Elsevier

Cognition

Volume 66, Issue 3, June 1998, Pages 215-248
Cognition

A cross-linguistic comparison of generic noun phrases in English and Mandarin

https://doi.org/10.1016/S0010-0277(98)00021-3Get rights and content

Abstract

Generic noun phrases (e.g. `bats live in caves') provide a window onto human concepts. They refer to categories as `kinds' rather than as sets of individuals. Although kind concepts are often assumed to be universal, generic expression varies considerably across languages. For example, marking of generics is less obligatory and overt in Mandarin than in English. How do universal conceptual biases interact with language-specific differences in how generics are conveyed? In three studies, we examined adults' generics in English and Mandarin Chinese. The data include child-directed speech from caregivers interacting with their 19–23-month-old children. Examples of generics include: `baby birds eat worms' (English) and da4 lao3shu3 yao3 bu4 yao3 ren2 (`do big rats bite people or not?') (Mandarin). Generic noun phrases were reliably identified in both languages, although they occurred more than twice as frequently in English as in Mandarin. In both languages, generic usage was domain-specific, with generic noun phrases used most frequently to refer to animals. This domain effect was specific to generics, as non-generic noun phrases were used most frequently for artifacts in both languages. In sum, we argue for universal properties of `kind' concepts that are expressed with linguistically different constructions. However, the frequency of expression may be influenced by the manner in which generics are expressed in the language.

Introduction

Generic noun phrases (e.g. `dogs chew bones', `a tiger is a wild animal') provide a window onto human concepts. They refer to a category as an abstract whole (e.g. dogs as an entirety), not to any individual or group of individuals (e.g. Fido; the dogs we saw yesterday at the park). Indeed properties predicated of whole categories (such as `extinct') require generic subjects. We can say that dinosaurs are extinct, but not that any particular dinosaur is extinct. Furthermore, generics refer to qualities that are relatively essential, enduring, and timeless – not accidental, transient, or tied to context (Lyons, 1977). Thus, generics imply that a category is a coherent, stable entity. We can refer to these coherent, stable categories as `kinds' (cf. Gopnik and Meltzoff, 1997).

Kinds play an important role in human cognition because they are used to organize knowledge and guide inferences. Recent psychological studies demonstrate that thinking about kinds leads people to make rich inferences about the world (e.g. Gelman and Markman, 1986; Shipley, 1993; Gopnik and Meltzoff, 1997). Once one learns that something is a member of a kind (for example, that a chihuahua is a dog), one tends to infer that the entity shares properties with others of the same kind (Gelman and Markman, 1986; Gelman and Markman, 1987).

Although kind concepts are often assumed to be universal, generic expression varies considerably across languages (Krifka et al., 1995). The present studies examine generics in two languages that express them quite differently: English, which conveys generics by means of obligatory morphosyntactic cues, and Mandarin Chinese, which does not. Our question is whether and in what ways these language differences affect how generics are used. In other words, how do universal conceptual biases interact with language-specific differences in how a concept is expressed?

In English, generics can be expressed with definite singulars, bare plurals, or indefinite singulars (cf. Lyons, 1977):

  • a. The bird is a warm-blooded animal.

  • b. A cat has nine lives.

  • c. Dinosaurs are extinct.

They can be contrasted with non-generic expressions such as the following:

  • d. The bird is flying.

  • e. A cat caught two mice.

  • f. There are some dinosaurs in the museum.

As can be seen in the examples above, generics in English are not uniquely identified with a particular form of the noun phrase, but instead are cued by a variety of morphosyntactic, semantic, and pragmatic cues, including the form of the noun phrase and of the verb. Specifically, there are at least four morphosyntactic cues that help a speaker identify an utterance as generic or non-generic: determiners, number (i.e. singular vs. plural), tense, and aspect. We treat determiners and number together, since they jointly operate to indicate genericity. In English, a plural NP preceded by the definite determiner (the) cannot be generic. For example, `bears are brown' is generic, but `the bears are brown' is not. Neither determiners nor number alone indicate whether or not an NP is generic. However, it is the interaction of the two (i.e. definiteness plus plurality) that provides information regarding genericity. Tense is also an indication of genericity. With the exception of historic past (e.g. `woolly mammoths roamed the earth many years ago'), past-tense utterances are not generic. For example, we distinguish between `a cow says `moo' (generic) and `a cow said `moo' (non-generic). Likewise, `the lion is ferocious' can have either a generic or a non-generic reading, whereas `the lion was ferocious' has only a non-generic reading. Finally, aspect is an important cue in English for distinguishing generic from non-generic interpretations. For example, a statement in the simple present, such as `cats meow', is generic, whereas a statement in present the progressive, such as `cats are meowing', is non-generic. Thus, in English, some of the cues relevant to whether an NP is generic include articles, plurality, tense, and aspect.

In contrast, Mandarin does not have three of these grammatical distinctions (articles, plurality, or tense). The fourth grammatical distinction (aspect) does not always appear in Mandarin, and is even ungrammatical with some verb types (Li and Bowerman, 1998)1. As a consequence, there are sentences in Mandarin that could be translated as either generic or non-generic in English (Krifka, 1995). For example, the following sentence:

  • g.: xiao3 ya1zi yao2yao2bai3bai3 de zou3 lu4

  • little duck waddlingly DE walk road

could be translated in any of three different ways:

  • h. The duck is waddling,

  • i. The ducks are waddling, or

  • j. Ducks waddle.

Only the third is generic. This does not mean that Mandarin fails to express generics. There are subtle semantic and pragmatic cues that may help clarify the status of an utterance (Krifka, 1995). For example, the absence of specific number, time, or place markers can imply a generic interpretation. Nonetheless, the lack of such morphosyntactic cues as articles, plurality, tense, and (sometimes) aspect, means that the expression of generics is less overt in Mandarin than English.

English generics are also detected in part on the basis of contextual cues, such as the absence of specific time and place markers. Thus, the marking of English generics is not wholly unambiguous, and the difference between English and Mandarin is a difference of degree rather than a qualitative difference. To summarize that difference: in English, there are some constructions for which it is possible to decide on the basis of morphosyntactic cues whether or not a generic interpretation is intended, whereas there are other constructions for which one must rely on pragmatic and contextual uses to determine whether or not a generic interpretation is intended. In contrast, in Mandarin all constructions are of the latter sort.

What are the implications (if any) of these linguistic differences for how generics are interpreted and expressed? It may be that there are no such implications. In other words, one possibility is that semantics are universal, and the form through which meanings are expressed is wholly incidental to language use or interpretation. However, there are both empirical and theoretical justifications for examining a link between generic use and means of generic expression. A variety of recent empirical studies have found that cross-linguistic differences in language form affect semantics: for words referring to motion events (Choi and Bowerman, 1991), spatial prepositions (Bowerman, 1996), object and substance terms (Imai and Gentner, 1997), number-marking patterns (Lucy, 1992a), and adjectives (Waxman et al., 1997).

For example, Bowerman (1996)finds that children express spatial prepositions in language-specific ways from very early in language learning. Whereas children learning English use notions of containment and support to guide their use of spatial prepositions, children learning Korean use notions of tight fit versus loose fit, which cross-cut the dimensions used by English speakers. Bowerman concludes, on the basis of her evidence, that: `children's semantic categories for spatial terms may already be profoundly language-specific even before the age of two' (p. 146).

With regard to generics, Chierchia (unpublished data) likewise suggests that there can be cross-linguistic variation in semantics. Based on linguistic analyses of different languages, he argues that the syntactic category N (noun) may map onto different semantic types across languages (including English vs. Chinese), and that this finding requires `giving up a widely shared idea, viz., that the syntax–semantics map is completely universal' (Chierchia, unpublished data).

Moreover, Lucy (1992a)argues that when a language signals a semantic distinction more obligatorily (e.g. plural in English vs. Yucatec), then speakers of that language should `habitually attend' (p. 87) more to that semantic distinction. In English, generics themselves are not obligatorily marked; however, both number and determiners are obligatory and, as noted earlier, contribute to presentation of a NP as either generic or non-generic. Thus, speakers of English should habitually attend to the status of an utterance as generic or not. In contrast, in Mandarin there is no obligatory marking of either number or determiners, thus contributing to the presentation of many NPs in a neutral way (neither generic nor non-generic).

For example, in a discussion of noun phrases in ancient Chinese, Dobson (1959); pp. 21–22) says the following:

An important observation must be made here about nominal usage in LAC [Late Archaic Chinese]. A word used nominally denominates indifferently both species and specimen or specimens of the species. Thus, jen in a nominal usage is indifferently `homo sapiens,' `man' as a species, or `a man, the man, men' as a specimen or specimens of the species. It is not merely that number (the difference between one, two, or more than two instances of) is not differentiated, but that both class and member are comprised in one term. Certain of these distinctions are imposed in determination, and the noun is then said to be `committed'; otherwise, it remains neutral in its indications as to these distinctions.

Although Dobson's analysis was of Late Archaic Chinese, the same can be said of modern Chinese as well, when additional markers or phrases (e.g. time adverbs, deictic pronouns, etc.) are not added to specify the meaning.

This analysis of generics contrasts with the case of simple ambiguity. Our claim is not that Mandarin speakers always have either a generic or a non-generic interpretation in mind, and simply express both with forms that are similar on the surface (as is the case of homonyms, such as bank). Rather, we suggest that consideration of a NP as generic or not can at times be bypassed in Mandarin. This is roughly analogous to the distinction between formal and informal presentations of `you' in languages such as French. The French language forces speakers to select either a formal or an informal interpretation, with every use of vous or tu. In contrast, although English certainly has devices for signaling level of formality (e.g. `sir' or `ma'am' as second-person address forms), they are not obligatory, morphosyntactic devices. Correspondingly, uses of you (in English) are typically neither formal nor informal, but rather neutral with respect to this dimension.

To summarize, then, we assume that the generic–non-generic distinction is a universal conceptual distinction that is expressed in languages with widely varying morphosyntactic systems. Nonetheless, we hypothesize that, for speakers of English, the use of obligatory markings to express generics heightens the salience of generic concepts, and thus leads them to be expressed more frequently in English than in Mandarin.

There is in fact a long-standing claim that linguistic differences between Mandarin and English may lead to conceptual differences. Typically, this claim is expressed in terms of broad differences along a concrete/abstract dimension (Lin, 1935; Granet, 1934; Wright, 1953; Hu, 1963; Bloom, 1981; Hansen, 1983; Bodde, 1991; see Moser (1996)for discussion). For example, Wright (1953)suggests: `the Chinese [language] was relatively poor in resources for expressing abstractions and general classes or qualities. Such a notion as `Truth' tended to devolve into `something that is true'. `Man' tended to be understood as `the people' – general but not abstract. `Hope' was difficult to abstract from a series of expectations directed toward specific objects' (p. 287). Relatedly, others have suggested that there may be cultural differences in how Chinese versus American people reason about categories, with greater accessibility of categories among Americans (Choi et al., 1997). For example, in their literature review, Choi et al. (1997)state, `Some ethnographers, philosophers, and historians of science have proposed that there are differences in reasoning style between Westerners and East Asians of a very broad nature that include a different emphasis on the role of categories in thought.'

Generics are but one example of this broader set of claims. The literature that specifically focuses on generics in Chinese versus English includes the claim that linguistic differences in expression of generics lead to corresponding conceptual differences in how speakers of Mandarin versus English think about abstract kinds (Bloom, 1981; Moser, 1996). However, as we review briefly below, there are no clear data available that are relevant to the issue.

Bloom (1981)states the linguistic relativity hypothesis most starkly: `... the Chinese language does not have any mechanism... with which to signal the generic concept...' (p. 38).

Perhaps the fact that English has a distinct way of marking the generic concept plays an important role in leading English speakers, by contrast to their Chinese counterparts, to develop schemas specifically designed for creating extracted theoretical entities, such as the theoretical buffalo, and hence for coming to view and use such entities as supplementary elements of their cognitive worlds (p. 36).

Thus, Bloom's suggestion has two parts: (1) that generics have no means of expression in Chinese, and (2) that this linguistic difference leads to corresponding conceptual differences.

Yet Bloom's evidence for this position was insufficient, on his own admission (p. 36). Specifically, to provide preliminary support, he presented a single sentence to 110 Taiwanese subjects: `dai4shu3 shi4 chi1 luo2bo de dong4wu4'2 ((The) kangaroo(s) is/are (a) turnip-eating animal(s)). Subjects were asked whether the sentence, `in addition to referring to an actual kangaroo, to some actual kangaroos or even to all actual kangaroos, might have an additional interpretation, for example, as a conceptual kangaroo' (guan1nian4 shang4 de dai4shu3). A positive response to this question was deemed evidence of a generic interpretation. Thirty-seven percent of the subjects said yes, which Bloom interpreted as a low number, and most of the `yes' responders already knew English, suggesting that their interpretation could have been influenced by their knowledge of English, rather than a functional knowledge of generics in Mandarin. Bloom inferred from these results that Chinese speakers have difficulty grasping generic concepts, although he cautioned that further research is needed.

This initial study, though intriguing, raises a variety of questions (see also Lucy (1992b)pp. 230–232, for a related critique). First, judgment of the test sentence posed a difficult metalinguistic task. The question itself is ambiguous and open to alternative interpretations (i.e. what is meant by a `conceptual kangaroo'?), and thus is likely to lead to confusion. In other words, the task may not have been a sensitive index of subjects' generic concepts. Moreover, the experiment included only Mandarin speakers tested in Mandarin, and did not test the same speakers in English, nor did it include a separate group of monolingual English speakers. Because there was no comparison group of English speakers, it is (a) difficult to know whether 37% agreement is relatively high or low, and (b) impossible to draw cross-linguistic conclusions. Thus, we cannot assume that English speakers would perform any differently.

Another limitation is that, in his analysis of English, Bloom focused on the definite singular form of the generic (`the X'), and did not discuss either the indefinite singular form (`an X') or the bare plural construction (`Xs'). One can form generics in English using any of these three forms (e.g. `the kangaroo is a turnip-eating animal'; `kangaroos are turnip-eating animals'; `a kangaroo is a turnip-eating animal'). We suggest that the notion of `a conceptual kangaroo' is sensible only when one focuses on the definite singular form of the generic (which, incidentally, is the least commonly found form in ordinary speech (Pappas and Gelman, 1998)). When considering plural generics (e.g. `kangaroos are turnip-eating animals'), it no longer seems apt to regard the generic noun phrase as referring to a conceptual individual. In other words, plural generics may call to mind the category as a whole (e.g. the category that includes most or all kangaroos), rather than a single abstract entity (e.g. a `conceptual' kangaroo).

Finally, in an anecdotal example provided by Bloom to illustrate lack of generic understanding, one speaker provided evidence that we interpret as potential sensitivity to generic concepts. Specifically, an adult speaker with `a very modest command of English' claimed not to understand the question about a conceptual kangaroo. However, `from the generality of the content of the statement and the lack of any kangaroos in the vicinity or previous mention of any, she inferred that the sentence must be referring to plural kangaroos; in fact, to all kangaroos' (suo3you3 de dai4shu3). We believe that this capacity to differentiate between particular kangaroos in specific contexts and the set of all kangaroos nicely captures the distinction between non-generic and generic interpretations. However, it is not clear how frequently people reached this interpretation, as Bloom did not report (and appears not to have asked) how many of the 110 speakers interpreted the sentence as referring to the set of all kangaroos.

In sum, Bloom's argument is that generic concepts are less available to speakers of Mandarin. The evidence used to support this hypothesis was a comprehension task, with findings that could be interpreted as showing that speakers of Mandarin reach non-generic readings of utterances that could be considered generic when translated into English. However, support for this hypothesis was weak. Moreover, if generic meanings are less available to speakers of Mandarin, then we should also expect to find that speakers express them less frequently in productive speech. Until now, no evidence was available on this issue from production data.

Moser (1996)put forth a different set of claims, focusing on both ancient and modern Chinese. In contrast to Bloom, he suggested that generics can be produced in Mandarin, but that generic concepts are less accessible to metacognitive awareness. (Another way of phrasing this is to say that overt marking of generics in languages such as English increases their availability to conscious awareness.) He suggested: `... there is evidence for at least count, mass, and generic nouns [sic] distinctions in classical Chinese. But since these distinctions were dealt with as cryptotypes in the language, and because their application was less obligatory, there seems to have been no meta-linguistic awareness of the distinctions on their part' (p. 167). This aspect of Moser's argument, although intriguing, remains untested.

Instead, Moser presented some preliminary evidence that speakers of Mandarin are not random in their interpretation of unmarked Ns. When asked to substitute either singular or plural third-person pronouns for nouns that English-speakers would consider generic, Mandarin speakers were often consistent across speakers for a given sentence (although variable across sentences). For example, for the word `kangaroo' in the sentence `kangaroo is/are turnip-eating animal', participants selected the third-person singular pronoun (ta1) 66% of the time, whereas for the word `panda' in the sentence `panda is/are on the verge of extinction' they selected the third-person plural pronoun (ta1-men) 63% of the time. However, these data are difficult to interpret, without either baseline comparisons of some sort (e.g. from speakers of a language that marks generics overtly), or predictions regarding which interpretations will be reached under various conditions. Most problematic for our purposes, either singular or plural pronouns could be consistent with a generic reading, and either could be consistent with a non-generic reading.

Thus, we are left with some powerful claims in the literature regarding the semantic and/or conceptual implications of how generic expression varies cross-linguistically – but little in the way of compelling or relevant data.

In the present set of studies, the primary question is whether different formal/linguistic means of expression across languages influence the frequency and use of generics in natural speech. Although we pose this as a general question, it is also of particular interest from a developmental perspective. Children's category knowledge is not just directly perceived or innately pre-wired (Wellman and Gelman, 1997). How, then, do children come to acquire this knowledge of kinds? Perhaps a major influence on the development and expansion of children's knowledge about a category is caregiver talk about the category, and that occurs most obviously with generic noun phrases (Gelman et al., 1998). In particular, generics have the potential to enrich children's categories in two ways. (1) They convey specific information concerning properties that members of a kind have in common, thus contributing to an elaborated knowledge base. For example, `bats fly at night' conveys a specific fact about bats that might never be observed directly by the child, thereby enriching the child's knowledge of a particular property typically shared by bats. (2) Generics imply that members of a category are alike in important ways, even when little or no specific information is conveyed. For example, the statement `bats are brown' may not tell the child anything new that he or she hasn't already observed, but nonetheless implies that the category (bats) is a kind about which one can make a broad generalization. Indeed, we suspect that even generic questions (e.g. `are bats brown?'), which themselves provide no information concerning the category in question, may convey to children that the category in question (bats) is richly structured, because they make reference to the category as a kind.

Thus, even if children have a propensity to construct kind concepts on their own (Gopnik and Meltzoff, 1997), generics in caregiver speech may play an important role in helping children to determine which categories are kinds, and in helping children to amass specific knowledge concerning kinds. In English, this may be facilitated by specific grammatical structures. In Mandarin, this acquisition issue is potentially problematic given that there is little explicit formal marking of generic forms in adult speech.

To examine the frequency and use of generics in the natural speech of caregivers, we first ask whether generic NPs can be reliably identified in Mandarin, and if so, how frequently they appear, relative to English. One position, the Conceptual Universals position, suggests that generic usage should be influenced by universal conceptual underpinnings and independent of how generics are expressed in the language. Thus, frequency and distribution of generic concepts should be substantially similar across English and Mandarin. In contrast, a second position, the Linguistic Relativity position, would argue that means of linguistic expression would influence speakers' tendency to produce generics (and perhaps even their tendency to think generically, although this is not assessed in the present paper). On this view, generics should be less frequent in Mandarin versus English.

The second major question we ask concerns how generics map onto semantic domains. Recent psychological work suggests that concepts are organized differently in different domains (e.g. Gelman, 1988; Keil, 1989; Wellman and Gelman, 1997). Specifically, animal concepts (such as `dog' or `skunk') appear to be more richly structured than artifact concepts (such as `chair' or `hat'). Animal concepts have more shared features, support more inferences about novel properties, and retain identity across striking transformations. Given these conceptual differences between animals and artifacts, we predicted that generics would show the same domain differences. In particular, we expected generics to be favored in the animal domain.

Past linguistic analyses of generics have tended to ignore semantic domain, perhaps because there are no formal restrictions on which domains can be used. It is certainly possible to form generics concerning a broad array of concepts (including animals, artifacts, plants, inanimate natural kinds, food, etc.). Therefore, if subjects differentiate among domains in generic usage, this would presumably reflect something about the mapping between semantics and conceptual structure. There is preliminary support for the idea that English-speaking mothers preferentially use generics for the domain of animals (Gelman et al., 1998). The comparison between English and Mandarin is especially useful here, as it tells us indirectly how generics function across languages. If we find clear domain differences in the frequency of generics, and if these differences are consistent across the two languages, this would imply that there is a common conceptual basis for generic use in English and Mandarin.

Studies 1 and 2 examine productive use of generics among adult caregivers using child-directed speech. Study 1 uses naturally-occurring speech samples of children interacting with their caregivers in or around the home. Study 2 uses natural speech gathered within a laboratory setting, thus providing more control over context, available props, and length of interaction. Study 3 examines generic comprehension in a sample of adults, and serves as a comparison to Study 2. All three studies provide evidence regarding Bloom's hypothesis that generic concepts are less available to speakers of Mandarin, with a focus on either language comprehension (Study 3) or production (Studies 1 and 2).

Section snippets

Mandarin

Ten children, eight boys and two girls, and their families participated. All resided in Beijing, China. Children ranged in age from 21 to 23 months; mean age 21.7 months; child mean lengths of utterance (MLUs, see Brown (1973)) ranged from 1.12 to 2.82, mean MLU 1.82. Of the ten children, five had parents who were classified as `intellectuals' (zhishifenzi: college-educated or above, professional and semi-professional occupations), and five had parents who were classified as `workers' (gongren:

Subjects

Twenty-four English-speaking children (12 first-born or only children, 12 later-born) and their mothers were recruited from the subject pool in a midwestern university town in the United States. Half of the children were male and half were female. Their mean age was 20 months, 20 days (SD=1 month, 1 day). The mean level of educational attainment for their mothers and fathers was 16.8 years (SD mothers=1.6, SD fathers=2.3), or almost 5 years of post-secondary schooling. One additional child was

Study 3

To this point, the data examine production of generic noun phrases, but we have no evidence regarding generic comprehension. Yet comprehension data are crucial if we wish to make claims regarding the psychological reality of the generic coding in Studies 1 and 2. In other words, comprehension data are needed to help establish that the utterances our coders considered generic are interpreted as generic by ordinary speakers. To some extent we have implicitly addressed this issue by finding that

General discussion

The results of the present three studies suggest an interaction between cognitive universals and language-specific effects. On the one hand, we argue for universal properties of `kind' concepts that are expressed with linguistically different constructions. On the other hand, the frequency of expression may be modified by the manner in which generics are expressed in the language.

Acknowledgements

We thank the parents and children who participated in the study, and Ryan Kettler, Simone Nguyen, Shanping Qiu, and Elizabeth Kohn for their able research assistance. We are very grateful to Erika Hoff-Ginsberg for permitting us to analyze her transcripts of mother-child conversations in Study 1 and to the Institute of Psychology, Chinese Academy of Sciences, for their assistance in collecting the Mandarin data in all three studies. We also thank three anonymous reviewers for their helpful

References (48)

  • Brown, R., 1973. A First Language. Harvard University Press, Cambridge,...
  • Carlson, G.N., Pelletier, F.J. (Eds.), 1995. The Generic Book. Chicago University Press,...
  • Croft, W., 1990. Typology and universals. Cambridge University Press, New...
  • Dobson, W.A.C.H., 1959. Late Archaic Chinese. University of Toronto Press,...
  • Fenson, L., Dale, P.S., Reznick, J.S., Bates, E., Thal, D.J., Pethick, S.J., 1994. Variability in early communicative...
  • Gelman, S.A., Coley, J.D., Rosengren, K.S., Hartman, E., Pappas, A., 1998. Beyond labeling: the role of maternal input...
  • S.A Gelman et al.

    Young children's inductions from natural kinds: the role of categories and appearances

    Child Development

    (1987)
  • Gentner, D., Boroditsky, L., 1998. Individuation, relativity, and early word learning. In: Bowerman, M., Levinson, S....
  • Gopnik, A., Meltzoff, A.N., 1997. Words, Thoughts, and Theories. MIT Press, Cambridge,...
  • Granet, M., 1934. La Pensée Chinoise. Éditions Albin Michel,...
  • Greenberg, J.H., 1966. Language Universals. Mouton, New...
  • Hansen, C., 1983. Language and Logic in Ancient China. University of Michigan Press, Ann...
  • E Hoff-Ginsberg

    Mother–child conversation in different social classes and communicative settings

    Child Development

    (1991)
  • Hu, S., 1963. The Development of the Logical Method in Ancient China. Paragon Books, New...
  • Cited by (0)

    View full text