Noun Classifier Extension in Q ’ anjob ’ al ( Mayan ) : Acquiring a Gender Stereotype 1

Linguists have not fully explored the correlation between sociolinguistic variation and pragmatics of nominal classifier systems. When pragmatics is discussed, the focus is usually on anaphora under a restricted conceptualization of “discourse” that closely resembles the co-text of an utterance, that is, the surrounding text context (e.g., Craig 1994; Aikhenvald 1994; Sands 1995; Zavala 2000). Zavala (2000:139) discusses at length the “discourse-pragmatic” properties and functions of classifiers in Akatek Mayan which control “[t]he presence or absence of noun classifiers.” Under this framework, “noun classifiers are used to mark third-person nominals as individuated, referential and thematically important items in discourse” (Zavala 2000:140). However, this notion of “discourse-pragmatic” is quite constrained, in that “discourse” is generally limited to the immediately surrounding utterances (i.e., the co-text) and “pragmatic” largely operates under the sense of other-than-syntactic, such as overt use of a noun classifier in early discourse for purposes of foregrounding that changes to zero anaphora in subsequent utterances. Q‟anjob‟al is a Mayan language principally spoken in the department of Huehuetenango in Guatemala (Mateo Pedro 2004) and, along with Akatek, Chuj, and Jakaltek, is part of the Q‟anjob‟alan branch (Campbell 1997:163; Robertson 1992:3). Within the last decades, some speakers of Q‟anjob‟al have been employing noun classifiers in a manner that extends them and is coupled with a particular intentional use (Mateo Pedro 2004). This use exhibits characteristics that suggest a broadening the aforementioned concepts of “discourse” and “pragmatics” is necessary in order to capture classifier function and use as a gender-based variation. This paper investigates noun classifier use in one dialect of Q‟anjob‟al and argues that recent sociolinguistic change of canonical classifier use by adult male speakers is principally pragmatic and ideological, discursively (re)producing social constructions of gender. This pragmatic and ideological usage by men informs child overextension of noun classifiers, though the latter does not exhibit entirely similar pragmatic and ideological function. The Santa Eulalia dialect of Q‟anjob‟al (Q‟SE) is of particular interest with regard to the phenomenon of gender-based noun classifier extension. Speakers of the Q‟SE dialect seem to have triggered this “innovation,” though it is also current among speakers of Q‟anjob‟al of Soloma because of contact and diffusion (Mateo Pedro 2004:1).


Introduction
Linguists have not fully explored the correlation between sociolinguistic variation and pragmatics of nominal classifier systems.When pragmatics is discussed, the focus is usually on anaphora under a restricted conceptualization of "discourse" that closely resembles the co-text of an utterance, that is, the surrounding text context (e.g., Craig 1994;Aikhenvald 1994;Sands 1995;Zavala 2000).Zavala (2000:139) discusses at length the "discourse-pragmatic" properties and functions of classifiers in Akatek Mayan which control "[t]he presence or absence of noun classifiers."Under this framework, "noun classifiers are used to mark third-person nominals as individuated, referential and thematically important items in discourse" (Zavala 2000:140).However, this notion of "discourse-pragmatic" is quite constrained, in that "discourse" is generally limited to the immediately surrounding utterances (i.e., the co-text) and "pragmatic" largely operates under the sense of other-than-syntactic, such as overt use of a noun classifier in early discourse for purposes of foregrounding that changes to zero anaphora in subsequent utterances.
Q"anjob"al is a Mayan language principally spoken in the department of Huehuetenango in Guatemala (Mateo Pedro 2004) and, along with Akatek, Chuj, and Jakaltek, is part of the Q"anjob"alan branch (Campbell 1997:163;Robertson 1992:3).Within the last decades, some speakers of Q"anjob"al have been employing noun classifiers in a manner that extends them and is coupled with a particular intentional use (Mateo Pedro 2004).This use exhibits characteristics that suggest a broadening the aforementioned concepts of "discourse" and "pragmatics" is necessary in order to capture classifier function and use as a gender-based variation.This paper investigates noun classifier use in one dialect of Q"anjob"al and argues that recent sociolinguistic change of canonical classifier use by adult male speakers is principally pragmatic and ideological, discursively (re)producing social constructions of gender.This pragmatic and ideological usage by men informs child overextension of noun classifiers, though the latter does not exhibit entirely similar pragmatic and ideological function.
The Santa Eulalia dialect of Q"anjob"al (Q"SE) is of particular interest with regard to the phenomenon of gender-based noun classifier extension.Speakers of the Q"SE dialect seem to have triggered this "innovation," though it is also current among speakers of Q"anjob"al of Soloma because of contact and diffusion (Mateo Pedro 2004:1).
Q"anjob"al possesses a recently developed system of noun classification that is the result of grammaticalization (Mateo Pedro 2004:2;Zavala 1992).Noun classifiers in Q"SE are clitics that are historically of nominal origin, leading to semantic transparency between noun and noun classified.Q"SE noun classifiers typically function as adnominals within the scope of a noun phrase or as pronominal elements for reference tracking and management in discourse.For example, the human classifiers ix and naq co-occur with proper names, as in ix Dominga "Dominga" and naq Matin "Martin," or with nouns whose referent is either female or male, as in ix unin "the (female) child" and naq winaq "the (male) man."These examples demonstrate the standard, canonical use of the human noun classifiers.In such cases, the semantics of the classifier (e.g., ix [+human, +female]) corresponds to the semantics of the noun classified (e.g., Dominga [+human, +female]).
However, within the last fifty years, some male speakers of Q"SE have extended the use of the female classifier ix to contexts for the male classifier naq, thus departing from the standard, canonical use (Mateo Pedro 2004).Adult male constructions of this type are ix Matin and ix winaq, which both involve the co-occurrence of the female classifier ix a nominal that is [+male].Such constructions are most often contextually employed exclusively in the presence of male interlocutors to convey a variety of pragmatic meanings, including spatial deixis, indefiniteness or unfamiliarity, relational distance, referring jokingly from relational closeness, and referring derogatorily.In what follows, I propose that the principal impetus for this change from canonical to non-canonical usage is ideological, based on social constructions of gender.The negative reaction that Mateo Pedro (2004:5-6) notes among female Q"SE speakers to such manipulated use attests to this ideological component, which is pragmatic.At least some young female Q"anjob"al speakers in Guatemala claim that the exploiting of ix by male speakers in reference to other males implicates and predicates gender stereotypes towards women (Mateo Pedro 2004:6).
Moreover, this pragmatic use among male speakers also has motivated classifier extension that is currently prevalent among child language learners of Q"SE.Accordingly, this paper focuses not only on adult gender-based variation, but developmental properties of gender stereotypes among children, as well.Child noun classifier extension exhibits greater opacity and less analyticity by transferring the extensional properties to a broader group of classifiers within the noun classifier paradigm.The extensional use of ix among children is polysemous in this respect.Male child use patterns after the salient adult property of "male only," but without an animate/inanimate distinction among the nouns that children classify with the female classifier ix.Thus, the child data demonstrates that ix is being extended not only to naq, but also to ch'en, the stone/metal classifier and no('), the animal classifier.That is, children use what is canonically the female classifier in order to classify nouns that belong to distinct categories, which results in an overextension of its application and use.Pragmatic and ideological influences, then, such as gender-based intentional uses, appear to be a driving force influencing variation and the current development of noun classifier use in Q"SE.
In the discussion that follows, I first treat the semantics of Q"SE noun classifiers according to canonical use (Section 2).In Section 3, I turn to the notions of noun classifier extension and overextension.Based on Mateo Pedro"s (2004) original discussion of noun classifier extension in Q"anjob"al, I provide a somewhat formal apparatus for conceptualizing and taxonomizing these phenomena.This provides a conceptual framework for investigating the properties of adult noun classifier extension and child noun classifier overextension in Sections 3.1 and 3.2.Following this analysis, in Section 4 I conclude by summarizing the differences between adult and child (over)extensional usage, as well as discussing implications for future research and potential trajectories for further development of the Q"SE system of nominal classification.
The Q"anjob"al Santa Eulalia (Q"SE) noun classifier paradigm contains 13 noun classifiers.They function either adnominally, preceding the noun they classify, or pronominally, for purposes of reference tracking and management in discourse.Noun classifiers in Q"anjob"al are clitics that derive from nominals (Zavala 1992).For example, the female human noun classifier ix derives from the word ix "woman," and male human classifier naq derives from the word winaq "man." The 13 Q"SE classifiers, along with their nominal origin and examples of each according to canonical usage are provided in Table 1 below (González et al. 2000 tx'anej Fiber/rope tx'an tx'anej "the rope" q'a(q)' q'aq'ej Fire q'a q'aq'ej "the fire" tz'am atz'am Salt tz'am atz'am "the salt" As clitics, these noun classifiers are largely phonologically reduced forms of a nominal within the current lexicon, though some are "repeaters," or noun classifiers that are homophonous ("identical") with the nouns they are used to classify (Grinevald 2004(Grinevald :1026)).The relation between classifier and noun origin, and resultant taxonomies, leads to a fair amount of semantic transparency.Speakers use noun classifiers to select nominals with particular salient or perceived features that are readily accessible.The use of a particular classifier with a corresponding noun typically encodes some perceptually or ontologically inherent characteristic of the noun classified (Allan 1977:285;Zavala 2000:143).So, while semantics are considered a contingent property for nominal classification (Aikhenvald 2000a:81) noun classifiers in Q"anjob"al currently rely heavily on semantics in the selection of which classifiers are employed to modify a particular noun.
The semantic relation between a noun classifier and the noun used to refer is frequently transparent, such as the use of the Q"SE noun classifier ix with a female proper name, which encodes some semantic feature roughly corresponding to [+female, +human].When, according to standard use, a speaker employs ix, the femininity and humanity of the referent are necessarily in focus (Aikhenvald 2000b:100).However, in some languages cross-linguistically, the same noun classifiers can be used in conjunction with differing nouns in order to specify meaning (Aikhenvald 2000a:84).In the Minangkabau (Austronesian) language, a noun meaning "lemon" can be used with either the noun classifier for tree or for fruit (Aikhenvald 2000a:84).In the language Yidiny (Australian), speakers classify a particular tree species, diwiy, according to the tree classifier jugi or the drinkable classifier bana to refer to the water that can be tapped from its bark (Aikhenvald 2000a:84).Nevertheless, this type of variable use between noun classifiers and the nouns they classify does not correspond to the extensional use observed in Mayan Q"SE case, which we will see is unique in that human noun classifiers are not typically interchanged to signify a particular, separate meaning.Rather, they are ordinarily strictly monosemous and (again, ordinarily) semantically fixed or frozen.Such use leads to the creation of fixed and conceptually salient semantic boundaries between individual noun classifiers, as well as the taxonomies that each classifier creates according to canonical use.
Figure 1 below illustrates the semantics of noun classifiers in Q"SE, with specific attention given to human noun classifiers, as they are of particular import to the present topic: As the above figure illustrates, human noun classification for Q"SE is rigidly sex-based.In other words, biological sex is always in focus as the salient property when speakers use a noun classifier to select for a particular human noun; there is no general noun classifier for human where sex is not in focus.Because of this specialized focus, potential manipulation of human noun classifiers, which ultimately transgresses the well-defined boundaries that ix and naq typically operate within, creates a situation that makes the feature of gender audible interaction (Sunderland & Litosseliti 2008:14).Thus, the extension of ix to naq constitutes warrant for the relevance of gender as a variable for detailed consideration.

Noun Classifier Extension and Overextension
Under standard, canonical use, human noun classifiers operate within a fixed semantic realm, governed by the semantic boundaries rooted in the salient or perceived characteristics of the referent.Noun classifier extension in Q"anjob"al Santa Eulalia (Q"SE) violates the typical semantic boundaries, which in the case of ix and naq, is sex-based agreement between noun and noun classified.Figure 2  The basic notion behind the concept of extension is that a set of nominals classified by a given element becomes expanded to include nominals from a separate set.So, if a classifier, x, fixes adnominally to any element in the set [x 1 , x 2 , x 3 , …, x n ], and a second classifier, y, likewise co-occurs with any element in the set [y 1 , y 2 , y 3 , …, y n ] then we may say that x extends to y if the set of nouns which x classifies comes to include both [x 1 , x 2 , x 3 , …, x n ] and [y 1 , y 2 , y 3 , …, y n ].The result is a novel, nonstandard taxonomy for x.The following diagram expresses this type of extension, where x extends to y: Within this description, a specific semantic relation holds between noun classifier and noun classified, as expressed by, for example, x:x 1 or y:y 3 .However, extension creates a separate, novel relationship, such as the one between noun classifier and noun classified in x:y 2 .The extension of x to y does not dissolve the use of y in application to its original set, nor the relation between y and the nominals with which it co-occurs.Under extension, the set, or a portion of a set, typically belonging to a particular noun classifier becomes subordinated to another classifier for intentional purposes, thus altering the "fixed" boundaries between classifiers as stipulated by standard, canonical use.Similarly, noun classifier overextension expands the set of nominals over which a particular classifier operates through co-ocurrence with or anaphoric reference to a given nominal element.The difference stems from the observation that overextension builds upon extensional properties that are previously established, applying it in a more broad and systematic manner.Using again the above framework and semantic map in Figure 3, overextension involves the subordination of one or more sets to the extensional values already in use.So, if x becomes overextended, it would classify the set of nominals in both [x 1 , x 2 , x 3 , …, x n ] and [y 1 , y 2 , y 3 , …, y n ] as well as one or more additional sets, such as [z 1 , z 2 , z 3 , …, z n ].In the discussion and data below, we will see that extension associates with speaker intention, as a way of manipulating the semantic boundaries between two noun classifiers.This use characterizes the quality of adult speakers, and retains a relative amount of transparency.On the other hand, overextension is characteristic of child use and is associated with systematicity and opacity, and is polysemous.

Adult Noun Classifier Extension in Q'anjob'al Santa Eulalia
To understand the above conceptualization of extension, we will turn to the recent phenomena and discuss specific examples of noun classifier extension in Q"anjob"al Santa Eulalia (Q"SE).When noun classifiers co-occur with nouns as adnominals, canonical noun classifier use (see Table 1 above) of the human classifiers in Q"SE involves a correspondence of semantic features: the female classifer ix co-occurs with nouns that are [+human, +female], as in ix Dominga "Dominga," while the male classifier naq co-occurs with nouns that are [+human, +male], as in naq Matin "Martin."However, when the classifier ix is extended to naq, ix co-occurs with nouns that are [+human, +male], as in ix Matin "(female) Martin," and such usage is non-canonical.Among adult speakers of Q"SE, noun classifier extension is not only relegated to the human domain, it is also unidirectional: male speakers do not extend naq to ix, as in *naq Dominga "(male) Dominga." Noun classifier extension in Q"SE corresponds to diverse pragmatic and ideological functions.The following list summarizes the various usage features ix to naq extension by adult males (Mateo Pedro 2004:5-6): (1) Properties of adult ix to naq extension a. employed solely by male speakers b. used (almost) exclusively in the presence of male interlocutors c. used deictically in reference to other males not physically or immediately present at the time of utterance d. used deictically in reference to other males who are relationally distant to the speaker e. used to communicate indefiniteness or unfamiliarity f. used to refer to other males jokingly (indicating relational closeness) or derogatorily g. based on female reaction, (potentially) implies a negative attitude toward females According to Mateo Pedro (2004), noun classifier extension from ix to naq began in Q"SE by adult male speakers in Jolom Konob" approximately 50 years ago.While he does mention that "there is no documentation that can prove how and when the extension of ix to naq started," he refers to personal exposure to this type of usage as well as confirming its prevalence through conversations with other adult male native speakers of Q"SE (Mateo Pedro 2004:5-6).Mateo Pedro observes that the extension of ix to naq occurs primarily among male speakers in the exclusive presence of male interlocutors (Mateo Pedro 2004:5).The presence of a female in a conversational setting typically blocks such usage (Mateo Pedro 2009, personal communication).However, Mateo Pedro also notes that some young Q"anjob"alan women in Guatemala possess an awareness of this type of extension as well as a subsequent negative reaction towards its employment, despite its restricted use.He writes: There is a reaction of young female speakers of Q"anjob"al toward this extension.They argue that young male Q"anjob"al speakers using ix instead of naq are trying to imply that women are unskilled at doing activities that men do.Perhaps men use ix to classify a male who is not skilled, so they compare him with a wom[a]n (Mateo Pedro 2004:6).
This disapprobation towards male extensional use implies that female speakers perceive and interpret extension to include ideological attributes regarding socially constructed notions of gender, such as ascribing impoverished skill and performance for women.This additionally concords with a further observation that Mateo Pedro notices when, as a result of diffusion via contact, male speakers from Soloma used ix to naq extension in order to make jokes about other men (Mateo Pedro 2004:6).Furthermore, ix to naq extension typically occurs when the referent is either spatially or relationally distant, or for communicating some sort of indefiniteness or, similar to the "joke" usage, derogatorily (Mateo Pedro 2009, personal communication).In terms of relational distance, Mateo Pedro also indicates that speakers use ix in reference to men to either imply unfamiliarity or signify lack of physical relationship.Unlike the systematized multiple applications of specific classifiers in Minangkabau and Yidiny from Section 2 above, noun classifier extension in Q"SE transgresses the semantic boundaries between classifiers for intentional purposes.The utterances produced should, according to canonical use, result in ungrammaticality.Nevertheless, adult male extension of ix to naq results in the subordination of nominals classified by naq to the female noun classifier ix, dissolving the prior fixity of semantic boundary between female and male nouns, and creating a novel taxonomy of extensional use: Returning to the utility of broader notions of "pragmatics" and "discourse," these features clearly illustrate that the female noun classifier ix when used by males to refer to and individuate other males transcends discourse-pragmatic (in the sense of Zavala 2000:139, 143) conceptualizations of reference tracking and management.Moreover, context that can capture each of the above features, whether exemplified in isolation or some combination, becomes crucially important.Therefore, in order to contextualize a particular utterance that includes noun classifier extension, we must consider the gender and age of the speaker, the gender and age of the speaker"s interlocutors, and the surrounding co-text in which an utterance is situated, as well as other potentially relevant variables.
One critical aspect to consider is that the above discussion derives from observations and personal conversations (e.g., between Mateo Pedro and me or between Mateo Pedro and native speakers of Q"SE) that lack empirical exemplification.The single specific example that Mateo Pedro (2004) provides in regards to adult male classifier extension is an introspective memory of his father being referred to as ix Matin by his father"s friends, instead of the expected, typical construction of naq Matin or cham Matin (Mateo Pedro 2004:6).The degree of uncertainty in the treatment of adult male usage and its features demands further research, such as qualitative interviews that elicit use and native speaker interpretation, participant observation, or ethnographic research.

Child Noun Classifier Overextension
Nevertheless, Mateo Pedro (2004) does provide specific exemplification of child noun classifier extension.However, child data differs markedly from the description of adult use in that ix is not limited in its extension to naq by children.Instead, ix is overextended to nouns that canonically co-occur with a variety of other classifiers.Mateo Pedro (2004:8) provides examples that demonstrates child overextension not only involves extension of ix to naq, as in ix ronal "Ronald," but also ix to ch'en, the stone/metal classifier, in ix karo "the car" and ix to no('), the animal classifier, in ix mistun "the cat."This may be because children pick up on certain pragmatic properties of noun classifier extension, such as exclusive extensional use among males and the fact that ix is the only noun classifier extended to other categories.Additionally, children may not be noticing or acquiring the ideological properties of adult extensional use, which could allow them to overextend ix to a variety of nominals that are outside of the human domain.
So, child overextension not only transgresses the semantic boundaries between female/male nouns, but human/nonhuman, and animate/inanimate nouns, as well.Figure 5 shows the taxonomy created by child overextension of ix, and it illustrates that ix in child overextensional use functions as a somewhat all-purpose noun classifier that collapses canonical semantic categories.Adult noun classifier extension collapses the female/male distinction (see Figure 4), but child overextension collapses multiple distinct semantic categories: Concerning context, use is still limited to male speakers in the presence of interlocutors (Mateo Pedro 2004:7).The current significant differences, then, between adult extension and child overextension are twofold.The first difference is a quantitative one, since children extend the female classifier ix to nouns that lack the semantic features [+male, +human], while adults maintain a strict usage.The second difference is a qualitative one, since the child usage appears to lack an ideological impetus, despite its causal relation to the adult use.That is, child overextension is not limited to the semantic domain of human, which suggests that social constructions of gender are not motivating child overextensional use.Child data instead exemplifies a systematic characteristic wherein a variety of noun classifiers, not just naq, are subordinated to ix through overextension.Additional child acquisition data reinforces the initial observations from Mateo Pedro (2004), with the exception that exclusive male presence may not be a necessary controlling factor in child use.The following discourse in (2) was taken from a conversation between a male child (Speaker1) and his grandmother (Speaker2); Speaker3 was not identified in the transcript.Following Zavala (2000), portions not relevant to the analysis are provided in square brackets without glossing.Additionally, the symbol "" to the left of an utterance signals an instance of overextension: (2) Example discourse: "the female man"  a.In (2a-u) above, the multiple instances of overextension lead to greater opacity, and less analyticity, as well as appearing to be polysemous and possibly indicating semantic bleaching.These attributes stem from the dual overextension of ix to both ch'en (1a) and naq (1c, f, g, i, j, l, p, r, t).Consequently, given the simultaneous occurrence of overextension to two types of noun classifiers, the ideological component, which is characteristic of adult use, becomes invalid for child instantiation that is based on systematic subordination of multiple elements to a single noun classifier.
Another significant example of child noun classifier overextension involves the process of repair: (3) ix naq naq pre CL:woman CL:man CL:man enrique "(woman, woman, man) Enrique" (4) ix naq pap CL:woman CL:man father "(woman, man) father" In both (3) and ( 4), the speaker attempts to individuate a male referent, but he does so initially by attaching the female noun classifier then switching to canonical use.The co-occurrence of two different noun classifiers with a single noun signals that some form of repair is taking place.However, the most telling feature of these two utterances relates to the discourse context.As in (2a-u), the child"s interlocutor is his grandmother, but the instances of repair take place in the presence of an additional female interlocutor with whom the target child was not entirely familiar.This presence of an unfamiliar female may have triggered the repair on behalf of the speaker.
According to the above discussion, we may summarize child overextension as follows: (5) Child overextension of ix a. employed solely by male speakers b. characterized by overextension to a variety of classifiers (opaque & polysemous) c. used in the presence of both female and male interlocutors d. used deictically in reference to other males not physically or immediately present at the time of utterance e. restricted subordination of multiple classifiers to ix These features experience minimal overlap with the adult use of noun classifier extension and, apart from male-only production, the dominant characteristics (5b-c) are manipulations of the adult use which broaden the original parameters as the result of systematic application.Still, child overextension differs significantly from the adult extensional use described in Section 3.1, with the principal distinction resting on the systematic nature of the former versus the functional and ideological use of the latter.

Conclusions and Implications
Adult noun classifier extension of ix to naq and child overextension of ix to multiple classifiers are two recent innovations within the nominal classificatory system of Q"anjob"al Santa Eulalia (Q"SE).Adult extension corresponds to multiple pragmatic and ideological functions, which include situational elements governing use and implications based on gender stereotypes.Noun classifier overextension by male children appears to have been motivated by adult extensional use.Children who overextend the noun classifier ix are, in a sense, acquiring a gender stereotype.However, the child data suggests that children"s overextensions are characterized by relatively few of the properties that characterize adult extension.Moreover, at present there is no evidence to suggest that children"s overextensions are ideologically motivated, in contrast to adult usage that appears to be based on social constructions of gender.Thus, children are acquiring the form of a gender stereotype apart from its ideological function.
Comparing the attributes of adult extension with child overextension, we note the following properties: Table 2: Properties of adult noun classifier extension and child ovextension.
As Table 2 illustrates, the properties shared by both adult extension and child overextension are in the minority.Moreover, comparing these two innovative uses of noun classifiers reveals significant differences in terms of situational context, ideology, and amount of extension.
Children not only extend ix to naq, they also overextend ix to inanimate and non-human animate nouns.Moreover, child extensional usage occurs in the presence of other males as well as females, and they do not employ the non-canonical use of ix jokingly or derogatorily.The qualitative differences between adult noun classifier extension and child noun classifier overextension illustrate the need for broader conceptualizations of pragmatics, discourse, and context as they pertain to noun classifiers in Q"SE in order to provide a fuller account for genderbased variation and pragmatic functions.That is, a broader sense of pragmatic could allow for a more explicit understanding of intentionality and what speakers are doing with the particular extensional forms they employ.The ideological component of adult extensional use suggests that subsequent research not only "focus on […] isolated words and sentences," but on "larger units" such as "texts, discourses, conversations speech acts, [and] communicative events" (Wodak and Meyer 2009:2).With this in mind, the notion of discourse could be expanded to explore a variety of textual relationships, where discourse context includes not only the immediately surrounding utterances (i.e., cot-text) but also the larger communicative situations along with the social and historical contexts in which extensional use is embedded.This focus could additionally lead to a "study of action and interaction" between speakers of Q"SE as well as "the study of the functions of (social, cultural, situative and cognitive) contexts of language use" (Wodak and Meyer 2009:2, emphasis original), which in turn would aim at understanding the pragmatic and ideological forces that underlie adult extension, such as social constructions of gender.
Moreover, incorporating the broader sense of these concepts in future exploration of this topic could bring additional aspects into focus, such as how such language use by adults may relate to identity.For example, in this paper I have discussed the interaction between language (use), gender, and biological sex.I treated this interaction in terms of the function of ideology (e.g., derogatory reference implying negative attitudes toward females), but did not address the potentially performative nature of gender as a core component of identity (Butler 1990) through adult classifier extension.Recent sociological and linguistic research also suggests a close relationship between language, gender, and sexuality (Butler 1990(Butler , 2004;;Morrish and Sauntson 2007;Livia and Hall 1997), which could be a factor in adult male Q"SE speakers using classifier extension to refer derogatorily to other males.Future treatment of noun classifier extension by adult males could aim at investigating the relevance of performing identity with regard to gender and sexuality.
Although child overextension lacks ideological impetus, it stands in a causal relation to the more restricted adult extension of ix to naq.Pragmatic and ideological influences, then, are a driving force in child language acquisition and development, though at present children appear to only acquire a small portion of the pragmatic features and none of the ideological ones.Pragmatic and ideological influences also are informing language change as noun classifiers in Q"SE currently undergo development.Therefore, the present research pertains to the areas of language acquisition, language and gender, and linguistic typology by providing some insight to the uniqueness of non-canonical classifier use in Q"SE.
From a typological and historical perspective, Q"SE appears to be evolving through multiple stages in regards to its system of nominal classification, based on trends that are diachronically and synchronically observed.These stages, delineated below, are symbolized as synchronically represented language states (based on Croft 2003:244): States: no noun CL  noun CL , no extension  adult noun CL extension  child noun CL overextension  future change?Each of the above states represents respective stages of language change.The first stage is a synchronic view of Q"SE prior to the grammaticalization process that led to a developed system of nominal classification, which, among Mayan languages, was a Q"anjob"alan innovation (Craig 1990).The second stage represents typical noun classifier use.That is, this type of use follows the semantic, morphosyntactic, and discourse-pragmatic (i.e., in the sense of anaphoric reference) parameters previously mentioned, which are in accordance with canonical usage.Stage three emerged closer to the present time, and depicts the still current adult extension of ix to naq, a second Q"anjob"alan innovation which Mateo Pedro (2004) concludes to be a persistent phenomenon over the last 50 years.The fourth stage demarcates the emergence of noun classifier overextension by male child speakers.It is perhaps contestable as an actual "stage" of language change.That is, the present amount of child extension and overgeneralization does not definitively determine such usage as a stage of language change.Finally, the fifth stage looks toward the potential future of the Q"SE system of nominal classification as a result of present extensional and overextensional use.
Under previous typological and diachronic considerations of various nominal classification systems, at least two proposals have been offered as distinct evolutionary processes of grammaticalization that ultimately result in either a nomininally-derived classifier or a noun class/grammatical gender system (Grinevald 2002).These proposals are as follows (Grinevald 2002:264) Grinevald notes that first process in (6a) is prevalent in some Southeast Asian languages, while the second (6b) occurs on many continents (Grinevald 2002:264).The above processes that I propose align with the second of these evolutionary models.This is because noun classifiers in Mayan languages, and Q"anjob"al in particular, derive from nouns and not intermediary class terms.Furthermore, since noun classifier systems are dynamic and not static entities, change is to be expected as a direct result of this dynamism and the intermediate status of noun classifiers on a lexico-grammatical continuum (Carpenter 1986:23-24;Grinevald 2002:264-265).
Bearing in mind this sense of dynamism, below are two separate hypothetical future stages that could stem from the current diachronic trajectory of the Q"SE noun classifier system: generic CL? Process: noun > noun CL/discourse particle (anaphor) > pragmatic CL noun class?
In this representation, adult extension and child overextension are collapsed within a single designation of pragmatic classifier.In the proposed trajectory that results in a generic classifier, this would entail that child overextension continues and gains prevalence.In the alternate trajectory that results in a noun class, this would entail at least a higher degree of grammaticalization with morphosyntactic marking on the classified nominal (Grinevald 2002:260).However, the evolution of the innovative nominal classificatory system of Q"anjob"al has not resulted in the creation of a noun class, or system of grammatical gender.
Since the child noun classifier extension is trending toward polysemy, in the future this polysemy could trigger additional change if the use is common and occurs with high frequency.The single human noun classifier ix could become semantically bleached, leading to continued overextension.If this occurred as an outcome of child overextension, ix might begin to lose its selectional semantic properties of [+human] or [+female], or both, and we would expect ix to cooccur with a large variety of nouns, both animate and inanimate.This proposal is similar to what is currently taking place in child speech, although it would need to take place to a greater extent.Still, it is unclear how continued pragmatic and ideological usage of extension by adult males might interact with child overextension.For example, child learners" overextension might assimilate to the properties that characterize adult usage as the children get older.
On the other hand, language learners might continue to apply overextension, which could then result in further overextension.If this occurs, we would expect to see additional classifiers being overextended, and not ix exclusively.Yet no instances of this type have yet been documented.Any of these possibilities might signal the onset of a transitory language state, though the current situation cannot be termed such since the primary source of this type of extension comes from language learners.A further possibility is that the current extensional use would fluctuate and then revert back to ungrammaticality.For this case, we would expect to encounter fewer occasions of noun classifier extension as time progresses.
Much remains to be done in exploring this second of Q"anjob"alan innovations with regard to noun classification.The properties of adult noun classifier extension in contrast with child overextension suggest that continued research would benefit from investigating these complex processes by integrating multiple perspectives aimed at understanding the interaction between language (use) and ideology, especially with regards to social constructions of gender, as well as their effects on acquisition and language variation.

Figure 4 :
Figure 4: Sample adult male classificatory taxonomy of the extensional use of ix, along with the retained canonical use of ch'en and no(').

Table 1 :
Noun Classifiers in Q"anjob"al Santa Eulalia illustrates semantic boundaries between classifier types according to canonical use.Boundaries between classifier types are signaled by a vertical dashed line: Figure 2: Sample classificatory taxonomies according to canonical use of noun classifiers ix, naq, ch'en and no(').