Printer Friendly

On the assignment of gender to Chicano Anglicisms: Processes and results.

Robert N. Smead (*)

The creation of neologisms through linguistic borrowing is a complex process. For donor language models to be transformed into host language replicas, phonological and morphological adaptation must take place. Part and parcel of that adaptation is the assignment of gender to loanwords in a host language like Spanish. The present article examines gender assignment in two corpora of loanwords (New Mexican and Southern Colorado Spanish: N=212 and General Chicano Spanish: N=595). Through the application of variable rule methodology (GoldVarb 2), five factors are posited as significant. Two of these categorically determine gender assignment when present: (1) the biological sex of the referent and (2) the presence of a derivational suffix (or a sequence which can be interpreted as such), additionally, gender assignment for these Anglicisms responds variably to three other factors: (3) the terminal phoneme(s) of the loanword, (4) the gender of a common Spanish synonym, and (5) the gender of the Spanish hyperonym.


In a language like Spanish, gender assignment, i.e., the assignment of a noun/nominal to a particular category (such as masculine, feminine, or neuter) has morphosyntactic repercussions for the construction of a noun phase (NP). Though a process known as agreement, other noun phase constituents (e.g., determiners and adjectives) must demonstrate concord with the nucleus. Furthermore, anaphoric reference to the nucleus must also be marked for agreement. Thus, while gender may be considered a nominal classification system, its effects transcend the lexicon.

Corbett (1991) notes that all known gender systems have a semantic core in which many animates are assigned gender on the basis of biological sex. However, this criterion, known as genero natural in Spanish, plays only a limited, indirect role in the assignment of gender to nouns whose underlying referent is inanimate. For instance, some nouns may be treated as if their underlying referent were animate and assigned a gender based on characteristics or qualities perceived as masculine or feminine. The Chicano Anglicism plogue (mf.) 'plug' illustrates this principle; presumably the masculine refers to the plug end of the electrical cord while the feminine refers to the receptacle or outlet. This interpretation is reinforced by the semantic extension of plogue/ploga (f.) to mean 'kept woman or mistress' and the gender is derived directly from biological sex. It is also hypothesized that some nouns are assigned gender through metonymy; indeed, Espinosa (1914: 286) claims that some Anglicisms that take masculine gender in New Mexican Spanish do so because they denote "masculine articles of clothing, male occupations, or mechanical instruments or new machinery and inventions." As will be demonstrated, however, gender assignment in such cases responds to formal criteria that apply to nearly all nouns with inanimate referents. That is, phonological shape determines to a great degree what Spanish terms genero gramatical.

Spanish nouns/nominals must be marked as masculine or feminine (the neuter occurs only with pronouns, e.g., esto 'this' and in constructions like lo bueno 'the good thing' or lo de Maria 'that business about Mary' and refers either to unidentified items or a complex situation). Recent studies such as those by Bergen (1980); Teschner (1983) and (1987); Teschner and Russell (1984); Teschner and Alatorre (1984); and Arias Barredo (1990) note that both semantic and formal (principally phonological) criteria may determine gender assignment in Spanish. However, certain restrictions and complications exist. (1) With regard to animates, the biological sex of the referent normally takes precedence over the phonological shape of the noun: e.g., el camara 'the cameraman'; la modelo 'the female model' except in the case of:

Epicenes; that is, when the gender of the noun is invariable. Some examples include: angel 'angel' and individuo 'individual' (both m.), and estrella 'star' or 'celebrity' and persona 'person' (both f).

The names of certain animals in which biological sex can be distinguished only by the addition of the invariant forms macho 'male' or hembra 'female': la ballena macho 'the male whale' and el castor hembra 'the female beaver.' Such terms retain the grammatical gender assigned to them regardless of sex.

Collective nouns which, unlike pronominals, maintain the same gender regardless of whether the group is mixed or comprised of individuals of the same sex; thus a large gathering of people may be referred to despectively as la chusma or el gentio 'mob.'

Masculine plural nouns which may refer to a mixed group, a male/female pair, or a collection of males: los padres maybe glossed as 'the parents,' 'the father and mother,' or 'the fathers.'

Names which originally referred to a male individual and are applied metaphorically to females: un Hitler con faldas 'a Hitler in skirts' and un Franco femenino 'a female Franco.'

With regard to inanimates, the phonological shape of the word becomes the primary criterion for gender assignment. Applied linguists such as Bull (1965) and Bergen (1978) have suggested the acronyms NORSEL or LONERS as a mnemonic device for those endings which consistently take masculine gender. It should be noted that each letter represents a word-final phoneme or grapheme. In addition, -s excludes -is and -n excludes -ion. According to the summary provided in Whitley (1986: 146), these endings correlate with masculine gender in approximately 90-100% of the cases.

In a similar vein, Chastain (n.d.) has proposed A-D-ION-IS as an acronym for feminine gender. If one stipulates that -ion be expanded to -cion or -sion and -is to -sis/-tis, Whitley indicates that these endings correspond to the feminine in about 97-100% of the cases.

While both Bull's and Bergen's research has proven invaluable to beginning and intermediate students (and serves as an excellent point of departure for a discussion of Spanish gender), many contradictions remain. Other linguists have furthered our understanding and have revealed a complex set of patterns that respond to both phonological and morphological factors. For example, Teschner (1983), utilizing a reverse dictionary based on the Royal Academy's Diccionario de la /engua espanola (18th edition, published in 1956) found that some 60% of words ending in -z are feminine (note that the acronyms given above do not treat this ending despite the fact that it is a fairly common one). He posits subcategorization as an essential step: nouns ending in -ez (particularly when -ez is a derivational morpheme denoting 'quality or characteristic of') and in -briz/-briz/-triz are predominantly feminine while all others (-iz, -az, -oz, -uz) are predominantly masculine.

As the previous example illustrates, gender assignment may correlate only weakly with phonological shape. Other complexities that lead to contradictions in gender assignment among inanimates include such factors as:

Diachronic variation: In some cases gender assignment has changed from feminine to masculine. The older feminine assignment is now considered archaic (or rustic). Examples include puente 'bridge,' ca/or 'heat,'fin 'end,' andlisis 'analysis.'

Diatopic/diastractic variation: Gender may vary in accordance with regional and social factors. In some cases, the terminal phoneme(s) of the word may change to reflect the assignment: el/la sarten 'frying pan,' el/la mar 'sea,' el bombiio, la bombilla 'lightbulb,' etc.

Homophony: some pairs that share the same phonological form are distinguished by gender assignment: el corte 'cut,' /a cone 'court;' la manana 'morning,' el manana '(to)morrow,' etc.

Ellipsis or metonymy: an associated term, generally the term designating a superordinate category, determines gender assignment (here the term in parentheses is the one that determines gender): la (emisora) nzimero uno 'the number one (radio station)'; la (compania) IBM'the IBM (company)'; el (carro) Toyota 'a Toyota (car)'; el (equipo) Cruz Azul 'the (soccer team) Cruz Azul'; el (rio) Amazonas 'the Amazon River', etc.

The purpose of this paper is to examine gender assignment for two corpora of Chicano Anglicisms. It will be demonstrated that gender assignment is governed by hierarchically ordered competing factors. Two of these factors categorically determine gender assignment for all nouns within their purview, namely, (1) the biological sex of the referent and (2) the presence of a derivational suffix (or a sequence which can be interpreted as such); additionally, gender assignment for these Anglicisms responds variably to three other factors: (3) the terminal phoneme(s) of the loanword, (4) the gender of a common Spanish synonym, and (5) the gender of the Spanish hyperonym. Furthermore, it will be noted that while the process of gender assignment in the Chicano community has been regularized and simplified, it follows monolingual norms and tendencies with few significant deviations.

Review of the Literature

The question of whether native speakers have valid intuitions regarding the gender of unfamiliar nouns was studied by Natalicio (1983). In her study, 100 native Spanish-speaking undergraduates at the University of Texas El Paso were asked to assign gender to 44 uncommon Spanish nouns. Exactly one-half of the items corresponded to each gender. Furthermore, each set of 22 items was again halved into those that followed Bull's rules (regarding phonological shape and gender assignment) and those that did not. She found that her subjects were most successful in assigning the correct gender to those nouns that followed Bull's rules: 85% of the masculine nouns and 76% of the feminine were assigned to the correct gender. However, they were much less successful when the item in question deviated from the expected gender assignment: only 37% of the masculine and 20% of the feminine were correctly assigned. Moreover, Natalicio (1983: 53) states that "they disagreed significantly [with the normative assignment] in the ca se of two endings, -d and -sis/-itis." While Bull found that 97% of nouns ending in -d and 98% of those ending in -sis/-itis were feminine, Natalicio's subjects preferred the masculine at 67% and 60%, respectively. She concludes that with the exception of the two endings noted above, her subjects had internalized something akin to Bull's generalizations.

Chaston (1996) investigated the largely anecdotal claim that the Chicano community is characterized by widespread gender vacillation and nonstandard gender assignment (e.g., el/la calle (f.) 'street' and la programa (m.) 'program,' respectively). In order to quantify and contextualize the phenomena in question, sociolinguistic interviews were conducted with 15 Spanish-English bilinguals of Mexican heritage (ages 18-24) studying at the University of Texas. He discovered that only 7.3% of the 562 article/noun combinations extracted from the interviews deviated from the norm with regard to gender; only six nouns exhibited gender vacillation (defined as an initial error followed by a subsequent correct assignment made in the same stretch of discourse by the speaker him/herself). Five speakers exhibited standard usage; that is, their use of gender in article/noun combinations was judged 100% accurate. The other 10 speakers ranged from 97% to 75% accuracy. According to quantitative analyses (performed by GoldVarb 2), the language used by the consultants in conversation with their parents correlates significantly with standard usage. Individuals who speak only Spanish (that is, without resorting to code-switching or language mixing) rarely or never commit errors in agreement; errors increase with the amount of English spoken so that speakers near the end of the bilingual continuum who use English exclusively with parents tend to exhibit nonstandard (inaccurate) usage. In terms of linguistic factors, three terminal phonemes (TPs) were associated most often with agreement errors: -l, -d, -cion. Chaston concludes that while gender vacillation and nonstandard gender agreement do occur, they are not widespread among bilingual Mexican-American university students.

Garcia (1998) also addresses the issue of gender marking among speakers of a Southwestern variety of Spanish. Specifically, she analyzed 904 NPs produced in free conversation by 11 Spanish-English bilinguals residing in San Antonio. While no formal measure of language proficiency was administered, she describes her consultants as fluent in conversational topics in Spanish. Her results confirm Chaston's findings reviewed above: Among these speakers, there were few instances of gender vacillation (with a handful of nouns ending in -e and escuela 'school') and nonstandard gender assignment (principally with programa and other nouns of Greek origin with ending in -a; there was a marked preference for the feminine rather than the standard masculine). Garcia calculates that 94% of the determiners agreed with masculine nouns and 96% agreed with the feminine. Noun/adjective agreement was lower; 92% of adjectives agreed with their masculine heads (as contrasted with 83% accuracy for feminine agreement). She rightly c oncludes that gender marking is a feature of Southwest Spanish and parallels the norms and tendencies of standard Spanish for the most part.

The problem of gender assignment for loanwords has received attention in numerous languages. Both Ibrahim (1973: 51-62) and Corbett (1991: 70-82) provide accessible summaries of the factors thought to operate in a wide variety of language dyads. While some of the factors must be discounted in this case (e.g., assignment of the model's gender to the replica or assignment on the basis of morphological class), many of the factors mentioned have been observed or empirically tested in the case considered here (loanwords into Spanish). (2)

Some linguists who have studied gender assignment for loanwords in Spanish have concluded, as does Prado (1982: 259), that the process is highly predictable and results in a large majority of the replicas being assigned to a single gender. Most linguists, however, would agree with Wagner (1990: 62), who notes instances of variable assignment and suggests that the process is not straightforward. Indeed, as will be seen, it is beset by multiple and potentially conflicting factors.

Apart from Espinosa (1914), who adduces both analogical factors (as cited in the introduction) as well as phonological factors for gender assignment to English loanwords in New Mexican Spanish, Zamora (1975) represents the earliest attempt in recent years to elucidate the problem. While he also examines gender assignment for loanwords from three indigenous languages (Arahuacan, Nahuatl, and Quechua), only the sections that treat gender assignment for English loanwords and switches will be reviewed here.

Thirteen educated bilingual Puerto Ricans ranging in age from 20 to 55, and residing in the Northeastern United States were selected to participate in his study. All but two of these had studied Spanish formally. The subjects responded to two written surveys: the first required that they construct Spanish sentences utilizing 10 common English words that do not function as loanwords in Puerto Rican Spanish (in order to examine how incipient loanwords are assigned gender); the second consisted of a list of 20 Puerto Rican Anglicisms to which the respondents supplied the definite article and their reasons for gender selected. Zamora states that in the first survey the assignment was made with one exception on the basis of the gender of the translation equivalent while three factors were adduced for the second survey: phonological shape (primarily determined by the TPs), semantic association (gender of the translation equivalent), and the gender originally transmitted and learned with the loanword. (3)

While Zamora's research represents a valuable step in researching gender assignment for Anglicisms in Spanish, it is subject to several criticisms. First, as in other research which will be reviewed herein, the tasks which his subjects completed are highly artificial and may not reflect actual usage in some instances. In particular, the validity of the first task appears doubtful. Since the items in question represent either core vocabulary or high-frequency terms in both English and Spanish, it is unlikely that they would be borrowed. Second, while it is entirely possible that his respondents utilized a semantic approach to assign gender to the English terms in question, this remains unproven since, in many cases, the terminal phoneme also correlates with the gender assigned. For example, four of the terms utilized, floor, pen, glass, and pencil, take masculine gender on the basis of the TP employed (as well as being paired with masculine translation equivalents: piso/suelo, boligrafo, vaso/vidrio, and lapi z, respectively). Interestingly enough, he states that the term cup (also included in his list of ten terms) and pronounced [ko] by his subjects was assigned masculine gender (which does not correlate with the gender of the translation equivalent taza). As he himself states, gender assignment follows formal criteria in this instance, and is based upon the pronounced form and not its written equivalent. Therefore, the model source (written vs. spoken) and mode of transmission (often represented by the degree of phonological adaptation or integration) are also important variables to be considered.

Barkin (1980) investigated gender assignment to Anglicisms via a pictorial questionnaire and free conversation. Her informants consisted of 33 Mexican American migrant workers residing in Florida. She claims that phonological integration plays a role in assigning gender: unassimilated forms were either not assigned a gender or the informants vacillated in selecting a single gender, while partially assimilated/wholly assimilated loanwords were assigned gender in a more consistent fashion. These findings, however, have not been substantiated by other researchers.

Poplack et al. (1982a) examined gender assignment for Anglicisms in Montreal French and Nuyorican Spanish. (4) The Spanish corpus consisted of some 300 hours of taped speech recorded in East Harlem, New York, from 16 residents (children, parents, and elders). Utilizing variable rule methodology (which allows for all competing factors to be considered simultaneously), these researchers found that gender assignment for English loanwords in this variety of Spanish responds primarily to three factors: physiological gender (biological sex), phonological gender (based on TPs), and analogical gender (gender of the translation equivalent), in that order. They also discovered that the degree of phonological integration influenced the assignment of gender in that unintegrated forms were more likely to be assigned masculine gender. While easily the best executed and most comprehensive study of its kind, we shall have occasion to question two of the conclusions reached by these researchers.

Utilizing visual stimuli, Banfield (1994) elicited gender assignment for 54 loanwords among 29 Mexican migrant farm workers (all with minimal English skills) from San Luis Obispo County; California. While some unassimilated loans received ambivalent treatment, the masculine gender was assigned most often (52% of the loanwords were accompanied by a masculine determiner or adjective as opposed to 35% for the feminine). He provides evidence that two factors, namely phonological shape and analogical gender, operate for these incipient bilinguals. His study is marred by two flaws, however. First, he appears to confuse phonological integration with lexical adoption/assimilation. Second, it is unclear whether the oral model (provided when the informant could not identify the object pictured) was presented with English or Spanish phonology. If it was presented with English phonology, the informants, due to their limited English proficiency, may have failed to establish phonemic correspondences between the two langua ges (thus misidentifying the phonological shape of the loanword).

In her study of English loanwords appearing in the Spanish and Mexican press, Sanchez (1995: 134-37) catalogues various factors as determinants of gender assignment. Most of these have been discussed already and do not merit further treatment here. She does, however, make three important contributions to a list of possible factors. Sanchez notes that the masculine functions as the default gender--that is, if other formal or semantic criteria do not unequivocally identify the loanword in question as feminine, it is assigned to the masculine. She also observes that not only can a loanword be assigned a gender on the basis of its translation equivalent, but it may inherit the gender of its hyperonym. Thus, since deporte 'sport' is masculine, the specific sports themselves are assigned masculine gender. These include some loans which would be assigned to the masculine on the basis of their TP: waterpolo, futbol, and tenis, as well as others for which the masculine could not be predicted: golf, surfing, etc. This factor is one that operates in monolingual varieties of Spanish (often termed elliptical or metonymic gender as noted in the introduction) and has not been empirically investigated or differentiated from analogical gender (gender of the translation equivalent). Lastly, she also posits morphological adaptation as a pertinent factor. That is, the addition of a derivational suffix which is associated with a particular gender (e.g., -iza in goliza 'a series of goals') determines the assignment of feminine gender to an originally masculine noun gol. Again, I am not aware of a previous study that takes into account this factor.

Clegg (1997) reports on an experiment with 40 native Spanish speakers (20 males and 20 females). Each participant assigned a gender to 20 potential English non-words, selecting either the masculine or feminine definite article after reading the item (which included a definition) and attempting a Hispanicized pronunciation of the term in question. Fourteen words represented the decidedly infrequent TPs (two in each case): /-t/, /-b/, /-p/, /-k/, /-h/, /-g/, and /-m/ and 6 words (two each ending in /-o/, /-a/, and /-1/) were employed as distractors. For all TPs except /-a/, masculine gender assignment ranged from 82.5 to 92.5%. In some instances, the attempted pronunciation (and written representations of the same) evidenced phonological adaptation/integration. While the informants' age and amount of contact with English (in terms of formal study and residence in the United States) were also taken into account, Clegg determined that these sociolinguistic factors did not play a role in the assignment of gender.

To summarize, then, it may be assumed that native speakers of Spanish are in general agreement regarding the gender of unfamiliar nouns (such as loanwords) which employ common TPs and/or have an underlying animate (human) referent. Furthermore, while gender vacillation and nonstandard gender assignment do occur in the Chicano community; the effects of these processes on loanword gender assignment by competent speakers of the variety are probably negligible. Finally, the research points to several other factors, most variable in nature, that appear to bias gender assignment. On the semantic side, the gender of the translation equivalent (synonymic gender) as well as the gender of the super-ordinate term (hyperonymic gender) have been posited as important variables. With regard to formal factors, both the gender assigned to the derivational suffix as well as the strong possibility that the masculine operates as the default (when the TP or other semantic factors underdetermine assignment) warrant further conside ration.

Data and Methodology

The data for this research project derive from two sources, Galvan's (1995) Diccionario del espanol chicano and Cobos's (1983) Dictionary of New Mexico and Southern Colorado Spanish. (5) English loanwords, glossed as gendered nouns in their respective dictionaries, were entered into a Gold Varb 2 token file along with a coding string consisting of the gender assigned to the term in question (this factor is designated as the dependent variable); the biological sex of the underlying referent (if applicable); the TP; the gender of a common synonym (if one exists); the gender of the respective hyperonym (again, if applicable); and the gender assigned to the derivational suffix (if one was employed). Simon and Schuster's International Spanish Dictionary and Santamaria (1983) were utilized to determine both synonymic and hyperonymic gender in some instances. The data were entered into two token files and processed separately since the Spanish of northern New Mexico and southern Colorado is recognized as a regional variety distinct from that of the rest of the Southwest (Cobos 1983, viii-xvi). This resulted in two corpora: Cobos yielded 212 tokens while Galvan yielded 595. The data were cross-tabulated, allowing one to isolate any knockout factors (those that result in categorical application or non-application) or singleton factors (factor groups consisting of a single factor) which cannot serve as input to variable rule analyses. Additionally, the cross-tabs provide the descriptive analyses (raw numbers and percentages) included in the subsequent tables. The binomial step- up/stepdown variable rule analysis was performed on the remaining factors (some TPs, synonymic as well as hyperonymic gender). This procedure generates probability coefficients (estimates of the strength of individual factors) and also determines which factor groups operate in a statistically significant fashion in assigning gender.

While the data extraction and encoding processes were generally straightforward and uneventful, several complications did arise. First, it was necessary to limit the tokens to only loanwords since both dictionaries included calques and hybrid forms as well and these were not always distinguished. (6) This was done to restrict the scope of the inquiry and because gender assignment for calques and hybrids was viewed as unproblematic. For instance, gender assignment for a hybrid like accento where the [ks] is thought to derive from the English 'accent' does not differ from the assignment for the standard Spanish acento (m.). Similiarly, gender assignment for the calque word yarda (which under influence from U.S. English has come to mean 'the property on which a house sits') has not been affected: it remains feminine even though its meaning is no longer restricted to 'unit of measurement.' Additionaly, other terms that demonstrated variable adaptation (e.g., guaifa and huifa 'wife') but employed the same TP were counted as one.

Second, a few potential tokens in the Galvan corpus were glossed with no indication of gender. Thus, terms like bomper 'bumper', croum 'chrome', cdpsul 'capsule', and jaiscul 'high school' where gender was omitted were simply discarded. (7)

Third, multiple entries were made for forms assigned more than one related meaning (bil 'statement, account, or dollar bill') and double-gendered forms [esquimo mf. 'eskimo').

It should be duly noted that the advantages of utilizing these corpora far outweigh the disadvantages. Being highly accessible, authentic materials that incorporate only attested, established forms, they are a rich source of raw data. Apart from the difficulties discussed above, the fact that no sociolinguistic factor can be adduced from the lexical entries places a significant limitation on this study. For example, it is quite likely that one's placement along the bilingual continuum could have repercussions on the gender assigned to a loanword like biurichap (mf.) 'beauty shop'. I hypothesize that a Spanish-dominant speaker would assign masculine gender to the term in question while an English-dominant speaker may vacillate (thus giving the appearance of a double-gendered noun). Obviously, further research that incorporates such variables will need to be conducted.

Results and Discussion

As was stated in the introduction, two factors categorically determine gender assignment for all loanwords within their purview. The first of these, biological sex of the underlying referent, is summarized in table 1.

What is most noteworthy is that a minority (approximately one-tenth to slightly more than a quarter) of the loanwords respond to biological sex. Indeed, many of the terms extracted from Galvan exhibit either common gender or consist of forms derived from a common base for both the masculine and feminine versions (cf. esquimo cited above as well as chirion/-a and ofendor/-a). This has the overall effect of inflating the descriptive statistics the for Galvan corpus. Also worthy of comment is the disparity evident in the number of masculine and feminine tokens. Much of this must be ascribed to traditional male and female roles; thus, while the dictionaries do not register buquipa, petrolina, or bos as feminine (with an underlying female referent), these are certainly possible and if uttered in the appropriate context would be understood as such. Other terms, however, make inherent reference to one sex or the other: granma, guifa (both f); boi escout, guachiman (both m.).

Finally, while the DEC exhibited no forms with epicene gender, Cobos (1983) registers three invariably masculine terms (regardless of the sex of the animal referent): bole 'baldy, a horse or cow with a white head', mafrico 'maverick, unbranded or stray animal', and torque 'turkey'.

The second factor, which is categorical in its application, the gender assigned to a derivational suffix, requires some explanation. Recall that Teschner's study of /-z/ appeals to morphological criteria and that Sanchez posits morphological adaptation as a significant variable for loanword gender assignment. Remember, too, that the generalizations based on the TPs /-n/ and /-s/ also incorporate morphological caveats: NORSEL excludes the derivational suffixes -cion and -sis/-tis, which correlate with the feminine (A-D-ION-IS).

In other words, I propose the following hierarchy: regardless of phonological or morphological ending, biological sex predominates as an indicator of gender for animates (except in the case of epicenes and other related instances): the terms virgen 'virgin', modelo '(female) model' (both f.), and pianista 'pianist' (mf.) illustrate this. The suffix -ista may take either gender; the TPs -n and -o, which usually correlate with the masculine, are superseded by the semantic criterion of biological sex. While it is true that formal criteria apply almost categorically in all other instances, the gender accorded the suffix takes precedence over the terminal phoneme. Consider the examples summarized in table 2, where the TPs do not determine gender, but the entire morphemic sequence does.

In other words, while most terminal phonemes are assigned to one gender or the other, there are, nonetheless, common exceptions. Even omitting from consideration all animates, we find counterexamples to the general tendencies. Thus, /-a/ is classed as a feminine marker, but dia (m.) 'day' is a common counterexample. Similarly, while /-o/ is considered a correlate of masculine gender, common exceptions include mano 'hand' and foto 'photo'. Nevertheless, if either of these TPs constitutes the terminal element of a derivational suffix, the assignment is categorical--there are no exceptions. Terminal morphemes, then, take precedence over terminal phonemes.

Three points must be made regarding table 3. First, note that only a small percentage of loanwords (only slightly more than 10%) respond to this criterion. This may explain why this factor has been slighted in the other studies reviewed herein. Second, as will be seen in the next section, final /-n/ and /-o/ for inanimates are categorically assigned masculine gender, regardless of whether or not they constitute the terminal element of a derivational suffix. The TP /-a/, however, is assigned feminine gender in a categorical fashion only when it is part of a terminal morpheme. Third, I hypothesize that certain terminal sequences that are not morphemic may be interpreted as such. Thus, the sequence -in (colcrin, oyodin) or -ina (brillantina) cannot be classified as diminutive suffixes but quite possibly are interpreted as such. Similarly, while -on is an augmentative in the case of trocon, it is not in the case of creyon or saxofon; nonetheless, it is probably construed the same in all three instances.

As expected, nearly all the remaining forms ending in NORSEL are assigned to the masculine. In fact, while gender assignment for four of the TPs is variable in at least one of the corpora, no masculine TP is assigned to the feminine in more than four instances.

As will be seen, these exceptions are accounted for by synonymic or hyperonymic gender assignment.

Regarding the feminine TPs, one does not appear at all (-is) and both -ion and -d appear only twice. Only -a appears with any frequency. Both loanwords that incorporate -d follow the tendency that Natalicio and Chaston noted: they are assigned to the masculine. In the case of -a, a number of the exceptions are unaccounted for. Particularly intriguing are those that derive from [-er] and are assigned masculine gender; it is entirely possible that the graphemic form (retained in memory) influences their assignment while the (r-colored) schwa is assigned phonologically to /a/. This argument receives support from the fact that many of the forms appear as doublets: bompa and bomper 'bumper', for example.

Table 6 reports on TPs that occur infrequently in standard Spanish. As such, they represent forms that are unadapted phonotactically. Only in the case of /-i/ (graphemically, {-I} or {-y}) and /-u/ could one expect to encounter a sufficient number of nouns in Standard Spanish on which to base a statistically significant generalization. Nonetheless, for all 11 less common TPs, Bull as well as Teschner and Russell report masculine assignment.

Several researchers such as Luke (1990) and Chaston (1996) have noted similar trends. Luke reports that the masculine is preferred for unadapted English nouns in code-switched discourse among Chicanos in New Mexico. Chaston also observed this tendency with his consultants in Texas. In his study of gender vacillation and nonstandard agreement, 40/42 switches occurred with a form of the masculine article. He states that the masculine tends to be assigned to English switches as well as to new cognate vocabulary.

Prado (1982) offers considerable evidence that the masculine gender is unmarked in Spanish and could function as the default option in cases in which other factors underdetermine gender assignment. These facts contradict Poplack, et al. (1982a: 21) who discount the unmarked tendency "[as being] of limited explanatory value" in loanword gender assignment. In light of the present findings, however, it does appear that the masculine is preferred if no specific factor biases assignment toward the feminine.

The paucity of less common TPs encountered in Cobos (only 2 of the 11 found in Galvan) may have to do with the fact that older, established loanwords also tend to be more integrated phonologically (particularly among a population with strong host language dominance). Also, the Galvan corpus may represent a more socially integrated and highly bilingual segment of society.

Tables 7 and 8 summarize what many researchers including Poplack, et al. (1982a) have considered to be a single factor: namely, analogical assignment. However, utilizing a distinction advanced by Bookless (1982), we recognize both synonymic and unique loan-words. Synonymic loanwords are those that compete for the same semantic space with a native language term. On the other hand, unique loanwords designate a new referent/concept and fill what traditionally have been called lexical gaps. In terms of gender assignment, synonymic loans may be assigned the gender of the translation equivalent or that of the superordinate category; unique loans have no strict native language equivalent but may derive gender from a hyperonym. Furthermore, as noted in the introduction, hyperonymic gender functions in standard Spanish: the term which determines the gender is frequently omitted, giving rise to nomenclature like elliptical gender; see Whitley (1986: 147, 166) and Butt and Benjamin (1995: 11, 15) for further discussion.

While the Cobos corpus was coded for synonymic gender, no clear-cut examples emerge from the analyses; other more robust factors (biological sex and morphological and phonological gender) are posited as determinants in this case. Cobos himself, in the introduction to his dictionary, provides a plausible explanation for this. He states:

The lack of a continuous, day-to-day contact with the people of Mexico throughout the seventeenth, eighteenth, and nineteenth centuries, together with a dearth of books, dictionaries and other reading material, stagnated and impoverished the Spanish of the colonists. Many times these people could not recall the name of a particular article and had no recourse but to coin a new term. (1983: xii-xiii)

In other words, if synonymic gender plays a role at all for New Mexican loanwords, it is entirely possible that regional vocabulary provides the translation equivalent. Since the corpora were coded utilizing sources that reflect general or Mexican usage, this factor may have gone undetected.

As table 7 attests, several forms in Galvan that derive from a model with [--er] have masculine translation equivalents. In some cases, this may be coincidental--some forms such as estara and mafla probably predate their respective equivalents arrancador and silenciador and would have already been assigned to the masculine before the appearance of the native language synonym. Nonetheless, there can be little doubt that this factor operates for the loanwords assigned feminine gender. On the basis of the TP alone, all the examples provided would normally take the masculine.

Both friser (from table 7) and jey fiver show evidence of dual assignment based on competing processes: phonological assignment provides impetus for the masculine while semantic criteria contradict this tendency. These two loans provide evidence that speakers may not always agree as to which factor takes precedence in the assignment process.

Both corpora contain loans that derive gender from the Spanish term galleta, which is considered hyperonymic since it may be glossed as 'cookie or cracker' (similar to the British biscuit). While both forms that derive from a model with [-er] may not respond exclusively to the factor examined here (see above), the feminine forms do provide incontrovertible evidence as to the operation of hyperonymic gender in the two corpora examined.

Tables 9 and 10 summarize the results of the GoldVarb 2 analyses performed on the data.

While the binomial step-up and step-down proceeded in a normal fashion for Cobos, one item must be noted with regard to Galvan. Convergence did not obtain after 20 iterations (the maximum allowed in this version). According to Rand and Sankoff (1990: 24 fn. 10), "in this case, [the] estimations may not be as accurate as desired, or there may be some non-uniqueness in the variable rule model defined by the condition file." Nonetheless, the following observations are in order. In the Cobos analysis, only three factors show any bias toward the feminine. Two of these, the TP /-a/ and gender marking which coincides with a feminine hyperonym, are expectedly robust. The other factor, /-s/ as a TP, unexpectedly demonstrates a strong bias toward the feminine. A similar pattern occurs for the Galvan file, in which the TP /-a/, gender marking which coincides with either a feminine synonym or hyperonym is again expectedly robust. The TP /-u/, however, like /-s/, demonstrates a similarly strong bias toward the feminine, which is quite unexpected. The coefficients for these two TPs are interpretable only by taking into account the extreme bias towards the masculine (as demonstrated by the respective input probabilities) and by positing that the other factors included in the analyses do not completely account for the feminine gender assignment for some of the Anglicisms with these TPs.


It has been demonstrated that gender assignment for Chicano loanwords responds principally to a set of five factors. If the underlying referent is human, biological sex always determines gender assignment. For inanimates, both morphological and phonological criteria determine or strongly bias gender assignment. In a handful of exceptions, other semantic criteria play a significant role. This study of loanword gender assignment in Spanish contributes to our understanding of the processes involved for at least three reasons. First, it is one of two to incorporate variable rule methodology, which allows competing factors to be examined simultaneously and subsequently ranked. Second, factors previously postulated but not empirically tested (morphological, synonymic, and hyperonymic gender) have been investigated and found to significantly influence gender assignment. Third, it is the only study of its kind to examine gender assignment among New Mexicans as well as those of Mexican ancestry in the southwestern Uni ted States. It is expected that further research in this area will shed even more light on the process of gender assignment to loanwords and in particular, on the questions raised by this study.


Source M Examples F Examples Totals

Cobos 21 buquipa 5 granma 26/212
 'bookeeper' 'grandma' (12.2%)
 charque/sharque guaina
 'shark, hustler' 'wino'
 petrol(ino) jaina
 'highway patrol' 'girlfriend'
Galvan 99 boi escout 58 beibisira 157/595
 'boy scout' 'babysitter' (26.3%)
 bos guaifa/huifa
 'boss' 'wife'
 drama < drummer esquimo
 'traveling salesman' 'eskimo' (f.)
 guachiman ofendora
 'night watchman' 'juvenile offender'
 saico quira
 'psycho, crazy' 'kiddo'


Ending Status Gender Assigned Counterexample

-n Phonemic Masc. imagen (f.) 'image'
-on (*) Phonemic Masc. razon (f.) 'reason'
-ion Phonemic Fem. (**) sarampion (m.) 'measles'
-cion Morphemic Fem. None

-e Phonemic Masc. calle (f.) 'street'
-mbre Phonemic Masc. hambre (f.) 'hunger'
-dumhre Morphemic Fem. None

-d Phonemic Fem. cesped (m.) 'grass, lawn'
-dad/-tad Morphemic Fem. None

(*)-on is also an augmentative; the inclusion of forms like cucharon
(m.) 'serving spoon' causes the masculine to predominate.

(**)On the basis of the inclusion of -cion as well as words like region
or religion (both f.) -ion is a correlate of feminine gender.


Source M Examples F Examples Totals

Cobos 12 boguecito 9 gasolinera 21/186
 'baby buggy' 'gas station' (11.2%)
 saxofon (*) lonchera
 'saxophone' 'lunch pail'
 oyodin (*) marbolina
 'iodine' 'marble'
 trocon requera
 'semitrailer' 'wrecker, tow truck'

Galvan 34 colcrin (*) 24 brillantina (*) 58/438
 'cold cream' 'hair oil' (13.2%)
 creyon (*) checadita
 'crayon' 'check-up'
 fonazo gasofa (**)
 'a blast, a gasoline
 lot of fun'
 picheo guachateria
 'pitching' 'washateria,

(*)Terminal sequence identical to a derivational suffix (may be
interpreted as such).

(**)Uncommon suffix; probably represents word play.


TP Cobos N/(%) Galvan N/(%) Bull % (*)

-n fon 10/10 balun 27/27 96.3
 'fun' 'balloon' (no -ion)
 ton tochdaun
 'ton' 'touchdown'

-o dipo 18/18 jaipo 32/32 99.7
 'depot' 'syringe, hypo'
 jando caucho
 'handout, money' 'couch, sofa'

-r estiquer 4/4 quemper 23/26 99.2
 'sticker' 'camper' (88.4)
 treilar escurer
 '(semi)trailer' 'scooter'

-s cabus 5/7 fius 15/15 92.7
 'caboose' (71.4) 'fuse' (no -is)
 velis rices
 'valise' 'recess'

-e bonque 76/77 diche 91/95 89.2
 'bunk (bed)' (98.7) 'ditch' (95.7)
 dompe sinque
 'dump' 'sink'

-l guinchil 11/11 guafol 23/24 96.6
 'windshield' 'waffle' (95.8)
 sacarol suiminpul
 'sucker rod' 'swimming pool'

TP T&R % (**)

-n 48.3
 (includes -ion)

-o 99.8

-r 98.5

-s 57.3
 (includes -is)

-e 89.3

-l 97.8

(*)Percent masculine from Bull (1965)

(**)Percent masculine from Teschner and Russell (1984)


TP Cobos N/(%) Galvan N/(%) Bull%(*) T&R% (**)

-a baica 39/42 laira 69/78 98.9 96.3
 'bike' (92.8) 'lighter' (88.4)
 cuilta boila
 'quilt' 'boiler'
 ploga dona
 'plug' 'doughnut'
 cuara (m.) (+) chahua (m.) (+)
 quarter, 'shower;
 twenty-five bridal or
 cents' baby shower'

-d raid (m). 0/1 raund (m.) 0/2 97.0 97.5
 'ride' 'round'

Note: Only two terms ended in -ion (both assigned feminine gender),
complexion (Cobos); taxacion (Galvan). No term ended in -is, except for
an occasional plural: fonis 'funnies, comics'.

(*)Percent feminine from Bull (1965)

(**)Percent feminine from Teschner and Russell (1984)

(+)Several items in both corpora where the final /-a/ derives from {-er}
([[partial]] or [[partial]]) are assigned masculine gender


TP Cobos N/(%) Galvan N/(%)

-b No examples clab/clob 1/1

-k No examples bloc 12/14
{c} 'city block' (85.7)

t [integral] No examples blich 7/7
{ch} 'bleach'

-f No examples blaf 1/1

-g No examples ring 1/1

-i chiribi 4/4 ciodi 30/31
 'shivaree' 'C.O.D' (96.7)
 guini jotquey
 'wienie' 'hotcake'
 peni (mf.)
 'penny; penitentiary'

-u champu 1/1 chou 2/4
 'shampoo' 'show' (50)
 glu (f.)
 flu (mf.)

-m No examples coicrim 4/4
 'cold cream'
 'chewing gum

-p No examples teip 6/7
 'tape' (85.7)

 'pick-up tr'
 biurichap (mf.)
 'beauty shop'

-t No examples bloaut 19/19
 'flat tire'
 'jet (plane'

-ks No examples clorox 4/4
{x} 'bleach'
 'income tax'

TP Bull N/(%) (*) T & R N/(%) (+)

-b 6/6 12/12

-k 18/19 23/24
{c} (94.7) (95.8)

t [integral] 3/3 3/3

-f 4/5 4/4

-g 4/4 3/3

-i 222/236 398/427
 (94) {-i/-y} (93.2) {-i/-y}

-u 56/58 97/102
 (96.5) (95.1)

-m 32/33 30/30

-p 4/4 5/5

-t 35/37 39/42
 (94.5) (92.8)

-ks 29/29 30/33
{x} (90.9)

(*)Percent masculine from Bull (1965) (including number of tokens)

(+)Percent Masculine from Teschner and Russell (1984) (including number
of tokens)


Synonym (m.) Examples Synonym (f.) Examples

calenton jira congeladora friser (mf.)
 'heater' 'freezer'

arrancador estara bicicleta baic
 'starter motor' 'bike'

silenciador mofla coca cola couc
 'muffler' 'coke, coca cola'

jardin de ninos quinda leche milque
 'kindergarten' 'milk'


Hyperonym Cobos Hyperonym Galvan

galleta craques (f. pl.) galleta cuque (f.)
'cracker' 'soda crackers' 'cookie' 'cookie'

polvo bequenpaura (m.) pelota sofbol (f.)
'powder' 'baking powder' 'ball' 'softball'

camion treila (m.) alergias/fiebre jey fiver (mf.)
'truck' 'semi(trailer)' 'allergies/fever' 'hay fever'


Factor ProbC (*)

Input (**) .912
Terminal Phoneme: /a/ .006
Terminal Phoneme: /e/ .939
Terminal Phoneme: /s/ .222
Hyperonymic Gen: masc. .812
Hyperonymic Gen: fem. .159
p = .016

(*)Probability coefficient (above .500 progressively favors the
masculine; below .500 progressively favors the feminine).

(**)Input probability reflects the general tendency for the masculine to
occur in spite of or unaccounted for by the factors specified.


Factor ProbC (*)

Input (**) .870
Terminal Phoneme: /a/ .022
Terminal Phoneme: /c/ .537
Terminal Phoneme: /e/ .848
Terminal Phoneme: /i/ .905
Terminal Phoneme: /l/ .858
Terminal Phoneme: /p/ .593
Terminal Phoneme: /r/ .683
Terminal Phoneme: /u/ .201
Synonymic Gen.: masc. .666
Synonymic Gen.: fem. .293
Hyperonymic Gen: masc. .673
Hyperonymic Gen: fem. .254
p = .018

(*)Probability coefficient (above .500 progressively favors the
masculine; below .500 progressively favors the feminine).

(**)Input probability reflects the general tendency for the masculine to
occur in spite of or unaccounted for by the factors specified.

(*.) I acknowledge the tireless and invaluable help of Jens Clegg, my research assistant, without whom this project would not have been completed. I also thank the Spanish and Portuguese Department and College of Humanities for their financial support.


(1.) The following two sections rely heavily (but not exclusively) on Butt and Benjamin (1994: 1-16).

(2.) Lexical innovation due to linguistic contact involves both a model from the donor language (English, in this instance) and a replica in the host language (Spanish). Since nouns are not assigned grammatical gender in English, gender for English loanwords in Spanish cannot derive from the gender of the English model. Similarly, since Spanish nouns are not declined, morphological class cannot serve as a basis for assigning gender.

(3.) Zamora Munne and Bejar (1987) further this line of research and examine gender assignment to French loanwords in Spanish where both phonological shape and gender of the model are posited as principal determinants.

(4.) Poplack et al. (1982b) is the Spanish language version of the article.

(5.) The preliminary version of this article, presented at the Sixteenth Conference on Spanish in the United States and held in conjunction with the Seventh University of New Mexico Conference on Ibero-American Culture and Society: Spanish and Portuguese in Contact with Other Languages on 12-14 February 1997 in Albuquerque was based only on data from Galvan (1995). Pursuant to the suggestion of one of the conference participants we incorporated Cobos (1983) as representative of the New Mexican and Southern Colorado region. It was not possible, however, to incorporate the English loanwords from Espinosa (1914: 304-11) because he did not indicate gender on his word list.

(6.) For a more in-depth treatment of the general nature of calques and loanwords in Chicano Spanish, see Smead and Clegg (1996) and Smead (1998).

(7.) Capsul would have provided an interesting case in point due to the change in TP: The standard form is capsula (f.) while the form cited in the DEC could be assigned masculine gender. In the absence of any clear indication, however, the form was discarded.


Arias Barredo, Anibal. 1990. Genero gramatical y motivacion semantica. Estudios de Linguistica 6: 107-21.

Banfield, Robert L. 1994. Gender assignment of English loanwords in the Spanish language. Master's thesis, Brigham Young University.

Barkin, Florence. 1980. The role of loanword assimilation in gender assignment. Bilingual Review/Revista Bilingue 7, no. 2: 105-12.

Bergen, John J. 1978. A simplified approach for teaching the gender of Spanish nouns. Hispania 61: 865-76.

_____. 1980. The semantics of gender contrasts in Spanish. Hispania 63: 48-57.

Bookless, T. C. 1982. Towards a semantic description of English loan-words in Spanish. Quinquereme 5, no. 2: 170-85.

Bull, William. 1965. Spanish for teachers: Applied linguistics. New York: John Wiley & Sons.

Butt, John, and Carmen Benjamin. 1994. A new reference grammar of modern Spanish. Lincolnwood, IL: NTC.

Chastain, Kenneth. n.d. Spanish and English as communication systems. Unpublished ms.

Chaston, John M. 1996. Sociolinguistic analysis of gender agreement in article/noun combinations in Mexican American Spanish in Texas. Bilingual Review/Revista Bilingue 21, no. 3: 195-202.

Clegg, Jens. 1997. Native Spanish speaker intuition in gender assignment to English loanwords. Unpublished ms.

Cobos, Ruben. 1983. A dictionary of New Mexican and Southern Colorado Spanish. Santa Fe: Museum of New Mexico Press.

Corbett, Greville. 1991. Gender. New York: Cambridge University.

De Gamez, Tana, ed. n.d. Simon and Schuster's international Spanish dictionary. New York: Prentice Hall.

Espinosa, Aurelio M. 1914. Studies in New Mexican Spanish. Part III: The English elements. Revue de Dialectologie Romane 6: 241-317.

Galvan, Roberto A. 1995. The dictionary of Chicano Spanish/El diccionario del espanol chicano. 2d ed. Lincolnwood, IL: NTC.

Garcia, a, Maryellen. 1998. Gender marking in a dialect of Southwest Spanish. Southwest Journal of Linguistics 17, no. 1: 49-58.

Ibrahim, Muhammad Hassan. 1973. Grammatical gender: Its origin and development. Janua Linguaram, no. 166. The Hague: Mouton.

Luke, Chad. 1990. Gender agreement in Spanish-English language mixing: A preliminary study. Unpublished ms.

Natalicio, Diana. 1983. Native speaker intuitions as a basis for determining noun gender rules in Spanish. Southwest Journal of Linguistics 6, no. 1: 49-55.

Poplack, Shana, Alicia Pousada, and David Sankoff. 1982a. Competing influences on gender assignment: Variable processes, stable outcome. Lingua 57: 1-28.

Poplack, Shana, Alicia Pousada, and David Sankoff. 1982b. Un estudio comparativo de la asignacion de genero a prestamos nominales. In El espanol del Caribe: Ponencias del VI Simposio de Dialectologia, edited by Orlando Alba, 237-69. Santiago, Republica Dominicana: Universidad Catolica Madre y Maestra.

Prado, Marcial. 1982. El genero en espanol y la teoria de la marcadez. Hispania 65: 258-66.

Rand, David, and David Sankoff. 1990. GoldVarb 2.1. Montreal: Centre de recherches mathematiques, Universite de Montreal.

Sanchez, Maria F. 1995. Clasificacion y analisis de prestamos del ingles en la prensa de Espana y Mexico. Lewiston, NY: Edwin Mellen.

Santamaria, Francisco J. 1983. Diccionario de mejicanismos, 4th ed., revised and expanded. Mexico, DF: Editorial Porrua.

Smead, Robert N. 1998. English loanwords in Chicano Spanish: Characterization and rationale. Bilingual Review/Revista Bilingue 23, no. 2: 113-23.

Smead, Robert N., and J. Halvor Clegg. 1996. English calques in Chicano Spanish. In Spanish in contact: issues in bilingualism, edited by Ana Roca and John B. Jensen. Somerville, MA: Cascadilla.

Teschner, Richard V. 1983. Spanish gender revisited: -Z words as illustrating the need for expanded phonological and morphological analysis. Hispania 66: 252-55.

_____. 1987. Noun gender categories in Spanish and French: Form-based analyses and comparisons. In Current Trends and Issues in Hispanic Linguistics, edited by Lenard Studerus, 81-109. El Paso: University of Texas at El Paso.

Teschner, Richard V., and William M. Russell. 1984. The gender patterns of Spanish nouns: An inverse dictionary-based analysis. Hispanic Linguistics 1: 115-32.

Teschner, Richard V, and Yiyuk Estela Alatorre. 1984. The gender of Spanish nouns in -s: A description refined. Hispania 67: 409-11.

Wagner, Claudio. 1990. El enfoque linguistico de la normativa: el caso de los prestamos. Estudios Filologicos 25: 55-65.

Whitley, M. Stanley. 1986. Spanish/English contrasts: A course in Spanish linguistics. Washington, DC: Georgetown University Press.

Zamora, Juan Clemente. 1975. Morfologia bilingue: la asignacion de genero a los prestamos. Bilingual Review/Revista bilingue 2.239-47.

Zamora Munne, Juan C., and Eduardo C. Bejar 1987. El genero de los prestamos. Revista Espanola de Linguistica 17, no. 1: 131-37.
COPYRIGHT 2000 Bilingual Review Press
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2000 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Smead, Robert N.
Publication:Bilingual Review
Article Type:Statistical Data Included
Geographic Code:1USA
Date:Sep 1, 2000
Previous Article:Alternative education: The dark side of education or new tools for Latinos to achieve self-actualization.
Next Article:A conversation with Rolando Hinojosa.

Related Articles
Chicano Narrative: The Dialectics of Difference.
Voicing Chicana Feminisms: Young Women Speak Out on Sexuality and Identity.
OCC study: no evidence of age/gender discrimination.
The Chicano treatise.
Anglicisms in German; borrowing, lexical productivity, and written codeswitching.
Gender and national statistics.
UAE launches survey to compile comprehensive employment data.
Executive summary.
Gender differences in perception of effectiveness of using statistical software in learning statistics.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters