed by the pronunciation of Blacks ..., hybrid voices containing both Spanish or Indian and African elements; and finally, Spanish or Native American voices that present Afro-American suffixes". 4 The following words are included in Pereda Valdés' lists (1937;, but not mentioned by Laguarda Trías (1969): bunda, bundo, cachumba, candombero, catinga, catonga, luanda, mina and mondongo.
Pereda Valdés' work from 1965 is quite explicit with respect to his sources compared to his publication from 1937, as it provides a bibliography; it also includes a short introductory essay on Kimbundu linguistic features, based on Mendonça (1936). However, not all the references that appear in the list are included in the bibliography and vice-versa. 5 Pereda Valdés (1937) quotes literary sources and presents evidence from foreign travelers, such as Don Pernetty and Alcide D´Orbigny. Moreover, at the end of the bibliography he provides the names of eight informants.
The secondary sources cited by Laguarda Trías are more varied than those presented by Pereda Valdés. 6 His bibliography includes works on Spanish and Portuguese, as well as on different dialects and varieties of these languages. The bibliography includes works dealing with topics such as rural speech, slang, Americanisms, etc. He also appealed to chronicles and periodicals, literary sources, historical and anthropological research, and the scarce "bibliography about Blacks in the Rio de la Plata", among which Vicente Rossi (1926) and Horacio Jorge Becco (1953) can be found. Laguarda Trías manages an extensive range of reference 4 Our translation. 5 For example, Vicente Rossi (1926) appears in the bibliography, but not in the text. On the other hand, Luis Carlos de Morais, among others, is quoted in the text, but does not appear in the bibliography. Both in 1937 and 1965, Pereda Valdés appeals to Brazilian and Portuguese sources such as Rodrigues (1935), Mendonça (1936) and Ramos (1940), which appear in his bibliography. He also mentions works without specifying the source: João Ribeiro (without the year; might be his Diccionario Grammatical, 1906), Beaurepaire-Rohan (probably Diccionário de vocábulos brazileiros pelo Tenente-General Visconde de Beaurepaire-Rohan, 1956), Saraiva's glossary (probably Glossario de vocabulos portugueses derivados das linguas orientaes e africanas, excepto a árabe, 1837), and the Grammatica da língua portugueza by Silva Junior (probably 1894). In 1937, the author also quotes P. do Nascimento (probably Diccionario Portuguez-Kimbundu from 1907), and in 1965 he adds Raymundo (1936, probably O Negro Brasileiro e Outros Escritos) and Morais (probably Moraes' Vocabulário Sul-Riograndense, 1935). 6 In addition to the works of Granada (1957Granada ( [1889), Ortiz (1924) and Mendonça (1973Mendonça ( [1933), cited by his predecessor, Laguarda Trías bases his study on lexicographic publications such as Segovia (1911), Raymundo (1933), Corominas (1954Corominas ( -1957 and Gobello/Payet (1959).
Registers of African-derived lexicon in Uruguay literature that gives consistency to his work. 7 He also appeals to several literary sources that describe rural settings or belong to the so-called "gauchoesque" literature, including poetry and narratives. The majority of them are Uruguayan, although there are a few Argentinian sources. Additionally, there are a few references to chronicles or historical documents from the Colonial period.
Summing up, we could say that both Pereda Valdés and Laguarda Trías use some of the most important sources available in their time in Spanish and Portuguese, considering not only lexicographical materials but also literary and historical publications. However, Laguarda Trías seems to present them in a more detailed and specific way.

Demographic and linguistic profiles of the
African population in the region (19th century) In 1778, 29% of the inhabitants of Montevideo were slaves or freedmen (Borucki 2011, 99). In 1791, Montevideo became the only port in the present Southern Cone with the authorization to introduce enslaved Africans from various parts of Africa, and, from that moment, the number of Africans increased in the territory that today is Uruguay. In 1812, 70% of the Afro population in Montevideo was born in Africa, and individuals from West-Central Africa predominated, mainly individuals from present-day Gabon, the Republic of Congo, the Democratic Republic of Congo, and Angola (Borucki 2011, 101;Eltis n.d.). The introduction of Africans to the Rio de la Plata between 1777 and 1812 has been described as the most important demographic event after colonization: "The increasing number of slaves arriving in the Viceroyalty of the Río de la Plata was the most important demographic event since the Iberian colonization to this region. At least 70,000 slaves arrived in the Río de la Plata from Africa and Brazil between 1777 and 1812, which is surprising given that Buenos Aires had only 43,000 inhabitants by 1810 and Montevideo no more than 12,000 by 1803 (Arredondo 1928;Johnson and Socolow 1980). Between 1778 and 1810, the population of Buenos Aires grew 34 percent, while the slave population increased 101 percent. In Montevideo, the total population grew 119 percent between 1791 and 1810, while the slave population increased 486 percent (Campagna 1990)" (Borucki 2011, 52).
It is important to note that of these 70,000 individuals, a total of approximately 28,000 were imported directly from Africa, with the rest coming from Brazil, mainly from Rio de Janeiro (Borucki 2011, 59). Brazilian researchers reveal that nine out of ten of the trips performed by Rio de Janeiro's major traffickers had targeted the Congo-Angolan region in West-Central Africa, and that two of three of their expeditions were directed to a single African port, Luanda, Benguela or Cabinda, in present-day Angola (Florentino/Vieira Ribero/Domingues da Silva, 2004, 101). According to Borucki (2011, 66), 97% of the slaves who entered Rio de Janeiro between 1777 and 1812 had been shipped from West-Central Africa (especially from the present republics of Angola and Congo), and, among these, 85% were brought from Luanda or Benguela. Many of these captives were transported and sold in the south and ended up in Montevideo. In that same period, a minority of Africans, who came to Montevideo mainly from Salvador, Bahia, were probably from the Gulf of Benin in West Africa (especially Ghana, Benin, Togo and Nigeria) as were the enslaved Africans introduced to Salvador during the same period (ib.). Table 1 shows, in the first column, the African regions from which the captives came to Montevideo, and the second column includes the names of socalled African "nations" present in the census of 1812. The third column identifies the language that this group supposedly spoke, and the fourth column specifies the number of individuals from each African nation found in the demographic data of Montevideo (Borucki 2011, 131 s.;Álvarez López 2012, 60-62). Although it is difficult to identify the languages based on the registered nation names that were used as ethnonyms in colonial times to designate enslaved Africans, it is a reasonable way of limiting the number of African languages that come into play in etymological studies for a certain geographic and historical space (cf. Álvarez . These data show that the groups with the greatest demographic weight come from West-Central Africa and are speakers of Bantu languages such as Kikongo, Kimbundu and Umbundu. This group was followed by the so-called Mina, speakers of Gbe languages (such as Ewe, Fon), among other languages (cf. Álvarez López 2015, 59-62), and the group called Mozambique, speakers of Bantu languages in Southeast Africa (such as Yao and Ronga, among other). Considering all groups presented in the secondary sources that were available, the data include nation names applied to a total of 706 individuals in Montevideo. Although it does not appear in the Montevideo census, cabinda is included because it appears in other historical sources (de María 1957(de María [1887(de María -1895; Kandame 2006).
In brief, the presence of speakers of African languages, mainly from West-Central Africa, was considerable in Uruguay from the end of the 18th century, and new speakers of African languages were being introduced up until the mid-1830s (Borucki 2012, 16). 8 Montaño (2008,(291)(292)(293) gives a figure of 4,220 individuals of African origin (including those born in the Americas). We considered Borucki's (2011) data in order to limit the numbers to individuals born in Africa, that is, possible speakers of African languages.

Methodology for the lexical study 4.1 Data selection
The lexical database in this study is based on a total of 100 words and expressions presented by Pereda Valdés (1937; and Laguarda Trías (1969). Words that presented orthographic, phonetic or morphological variants, but the same meaning, were put together and counted as one entry: cachimbo was joined with cachimba; cacunda with carcunda. Words belonging to different word classes were maintained apart, as in the case of bombear (verb) and bombero (noun).
The words mentioned above do figure as Africanisms in other sources, however, and we therefore decided to explore their origins. For the present paper, we excluded 18 words that were clearly not Sub-Saharan (may be from Arabic, Indigenous languages, Italian, Latin or Spanish) and words that were not attested in the Rio de la Plata region by the authors or their sources: bombo, bozal, bugía, cachafaz, calenda, cebra, escobero, gramillero, mangangá, mataco, moreno, mota, motoso, ña/ño, papagayo, tarimba, tía/tío, zarambeque. Since Laguarda Trías (1969) does not give the meaning of all the words that he excludes and that we decided to include, we consulted two dictionaries that focus on Rio de la Plata Spanish: Ortiz Oderigo (2007), compiled before the author died in 1996, and Bermúdez/Bermúdez (1890-1947, also a dictionary that focuses on Rio de la Plata Spanish, in order to identify their meaning in the Rio de la Plata region during the first half of the 20th century. The result of this procedure is a corpus of 82 words (see Appendix).

Verification of the list
To verify that the words are known in the region, we consulted and compared registers of the words in recent and reliable dictionaries of Spanish and Portuguese, as well as glossaries with African-derived lexical items registered in Argentina and Brazil.
The three Spanish dictionaries considered suitable for our purposes are the dictionary of the Asociación de Academias de la Lengua Española, Diccionario de la Lengua Española (DLE, online), and two dictionaries of Uruguayan Spanish: one edited by the Academia Nacional de Letras del Uruguay, Diccionario del Español del Uruguay (DEU, 2011); the other belonging to the collection Nuevo Diccionario de Americanismos, by Günther Haensch & Reinhold Werner: the Nuevo Diccionario de Uruguayismos, by Kühl de Mones (NDU, 1993).
The Brazilian sources consulted were the updated general dictionary Grande Dicionário Houaiss da Língua Portuguesa [online] and Angenot et al. (2013), a glossary with 3,704 entries limited to Bantu-derived words found in 112 different studies conducted in Brazil. In order to include non-Bantu derived African lexicon, we also consulted Castro (2001) and Álvarez López (2004), two studies based on data collected in Brazil. For Argentina, we consulted a dictionary of Africanisms in the Rio de la Plata region, written by Ortiz Oderigo (2007), but published after his death. Ortiz Oderigo (ib.,24) mentions that he consulted various old Afro-Argentinians who shared their knowledge with him. These registers will be compared in section 5.1.

Identification of etyma and word classes
The methodology for the etymological study aims at identifying the origins of the collected words (see 5.2). The dictionaries were selected to cover African languages spoken in the regions from where Africans were brought to Montevideo, as presented in section 3. This will make it possible to confirm the existence of the words in Africa and to shed light on their possible etymologies.
We included dictionaries of three languages spoken in Angola (Kikongo, Kimbundu, Umbundu), Ghana (Ewe) and Mozambique (Yao), in order to at least consult one dictionary from each relevant region or language group presented in section 3 (Cannecattim 1804(Cannecattim , 1859Dupeyron 1880;Sanders/Fay 1885;Matta 1893;Delafosse 1894;Bentley 1895;Nascimento 1907;Johnston 1919;Laman 1936;Júnior s.a.;Maia 1961). Almost all these dictionaries were published near the end of the 19th century or the beginning of the 20th century, which is important, given that we are discussing African-derived words that were probably introduced by speakers of African languages during the 19th century.
We believe that word classes are relevant since earlier studies suggest that nouns are more easily borrowed and exchanged than terms in other word classes, and that they usually predominate among the African-derived words in contact varieties (Bartens/Baker 2012), as well as in our corpus. For nouns that can also be used as adjectives, such as the case of the nation names, we computed them as nouns since we believe that they were borrowed as such. Bartens/Baker (ib.) relate the study of word classes and etymologies by affirming that etymologies of lexical items other than nouns can reveal which languages were involved during the period of formation of a linguistic variety, and such results can be related to demographic data from the region. However, our data does not necessarily represent a specific Afro-Uruguayan variety, but rather loanwords identified as African-derived lexicon in earlier studies with a focus on Spanish.

Classification in semantic domains
Considering our classification in word classes, it is worth noting that nouns tend to be borrowed particularly when new or unknown objects and concepts need to be named (Thomason/Kaufman 1988, 77). Consequently, basic vocabulary should be less likely to be borrowed than more culture-specific African vocabulary or specific words and expressions that may be associated with, for example, slavery. In order to verify if the words in our list can be associated with the specific sociohistorical context as opposed to the lexical items that may be identified as part of a basic vocabulary according to Swadesh's (1971) 9 classical set of basic concepts for comparative linguistic studies, we compare our list to Swadesh's in section 5.3.
To further comment on the semantic aspects, we organized the lexical items in the following domains: universe and creation (animals, nature, etc.), person (people, body, etc.), language and thought (emotions, communication, etc.), social behavior (religion, relationships, etc.), daily life (home, food, etc.), work and occupation (work activities, tools), physical actions (posture, movements, etc.), and states (quantity, quality, time, etc.). These domains are clearly specified with subdomains in the Fieldworks Language Explorer software used to organize the lexical database (<http://fieldworks.sil.org/flex/>). Such classification left no "unclassified" items and makes it possible to compare the results with those of an analysis based on the same methodology regarding an Afro-Brazilian variety (Álvarez López/Jon-And 2017).
The analysis in 5.3 will also approach the specific semantic changes observed due to negative values associated to Afro-Latin-American cultural expressions. To identify semantic changes, we consulted the sources used by the two authors and the dictionaries and vocabularies mentioned above.
After clarifying the methodology related to data selection, and the study of semantic domains, words classes and etymologies, we proceed to our analysis.

The lexical items in Spanish dictionaries
In this section, we present the results of a search for the lexical items in the selected Spanish dictionaries.
The Diccionario del Español del Uruguay (DEU), 14 published by the Academia Nacional de Letras del Uruguay, in 2011, registers 35/82 (43%) words with equal or similar meanings to the words in our corpus (different meanings were registered for catanga, conga, mondongo). 15 It is interesting to note that 13 of the words have an explicit indication of African origin: cachimba, cachiquengue, canyengue, 16 catanga, matungo, muleque, mucama, muleque, muyinga, quibebe, quilombo, quitanda, yimbo. Those for which an African etymology is not given often have a Portuguese origin or a controversial etymology, or an etymology that is being studied. 17 The search in the DLE shows that approximately 50% of the words are not registered for Spanish, and that three words marked as Uruguayan or Argentinean do not appear in the dictionaries of Uruguayan Spanish: cantimpla, dengue and 12 The words registered as being used only in Argentina are cantimpla, dengue, malambo, milonga. Catanga appears to be from Argentina but with a different meaning. 13 The 26 words in NDU are banana, batuque, bombear, bunda, cachimba/cacimba/casimba, cachimbo/cachimba, cafua, candombe, canga, canyuengue, capanga, catinga, cuco, fula/fulo, lubolo, malambo, mandinga, marimba, matungo, milonga, mucama, quibebe, quilombo, quimbo, tango, tilingo. 14 It must be taken into account that the DEU only registers words and expressions that are not part of general or standard Spanish, as registered in the 22 nd edition of the Diccionario de la Real Academia Española (DRAE). It is not strictly a dictionary of words used only in Uruguay, since it includes words that are exclusive for the country, but also those that are shared with Argentina, Chile, Paraguay and other Spanish-speaking regions (DEU, 15). Muleque was included in the DEU, as in our corpus, which means that the DLE (online) probably made a mistake when labeling the word only as Cuban. For this paper, we consulted the 23 rd edition of DRAE, which has changed its name to DLE (online zamba. These three dictionaries are not completely independent from each other, although there are obvious differences in which words they consider to be shared by varieties of Spanish outside Uruguay. The 22/82 (26%) words present in our corpus that are not registered in any of the three dictionaries are banguela, bubango, bundo, cabinda, cachumba, calengo, camunda, capiango, carimbo, casanche, catonga, macota, magises, malungo, maxixa, mazagaya, menguengue, mina, ondú, pango, yimbo, zambomba. All those words were found in other sources.

The lexical items in Brazilian and Argentinian sources
Although many of the words are considered to belong to Uruguayan Spanish by the consulted Spanish dictionaries, the register of the same lexical items in Argentinean and Brazilian sources confirms that those items were used and that they circulated, and were most likely not unique to Uruguay, at least not before the 1960 s. All the lexical items in our corpus except eight 18 were found in Brazilian sources; sixteen 19 of them have other meanings in Brazil, and seven appear in similar but not identical forms. 20 The lexical items that do not appear in the Brazilian sources can be found in Argentina (cf. Ortiz Oderigo 2007). Moreover, Ortiz Oderigo (ib.) includes most of our words (apart from cachiquengue, calengo, camundá, casanche), and most of them appear with the same meaning.
The findings show that the lexical items registered by Pereda Valdés and Laguarda Trías are shared with both Spanish and Portuguese in neighboring regions, although there are cases in which the registered meanings seem to differ according to the regions or to the source.

Distribution in word classes
Not surprisingly, as shown in Table 2, nouns are the most representative word class.

TOTAL: 82 words
The interjection (cachumba) was not found in the consulted dictionaries of African languages.

Etymologies and word classes
We found possible etymologies for 63/82 (77%) of the words in the consulted dictionaries of African languages. Most of those 63 words considered to be African-derived are nouns and have possible Bantu origins, as shown in Table 3. In the dictionaries of African languages, we found 34 (55%) possible African etyms that matched forms and meanings with the words of our corpus. 21 Moreover, 18 (29%) possible etyms had similar forms and/or were considered as reasonably derivable from the original meaning. 22 In addition, nine possible etyms were considered dubious because the form and/or meaning was quite removed from the registered form and/or meaning. 23 Finally, two items, candonga and capanga, were considered dubious because they were not found in the dictionaries, but were only listed by Castro (2001) without specification of the meaning of the etyma. The sum of nine and two gives us a total of eleven (17%) dubious possible etymologies. Possible etymologies were not found for 19 items. 24 It is worth noting that 60/63 (95%) of the words with possible African etymologies are nouns, a word class that is more easily borrowed than verbs or adjectives. Among the 60 nouns, we have 10 cases of nouns derived from African toponyms in regions where Bantu languages are spoken, such as calengo or camunda. 25 Those have been classified as having possible Bantu origin, because they were not found in specific dictionaries. Two African "nation names" have been classified as having possible Gbe origin, since they refer to people from the Gbe-speaking areas (mina and magises). The proportion of nouns seems to be high when compared to the results that Bartens/Baker (2012) obtained from the analysis of 17 varieties of transplanted European languages, for which the average proportion of nouns in the African-derived lexicon is 74%.
Possible Bantu etymologies were also identified for one verb (Kimbundu) and two adjectives (Kikongo). According to our findings, both the adjectives and the verb may derive from nouns, as they are not registered as adjectives in the consulted dictionaries. Based on these findings, we have no support for Baker's hypothesis that groups with higher demographic weight in the earliest periods of colonization contribute more words in general and more lexical items that belong to word classes other than noun (Baker 2012, 283). Furthermore, the verb bombear and the Portuguese noun bombeiro were used in colonial Africa to refer to the action of 'to spy' and 'spies' in the sense of somebody looking for slaves in the countryside, and the term may have been introduced into the Rio de la Plata region by Portuguese slave traders.
The majority (59/63) of the words were identified as Bantuof which 11 are derived from toponyms. Without the 11 toponyms, we have 48 words identified as Bantu; 40 were found in Kimbundu dictionaries (22 of those were also found in Kikongo, 1 only in a Kikongo dictionary, and 4 in Umbundu dictionaries). This means that the lexicon is predominantly Bantu, and most of the lexical items could be found in Kimbundu dictionaries.

Relation to demographics
According to the census of 1812, there were 706 slaves living in Montevideo, of which 343 were denominated with nations' names that indicate that they were from the Congo and Angola regions, and probably spoke Bantu languages that could be found there (Álvarez López 2012, 65 s.). Moreover, 100 slaves came from Bantu-speaking areas in South-East Africa and may have spoken Bantu languages from Mozambique. Additionally, 159 came from areas where Gbe languages were spoken (including Akan and Ga-Adamngme), and 4 were registered as Fula and Hausa.
Given that 94% of the words were identified as having possible Bantu origins by consulting dictionaries of Kikongo, Kimbundu, and Umbundu, as well as a comparative Bantu dictionary (see Table 3), we can affirm that there is an expected relationship between demographic and etymological data.

Distribution in semantic domains
None of the words can be classified as being part of a basic vocabulary based on Swadesh's (1971) classical list of basic concepts for comparative linguistic studies. According to Thomason (2001, 69), "non-basic vocabulary items are the easiest to borrow, since in most languages a new noun (for instance), or even a verb, can be inserted readily into existing constructions". Most of our words fall within the domains of 'social behavior' and 'person', which may be associated with the sociocultural context.

Relation to sociohistorical and cultural context
Regarding social history, the distribution of the words in semantic domains shows that 58/82 (71%) of the words are found within the domain of 'social behavior' and 'person'. The majority of the lexical items related to social behavior (18/34) and designating persons (20/24) can be associated with Africa or Africans. There are also many words that may be associated with everyday life in colonial times, such as bombear 'to spy (look for slaves)' and cachimba ('water well'), and music and dance among slaves, such as candombe 'a dance', marimba 'an instrument'.
From a diachronic perspective, it clearly appears that a number of Africanderived words that refer to African culture or slavery have gained a negative or derogatory meaning in the Spanish of Uruguay. For example, batuque, cachicandonga, canyuengue and quilombo 26 have acquired a meaning of 'hullabaloo', 'confusion' or 'mess'. Candombero has a negative meaning in its figurative use. Some nicknames or address forms (like musinga, yimbo, zambaigo) are pejorative, although no negative value has been registered for curimba. Catinga refers to an African person's supposedly strong 'body odor', and mandinga is nowadays strongly associated with the 'devil'. The word milonga, on the other hand, might have a pejorative meaning in Kimbundu, but it is not clear that it had it in Spanish. Other African-derived words not associated with the African world or slavery have acquired new meanings related to prostitution, such as quitanda and quitandera, 27 as well as terms linked to criminal activities, such as capiango and capanga. Quibebe originally referred to 'sweet pumpkin porridge' in Kimbundu, but afterwards it developed a meaning closer to that of batuque, cachicandonga, canyuengue and quilombo (i.e. 'confusion, mess'), while banana, as is the case for many words that name fruits or vegetables, can be used as an insult (as in es un banana, meaning 'he is silly'; cf. English 'go bananas', Traora/Siaka 2016, 17).
In this section, we have seen that as much as the African words of our corpus can be associated with everyday life in colonial times, many of them have gained a pejorative meaning by being associated to Africans or Africa in the course of time.

Final remarks
The present paper confirms the reliability of the registers of African-derived lexicon in Uruguay. It also verifies that most words seem to be shared with varieties of Spanish outside the Rio de la Plata region, and most of them appear in Argentina and Brazil.
The result of the analysis of their African etymologies and their distributions in word classes showed that most loanwords are nouns with possible origins in Bantu languages spoken in West-Central Africa. Thus, the etymological analysis indicates that Bantu languages from West-Central Africa dominate, which corresponds to the available demographic data.
With respect to semantic domains, 'social behavior' prevails together with 'person': the African-derived lexicon is often used to denominate the 'other' as an African person or as a social behavior associated with Africa. From our study, we can also conclude that many of the words of African origin have gone through processes of semantic change and have developed pejorative or negative meanings in the Spanish of Uruguay.
Having analyzed various aspects of this lexicon from a historical perspective, we believe that it could be interesting to study the vitality of these words in Uruguay today. Alkmim/Petter (2008) show that many of the African-derived lexical items found in Brazilian registers are currently obsolete, and that may also be the case of a number of the words in our corpus, but this requires further study.