~fo i^rt&cctitrzitx ctrtctti/ntzi (keeping in mind that they are not always conscious of what they mean and do), citations from related literature, and collaboration with others to begin to achieve some degree of validity in regard to agreement. Note This book is an introduction to one particular approach to discourse analysis and I have made no attempt to compare and contrast this approach to others. For introductions to other approaches, see van Dijk (1985) and Scliiffrin (1994). Two volumes edited by van Dijk (1997a, b) contain articles on a wide variety of approaches to, aspects of, and topics in discourse analysis. "Conversational analysis" is a specialized approach to discourse analysis centered in sociology and is discussed in van Dijk 1997b (see the paper by Pomerantz and Fehr, pp. 64-91); see also, Goodwin and Heritage (1990); Ochs, Schegloff, and Thompson (1996), andPsathas (1995). Malone (1997) does a good job of combining conversational analysis and symbolic interactionism. Macdonnell (1986) and Mills (1997) are short introductions to discourse with a focus on feminist, poststructuralist, and postmodern work. Fairclough's (1989, 1992, 1995) "critical discourse analysis," though drawing on somewhat different tools of inquiry and a somewhat different linguistic tradition, none the less bears important similarities to the approach sketched in this book. Lemke (1995) and Kress (1985; Kress and van Leeuwen 1996) are both important approaches to discourse based on "social semiotics." Hicks (1995) overviews some approaches to discourse analysis as they apply to teaching and learning. Luke (1995) is a good discussion of issues of discourse, politics, and education. Judith Green and David Bloome have developed a distinctive approach to discourse in classrooms (see Bloome 1987; Bloome and Egan-Robertson 1993; Green and Dixon 1993; Green and Harker 1988; Green and Bloome 1997; Santa Barbara Discourse Group 1992; see also Gee and Green 1998). Duranti (1997); Duranti and Goodwin (1992); Gumperz and Levin-son (1996); and Levinson (1983) are good overviews of larger approaches to language that incorporate discourse analysis. 6 Processing and organizing language 6.1 Speech in produced in small spurts This chapter deals with a few aspects of how speech is produced and what this has to do with the sorts of meanings we speakers hope to convey and we hearers (always actively and creatively) try to "recover." We will deal here with a few technical details about the structure of sentences and of discourse. However, these details are not important in and of themselves. What is important is that the discourse analyst looks for patterns and links within and across utterances in order to form hypotheses about how meaning is being constructed and organized. What grammatical tenninology we choose to use is less important than the patterns we find and the hypotheses we form and test. Notions like "situated meanings," "cultural models," and "Discourses" will take a back seat here. In this chapter we are primarily concerned with some initial ways into a text. We are concerned with ways in which the analyst can start to organize his or her thinking about a piece of language. Of course these initial insights must quickly lead to thinking about situated meanings, cultural models, and Discourses. In turn, ideas about these will influence and, at times, change how the analyst thinks about the linguistic patterns in a text. Discourse analysis is a reciprocal and cyclical process in which we shuttle back and forth between the structure (form, design) of a piece of language and the situated meanings it is attempting to build about the world, identities, and relationships. Thanks to the way the human brain and vocal system are built, speech, in all languages, is produced in small spurts (Chafe 1979,1980,1994). Unless we pay close attention, we don't usually hear these little spurts, because the ear puts them together and gives us the illusion of speech being an unbroken and continuous stream. In English, these spurts are often, though not always, one "clause" long. In a rough and ready way we can define a "clause" here as any verb and the elements that "cluster" with it (see the appendix for more on grammar). So in a sentence like "Mary left the party because she was tired," we have two clauses, "Mary left the party" and "because she was tired." The 100 Processing and organizing language sentence "Mary left the party" contains only one clause. In a sentence like "Mary intended to leave the party," we also have two clauses, "Mary intended" and "to leave the party" (where "Mary" is understood as the subject of "to leave"). Here the second clause ("to leave the party") is embedded in the first clause ("Mary intended") as the direct object of the verb "intend." These two clauses are so tightly bound together that they would most often be said as a single spurt. In the example below, taken from a story told by a seven-year-old child, each spurt is one clause long, except lb and le where the child has detached parts of clauses to be spurts on their own (of course, children's speech units tend to be shorter than adults): la there was a hook lb on the top of the stairway lc an' my father was pickin me up Id an' I got stuck on the hook le up there If an' I hadn't had breakfast lg he wouldn't take me down lh until I finished all my breakfast li cause I didn't like oatmeal either To understand how these spurts work in English (they work differently in different languages), we need to discuss a set of closely interrelated linguistic concepts: function words, content words, information, stress, intonation, lines, and stanzas. We will start with the distinction between function words and content words. 6.2 Function words and content words Content loords (sometimes also called "lexical words") belong to the major parts of speech: nouns, verbs, and adjectives. These categories are said to be "open categories" in the sense that they each have a large number of members and languages readily add new members to these categories through borrowing from other languages or the invention of new words. Function words (also sometimes called "grammatical words") belong to smaller categories, categories which are said to be "closed categories" in the sense that each category has relatively few members and languages are resistant to borrowing or inventing anew such words (though they sometimes do). Such categories as determiners (e.g. "the," "a/n," "this/ that," "these/those" - these are also sometimes called "articles"), pronouns (e.g. "he/him," "she/her," "it," "himself," "herself"), prepositions (e.g. "in," "on," "to," "of"), and quantifiers (e.g. "some," "many," "all," "none") are function word categories. ':4Sfc ___________________ Processing and organizing language 101 Function words show how the content words in a phrase, clause, or sentence relate to each other, or how pieces of information fit into the overall on-going communication. For example, the definite determiner "the" signals that the information following it is already "known" to the speaker and hearer. Pronouns signal that their referents have been previously mentioned, or are readily identifiable in the context of communication or on the basis of the speaker and hearer's mutual knowledge. Prepositions link nouns and noun phrases to other words (e.g. in "lots of luck," of links luck to lots; in "ideas in my mind," in links my mind to ideas; and in "look at the girl," at links "the girl" to the verb "look"). I have not yet mentioned adverbs. Adverbs are messy and complicated. Very often they function in a way that is mid-way between a function word and a content word. Since function words show how content words relate to each other, they can help us make guesses about what categories (e.g. nouns or verbs) of content words accompany them and what these words mean. To see this consider the first stanza of Lewis Carroll's poem "Jabberwocky": Twas bryllyg, and the slythy toves Did gyre and gymble in the wabe: All mimsy were the borogoves; And the mome raths outgrabe. I have underlined the function words. I have also underlined the plural affix ("es" and "s") since it functions just like a function word, though it is not a separate word. In this poem, Carroll uses real English function words, but nonsense content words (how do we know they are content words? By how they are placed in relation to the function words). Despite the fact that half the "words" in this text are nonsense, any speaker of English can use the function words to unravel the grammar of the sentences and to make good guesses about what content word categories (noun, verb, adjective) the nonsense content words belong to. The speaker of English can even make some good guesses about what the nonsense words might mean or what they might refer to. Thus, we readily interpret the stanza as a description of an outdoor scene with creatures of various sorts frolicking or moving about. 6.3 Information Since function words carry less of the real content of the communication (their job being to signal the grammar of the sentence), we can say that they tend to be infonnationally less salient than content words. While they are certainly helpful, they are often dispensable, as anyone who has written a telegram knows. 102 Processing and organizing language Thus, let us make a distinction between two types of information in a sentence. First, information that is relatively new and relatively unpredictable I will call "informationally salient." The actual specific meaning of any content word in a sentence is unpredictable without knowing exactly what the content word means. In the Carroll poem, we vaguely know that "toves" are probably active little animate creatures, but we have no idea what exactly they are. Thus, content words are usually informationally more salient than function words. Second, information that is given, assumed already known, or predictable, I will call "informationally less salient." Very often even if you have not heard a function word you could pretty well predict where it should have been and what word exactly it would have been. For example, if you heard "Boy has lots ideas," you could predict that "the" is missing in front of "boy," and "of" between "lots" and "ideas." If, however, you heard "That man has lots of," you could not predict what content word should come after "of" (though "of" signals it will be a noun or a noun phrase). Thus, function words are usually informationally less salient than content words. In general, then, the content word-function word distinction is a distinction between two types of information. However, beyond this gross dichotomy, the distinction between information that is more or less salient is one that can only be drawn in the actual context of communication. We turn to tills matter now. 6.4 Stress and intonation Information saliency in English is marked by stress. In turn the different stress patterns in a spurt of speech set up its intonational contour. To see what these tenns mean, consider the little dialogue below: 1 Speaker A: Have you read any good books lately? Speaker B: Well, I read a shocking book recently. [Goes on to describe the book] How speaker B crafts her response is partially set up by the remark made by speaker A, which here represents part of the context in which B's response occurs. Let's think a moment about how the sentence uttered by B might have been said. English speakers mark the information saliency of a word by how much siress they give the word. Stress is a psychological concept, not a physical one. English speakers can (unconsciously) use and hear several different degrees of stress in a speech spurt, but this is not physically marked in any uniform and consistent way. Stress is physically marked by a combination of increased loudness, increased length, and by changing the pitch of one's voice (raising or lowering the pitch, or gliding up or down in pitch) on a word's primary Processing and organizing language 103 ("accented") syllable. Any one or two of these can be used to trade off for the others in a quite complicated way. In any case, English speakers unconsciously use and recognize stress, and it can be brought to conscious awareness with a little practice (some people are better than others at bringing stress differences to consciousness awareness, though we can all unconsciously use and recognize it). A word with more stress than another word sounds more salient (it often sounds louder, though it may not really be louder, but just be longer or have a pitch change on it, both of which will make English speakers think it sounds louder). So let's return to speaker B's response and assume it was said as one " spurt of speech. Its first word, "well," can be said with little stress, on a relatively low pitch and/or with little loudness, since it carries no content, but simply links speaker B's turn to speaker A's. This is not to say that words like "well" are not important in other ways; such words, in fact, have interesting discourse functions in helping to link and package information across sentences. Since "well" is the first word of speaker B's spurt of speech, and starts her turn, it will be said on a pitch that is taken to be close to the "basic pitch" at which speaker B will be speaking (perhaps, kicked up a bit from B's basic pitch and, too, from where speaker A left off, to mark B's turn as beginning). "I" is completely predictable in the context of the question speaker A has asked, and it is a function word. Thus, it is not very salient informationally and will receive little stress, just enough loudness to get it said and with a pitch close to the basic pitch speaker B has chosen (for this spurt or related run of spurts as she keeps speaking). The content word "read" is predictable because it has already occurred in speaker A's preceding question. So, too, for the word "book" later in B's remark. Both of these words will have a fairly low degree of stress. They will have more than the function words "well," "I," and "a," since as content words they do carry content, but certainly much less than the word "shocking" which carries new and non-redundant information. The indefinite article "a," of course, is informationally very unsalient and will get little stress. The speaker will mark what stress words like "read" and "book" have by bumping the pitch of her voice up or down a bit from the "basic pitch" she has established or is establishing and/or by increasing loudness a bit relative to words like "I" and "a." On the other hand, the word "shocking" is the most unpredictable, informationally salient, new information in the sentence. The speaker will mark this saliency by giving this word the most stress in the sentence. Such a word or phrase, which canies the greatest degree of stress in a sentence (or a given spurt of speech) is marked not just by bumping tire pitch of the voice up or down a bit in pitch and/or by increasing loudness, but by a real pitch movement (called a "glide"). 104 Processing and organizing language The speaker begins to glide the pitch of her voice up or down (or even up-then-down or down-then-up) on the word "shocking," allowing the pitch movement to continue to glide up or down (whichever she has chosen) on the words that follow it, here "book" and "recently." Of course, what sort of pitch movement the speaker chooses, that is, whether up, down, up-then-down, or down-then-up, has a meaning (for example, the speaker's pitch glide rises in certain sorts of questions and falls in certain sorts of statements). We are not now concerned, however, with these meaning differences. The pitch glide which begins on the word "shocking" marks "shocking" as the focus of the intonation unit. An "intonation unit" is all the words that precede a pitch glide and the words following it over which the glide continues to move (fall or rise). The next intonation unit begins when the glide is finished. The speaker often hesitates a bit between intonation units (usually we pay no attention to these hesitations) and then steps the pitch up or down a bit from the basic pitch of the last intonation unit on the first word of the next unit (regardless of whether it is a content word or not) to "key" the hearer that a new intonation unit is beginning. In B's response to A, the content word "recently" is fairly redundant (not too salient) because, while it has not been mentioned in A's question, it is certainly implied by A's use of the word "lately." Thus, it receives about as much stress, or, perhaps a little more, than the content words "read" and "book." The speaker may increase her loudness a bit on "recently" and/ or bump the pitch of her voice up or down a bit on its main syllable (i.e. "cent") as her pitch continues basically to glide up or down over "recently" as part of (and the ending of) the pitch glide started on the word "shocking." Below, I give a visual representation of how speaker B might have said his utterance: shock ing read book cent Well re I a ly There are, of course, other ways to have said this utterances, ways which carry other nuances of meaning. There is one last important feature of English intonation to cover here. In English, if the intonation focus (the pitch glide) is placed on the last content word of a phrase (say, on "flower" in the phrase "the pretty red flower"), then the salient, new information is taken to be either just this word or the material in the phrase as a whole (thus, either just "flower" or the whole phrase "the pretty red flower"). Of course, the context will usually Processing and organizing language 105 determine which is the case. If the intonation focus (pitch glide) is placed on a word other than the last word in the phrase, then that word is unequivocally taken to be the salient, new information (e.g. if the intonation focus is on "red" in "the pretty red flower," then the salient, new information is taken to be just "red"). In our example above, "shocking" is not the last word in its phrase (it is an adjective in a noun phrase "a shocking book") and, thus, is rmequivocally the new, salient information. An interesting situation arises when the intonation focus (pitch glide) is placed in the last (content) word in a sentence. Then, we cannot tell whether the salient, new information the speaker is trying to indicate is just that word or also other words that precede it and go with it in the phrase or phrases to which it belongs. So in an utterance like "This summer, Mary finished fifteen assigned books," if the speaker starts her glide on "books," the new salient information she intends to mark may be just "books" (answering a question like "Mary finished fifteen assigned whats?"), or "assigned books" ("Mary finished fifteen assigned whats?"), or "fifteen assigned books" ("What has Maiy finished?"), since "books" is part of the noun phrase "fifteen assigned books." The new salient information could even be "finished fifteen assigned books," since these words together constitute a verb phrase ending with, and containing, the word "book" ("What has Mary done?"). In fact, since "books" is the last word of the sentence, everything in the sentence could be taken to be new and salient ("What happened?"). Of course, in actual contexts it becomes clearer what is and what is not new and salient information. Ultimately, the context in which an utterance is uttered, together with the assumptions that the speaker makes about the hearer's knowledge, usually detenxiines the degrees of informational saliency for each word and phrase in a sentence. Speakers, however, can also choose to downplay or play up the information saliency or importance of a word or phrase and ignore aspects of the context or what they assume the hearer to know and not know already. This is part of how speakers actively create or manipulate contexts, rather than just simply respond to them. Of course, if speakers take this too far, they can end up saying things that sound odd and "out of context." In a given context, even a function word's information might become important, and then the function word would have a greater degree of stress. For example, consider the context below: 2 A: Did Mary shoot her husband? B: No, she shot YOUR husband! In this context, the information carried by "your" is unpredictable, new, and salient. Thus, it gets stressed (in fact, it gets extra stress because it is contrastive -yours not hers - and surprising). In fact, in 2B, given its context (2A), it will be the focus of the intonation unit. When speakers want to contrast or emphasize something, they can use extra stress (marked by more dramatic pitch changes and/or loudness) - this is sometimes called "emphatic stress." 6.5 Lines Each small spurt out of which speech is composed usually has one salient piece of new information in it that serves as the focus of the intonation contour on the spurt (e.g. "shocking" in dialogue 1 and "your", in dialogue 2). There is often a pause, slight hesitation, or slight break in tempo after each spurt. Speaking metaphorically, we can think of the mind as functioning like the eye (Chafe 1980, 1994). To take an example, consider a large piece of information that I want to communicate to you, such as what happened on my summer vacation. This information is stored in my head (in my long-term memory). When I want to speak about my summer vacation, my "mind's eye" (the active attention of my consciousness) can only focus on one small piece of the overall information about my summer vacation at a time. Analogously, when my eye looks at a large scene, a landscape or a painting for example, it can only focus or fixate on one fairly small piece of visual information at a time. The eye rapidly moves over the whole scene, stopping and starting here and there, one small focus or fixation at a rime (watch someone's eye as they look over a picture, a page of print, or at a scene in the world). The "mind's eye" also focuses on one fairly small piece of information at a time, encodes it into language, and puts it out of the mouth as a small spurt of speech. Each small chunk in speech represents one such focus of the mind's eye, and usually contains only one main piece of salient information. Such chunks (what I have heretofore been referring to as "spurts") have sometimes been called "idea units" when people want to stress their informational function, and "tone units" when people want to stress their intonational properties (Chafe 1979, 1980, 1994; Halliday 1989). I will refer to them here, for reasons that will become apparent later, as 'Tines" (Gee 1986). To see lines operating, consider the example below, taken from the opening of a story told by a seven-year-old African-American girl (we saw some of these lines at the outset of this chapter). Each line is numbered separately. Within each one, the word or phrase with the most stress and carrying the major pitch movement (i.e. the focus of the intonation contour), and which carries the new and most salient information, is underlined (in cases where more than one word is underlined, the last word in the phrase was where the pitch glide ocarrred and I am judging from context how much of the phrase is salient information): Processing and organizing language 107 3a last yesterday 3b when my father 3c in the morning 3d an' he . . . 3e there was a hook 3f on the top of the stairway 3g an' my father was pickin me up ("pick up" is verb + particle pair, a single lexical unit whose parts can be separated; the pitch glide starts on "pick" 3h an' I got stuck on the hook 3i up there 3j an' I hadn't had breakfast 3k he wouldn't take me down ("take down" is also a verb + particle pair) 31 until I finished all my breakfast 3m cause I didn't like oatmeal either Notice that each underlined word or phrase (minus its function words, which are necessary glue to hold the phrase together) contains new information. The first line (3a) tells us when the events of the story happened (in this child's language "last yesterday" means "in the recent past"). The second line (3b) introduces the father, a major character in the story to follow. The third line (3c) tells us when the first event of the story (getting stuck on a hook) took place. The fourth line (3d) is a speech dysfluency showing us the child planning what to say (all speech has such dys-fluencies). The fifth line (3e) introduces the hook; the sixth line (3f) tells us where the hook is. The seventh line (3g) introduces the action that leads to getting stuck. Thanks to having been mentioned previously in 3b, the father is now old information and thus "my father" in 3g has little stress. Therefore, "my father," now being old information, can be part of the line "my father was pickin' me up," which contains only one piece of new information (the action of picking up). The eighth line (3h) gives the result of the previous one, that is, that the narrator gets stuck. The rest of the lines work in the same way, that is, one salient piece of information at a time. Adults, of course, can have somewhat longer lines (thanks to their increased ability to encode the focuses of their consciousness into language), but not all that much longer. Notice, too, that once the child gets going and enough information has been built up (and thus, some of it has become old information), then each line tends to be one clause long. After line 3f all the lines are a single clause, except for 3i. And as the child continues beyond the point I have cited, more and more of her lines are a single clause. Most, but not all, lines in all speech are one clause long, though styles of speaking differ in interesting ways in this regard, with some styles having more single-clause lines than others. 108 Processing and organizing language When readers read written texts, they have to "say" the sentences of the text in their "minds." To do this, they must choose how to break them down into lines (which, thanks to the luxury of saymg-m-the-mind, rather than having to actually produce and say them anew, can be somewhat longer than they would be in actual speech). Such choices are part of "imposing" a meaning (interpretation) on a text and different choices lead to different interpretations. Writers can, to a greater or lesser degree, try to guide this process, but they cannot completely determine it. For example, consider the two sentences below, which I have taken at random from the beginning of a journal article by David Middleton entitled "The social organization of conversational remembering: Experience as individual and collective concerns." I have put slashes between where I, on my first "silent reading," placed line boundaries: My topic is the social organization of remembering / in conversation. My particular concern is to examine / how people deal with experience of the past / as both individually and collectively relevant. (Middleton 1997: 71) I find myself treating "in conversation" as a separate line in the first sentence - perhaps, because remembering can be socially organized in many ways, of which conversation is but one, though the one in which Middleton is interested. The way in which I have parsed the second sentence above into lines treats Middleton's main topic, announced in his first sentence ("the social organization of remembering in conversation," and referred back to by "my particular concern [in this topic] is to examine"), as having two parts: "how people deal with experience of the past" (one line) and "as both individually and collectively relevant" (another line). That is, he is going to deal a) with memory and b) with memory as both an individual and collective phenomena. Note that this bi-partite division is announced in the title of Middleton's article, where the colon separates the two themes. Lines reflect the infonnation structure of a text, whether that text is oral or written. 6.6 Stanzas The information embraced within' a single line of speech is, of course, most often too small to handle all that the speaker wants to say. It is necessary usually to let several focuses of consciousness (which lines represent) scan a body of information larger than a single focus. This is to say that the speaker has larger chunks than single focuses of consciousness in mind, and that several such focuses may constitute a single unitary larger block of information. Consider again the beginning of the young girl's story in the last section. These focuses of consciousness (lines) constituted the opening or setting of Processing and organizing language 109 her story, the background material one needs to know in order to situate and contextualize the main action of the story that follows. That is, these lines constitute a larger unitary block of information (the setting) within the story as a whole. However, within this block of information, there are smaller sub-blocks: the little girl devotes several lines to one topic (namely, getting stuck) and several other lines to another topic (namely, having breakfast). I will call such sets of lines devoted to a single topic, event, image, perspective, or theme a stanza (Gee 1986, 1991; Scollon and Scollon 1981; Hymes 1981, 1996). Below, I lay out the opening of the little girl's story in terms of its lines and stanzas: Setting of story: Stanza 1 (getting stuck): 4a last yesterday 4b when my father 4c in the morning 4d an' he . . . 4e there was a hook 4f on the top of the stairway 4g an' my father was pickin me up 4h an' I got stuck on the hook 4i up there Stanza 2 (having breakfast): 4j an' I hadn't had breakfast 4k he wouldn't take me down 41 until I finished all my breakfast 4m cause I didn't like oatmeal either Each stanza is a group of lines about one important event, happening, or state of affairs at one time and place, or it focuses on a specific character, theme, image, topic, or perspective. When the time, place, character, event, or perspective changes, we get a new stanza. I use the tenn "stanza" here because these units are somewhat like stanzas in poetry. Connected speech is like a set of boxes within boxes. The focuses of consciousness (lines), most of which are single clauses, are grouped together as one larger, unitary body of information, like the setting for a story. This larger body of information is itself composed of stanzas, each one of which takes a single perspective on an event, state of affairs, or character. Presumably this distribution of information has something to do with how the information is stored in the speaker's head, though speakers can actively make decisions about how to group or regroup information as they plan their speech. Processing and organizing language 6.7 Macrostructure Larger pieces of information, like a story about my summer vacation, an argument for higher taxes, or a description of a plan for redistributing wealth, have their own characteristic, higher-level organizations (Labov 1972b; Labov and Waletzky 1967; van Dijk and Rintsch 1980). That is, such large bodies of information have characteristic parts much like the body has parts (the face, trunk, hands, legs, etc.). These parts are the largest parts out of which the body or the information is composed. They each have their own smaller parts (ultimately body parts are composed of skin, bones and muscles, and the parts out of which a body of information is composed are ultimately composed themselves of stanzas and lines). The setting of the child's story we have been discussing is a piece of the larger organization of her story. It is a "body part" of her story. Below, I reprint this child's story as whole. Each larger "body part" of the story is numbered with a Roman numeral and labeled in bold capitals (SETTING, CATALYST, CRISIS, EVALUATION, RESOLUTION, and CODA). These larger "body parts" of the story as a whole can be called its "macrostructure," as opposed to its lines and stanzas which constitute its "microstructure." In order to see the patterning in the little girl's story all the more clearly, I do something a bit different in the way I represent lines and stanzas. I remove from the girl's story the various sorts of speech hesitations and dysfluencies that are part and parcel of all speech (and that tell us something about how planning is going on in the speaker's head). I also place the little girl's lines back into clauses when they are not full clauses (save for "last yesterday" which is a temporal adverb with scope over most of the story). What I have produced here, then, are what I will call idealized lines (Gee 1991). Idealized lines are useful when we are interested in discovering meaningful patterns in people's speech and in getting at their basic themes ahd how they are organized. Using them does not mean that we have totally ignored the more superficial patterns of the actual speech. In fact, we can use hesitations, pauses, dysfluencies, and non-clause lines as indicators of how planning is working, where stanza boundaries exist, and how the speaker views her information at a micro-level. In actual analyses we always shuttle back and forth between the actual lines and idealized lines. I. SETTING Stanza 1 1 Last yesterday in the morning 2 there was a hook on the top of the stairway 3 an' my father was pickin' me up 4 an I got stuck on the hook up there Processing and organizing language 111 Stanza 2 5 an' I hadn't had breakfast 6 he wouldn't take me down 7 until I finished all my breakfast 8 cause I didn't like oatmeal either II. CATALYST Stanza 3 9 an' then my puppy came 10 he was asleep 11 he tried to get up 12 an' he ripped my pants 13 an' he dropped the oatmeal all over him Stanza 4 14 an' my father came 15 an he said "did you eat all the oatmeal?" 16 he said "where's the bowl?" 17 I said "I think the dog took it" 18 "Well I think I'll have t'make another bowl" III. CRISIS Stanza 5 19 an' so I didn't leave till seven 20 an' I took the bus 21 an' my puppy he always be following me 22 my father said "he - you can't go" Stanza 6 23 an' he followed me all the way to the bus stop 24 an' I hadda go all the way back 25 by that time it was seven thirty 26 an' then he kept followin' me back and forth 27 an' I hadda keep comin' back IV. EVALUATION Stanza 7 28 an' he always be followin' me 29 when I go anywhere 30 he wants to go to the store 112 Processing and organizing language 31 an' only he could not go to places where we could go 32 like to the stores he could go 33 but he have to be chained up V. RESOLUTION Stanza 8 34 an' we took him to he emergency 35 an' see what was wrong with him 36 an' he got a shot 37 an' then he was crying Stanza 9 38 an' last yesterday, an' now they put him asleep 39 an' he's still in the hospital 40 an' the doctor said he got a shot because 41 he was nervous about my home that I had VI. CODA Stanza 10 42 an' he could still stay but 43 he thought he wasn't gonna be able to let him go This girl's story has a higher-order structure made up of a SETTING, which sets the scene in terms of time, space, and characters; a CATALYST, which sets a problem; a CRISIS, which builds the problem to the point of requiring a resolution; an EVALUATION, which is material that makes clear why the story is interesting and tellable; a RESOLUTION, which solves the problem set by the story; and a CODA, which closes the story. Each part of the story (except the evaluation and coda) is composed of two stanzas. In some ways this is the structure of all stories, regardless of what culture or age group is telling them. However, there are also aspects of story structure that are specific to one cultural group and not another. For example, devoting a block of information to an evaluation prior to a story's resolution is more common among some African-American (young) children than it is with some other groups of children. Adults tend to spread such evaluation material throughout the story or to place it at the beginning, though African-American adults engage in a good deal of "performance" features, which are a type of evaluation, and tend to use evaluation material to "key" a hearer into the point of the story, rather than to hit them over the head with the point bluntly indicated. Of course, such cultural information is Processing and organizing language 113 never true in any very exclusive way: there are many varieties of African-American culture, as there are of any culture (and some African-Americans are in no variety of African-culture, but in some other variety of culture or cultures). And other groups do similar or overlapping sorts of things. Another aspect of this story that is more specific to African-American culture, though also in a non-exclusive way, is the large amount of parallelism found in the way language is patterned within the stanzas. Note, to take one example of many, how stanza 3 says "an' then my puppy came" and then gives four things about the puppy, and then stanza 4 says "an' my father came" and then says four things (all of them speech) about the humans involved. This parallel treatment of the father and the puppy forces the hearer to see the story as, in part, about the conflict between the puppy as a young and exuberant creature and the adult world (home and father) as a place of order and discipline. As a seven-year-old child, the teller of the story is herself caught in the conflict between her own urges to go free and her duty to go to school and ultimately enter the adult world. Notice that the part of the story labeled evaluation makes clear that the essential problem with the puppy is that he wants to freely go places where he cannot go, just as, we may assume, a child often wants to go where she is not allowed to go and must go where she doesn't want to go. In line 21, the child says "My puppy he always be following me," and repeats this in the evaluation. This "naked be" is a form in African-American Vernacular English that means an action is habitual (regularly happens). Here it indicates that the puppy's urge to follow and go with the girl is not just a once or sometime thing, but a regular and recurrent event that follows from the nature of the puppy. It is a problem that must be resolved. The resolution of the conflict between the puppy and the adult world takes place at a hospital where a doctor (an adult) gives the puppy a shot and puts him to "sleep." Thus, the adult world dictates that in the conflict between home and puppy, the adult norms must win. The child is working through her own very real conflicts as to why she can't have her puppy and, at a deeper level, why she must be socialized into the adult world of order, duty, and discipline (by the way, the hook in the first stanza is just a dramatic device - the child is simply trying to say that her parents require discipline in the home; she is not, by any means, accusing anyone of mistreatment - for a fuller analysis of this story, see Gee (1985). The girl may also mean in Stanza 2 that the father would not get her down until she agreed to go finish her breakfast). This, in fact, is the basic function of narrative: narrative is the way we make deep sense of problems that bother us. Linguists and psychologists have proposed many other approaches to the higher-order structure of stories and other connected sorts of language (exposition, argument, description). But they all agree that such connected blocks of information are stored in the mind in terms of various "body parts" and that, in telling or writing such information, we often organize 114 Processing and organizing language the information in terms of these parts, though of course we can actively rearrange the information as we produce it and we often discover structure in information as we produce it. 6.8 Macro-lines So far I have used a young child's story as my source of examples of lines and stanzas. Lines and stanzas are often quite easy to find in children's language. With adults, complex syntactic structures within and across sentences sometimes make it harder to find the boundaries of lines and stanzas. Adults sometimes use the syntactic resources of their language to get lines and stanzas to integrate tightly with each other, to meld rather smoothly together. Indeed, in such language the beginning of a stanza is often constructed to link back to the last stanza and the end of the stanza to link forward to the next, with the "heart" of the stanza in the middle. And, of course, adults often have much more complex language than children. It is often said that, in speech, there are no such things as "sentences," that the sentence as a linguistic unit is a creature of writing only. I do not believe this is true. What is true is that sentences in speech are much more loosely constructed, much less tightly packaged or integrated, than in writing. None the less, people often use the syntactic resources of English to tie together two or more lines into something akin to a sentence. I will call these "sentences" of speech, macro-lines, referring to what we have so far called "lines" (i.e. international units, idea units, tone groups) as "micro-lines" when I need to distinguish the two. Let me give an example of what I mean by "macro-lines." The example is part of a much longer stretch of speech from a woman in her twenties suffering from schizophrenia. As part of a battery of tests, this woman (who is, like many schizophrenics, poor and not well educated) was placed in a small room with a doctor in a white coat and told to talk freely for a set amount of time, the doctor giving her no responses or "feedback cues" the whole time. This "language sample" was used to judge whether she showed any communication disorders connected with her mental state. Not surprisingly (given the limitations of collecting data in this way) the doctors (with little sophistication in linguistics) concluded the woman's text was "disturbed" and not fully coherent. In fact, I have argued elsewhere (Gee 1991) that the text is wonderfully coherent and a typical, if striking, example of human narrative sense making. Below I reprint just the first two stanzas of this young woman's long series of narratives. Below, each unit on a numbered line (e.g. "la" and "lb") is a micro-line. I include unfinished (cut off) micro-lines as separate micro-lines. I underline the focus of each micro-line. Each unit that has a Processing and organizing language 115 single number (e.g. "1" or "2") is a macro-line (thus, la and lb together constitute a macro-line): Stanza 1 (Play in thunderstorms) la Well when I was little lb the most exciting thing that we used to do is 2a There used to be thunderstorms 2b on the beach that we lived on 3 And we walked down to meet the thunderstorms 4a And we'd turn around and run home 4b running away from the 4c running away from the thunderstorms Stanza 2 (Play in waves from storms) 5a That was most exciting 5b one of tire most exciting times we ever had 5c was doing things like that 6 Besides having like - 7a When there was hurricanes or storms on the ocean 7b The waves 7c they would get really big 8 And we'd go down and play in the waves when they got big Consider stanza 1 (the grammatical details to follow in the next few paragraphs are not important in and of themselves - the point is simply to ask oneself how various intonation units or micro-lines are related to each other). Line la is a zo/ieri-clause that is syntactically subordinated to lb as its main clause. So la and lb together constitute a sentence. Lines 2a and 2b are clearly part of one sentence, since 2b is an argument of the verb ("to be") in 2รก. Line 3 is a two clause sentence ("we walked down" and "to meet the thunderstorms") that has been said as a single intonation unit (micro-line). Line 4b is an incomplete micro-line that is said completely in 4c. Line 4c is a participial clause (an -ing clause) that is subordinated to 4a as its main clause. Now turn to stanza 2. Line 5a is an incomplete micro-line. Line 5b is the subject of the predicate in 5c, the two together making up a single sentence (5b contains the phrase "one of the most exciting times" and the relative clause "we ever had"). Line 6 is a false start that does not get continued. Line 7a is a irf-ien-clause that is subordinated to 7b/c as its main clause. In 7b, the speaker has made "the waves," the subject of the sentence "The waves would get so big," a separate micro-line and then repeated this subject as a pronoun in the full sentence in the next micro-line ("they would get really big"). This pattern, common in speech, is called "left dislocation." Line 8 is a single sentence with two clauses in it 116 Processing and organizing language ("we'd go down and play in the waves" and "when they got big"). The speaker could have chosen to say this sentence as two micro-lines (intonation units), rather than one. In many oral texts, it is possible, then, to identify "sentences" (macro-lines) by asking how various micro-lines (intonation units) are syntactically connected to each other, though the connection may be rather loose. In any case, the whole point of macro-lines is get the analyst to think about how syntax is used to stitch intonation units (micro-lines) together. Let me give one more example of macro-lines. My example comes from the first formal meeting of a project sponsored by an educational research institute. The meeting was attended by a researcher from the institute, several undergraduate and graduate students and research assistants, six elementary school teachers, a university professor, and two curriculum consultants. The purpose of the meeting was to start a joint insritute-university-schools project on teaching history in elementary schools in the town in which the meeting was being held. The text below comes from the opening remarks of the researcher from the institute who was leading the meeting and the project (for a full analysis, see Gee 1993): la I'm sort of taking up a part of lb coordinating this project lc bringing the two schools together Id and trying to organize le well what we're going to do in these meetings 2a what it means 2b for teachers and researchers and historians and curriculum people 2c to come on and try to organize a team 2d and students interested in history and other things 2e to try to organize a team to get a piece of curriculum 2f essentially up and running and working in the schools Line lb is, of course, the object of the preposition in la. Lines lc and Id are coordinate clauses (two clauses connected by "and") that are subordinated to the main clause in la/b. Line le represents the complement of the verb "organize" in Id (note that thanks to "well" it is only quite loosely integrated with Id). All of 2 is a recast of le. Since all of 2 can also be seen as an appositive on the sentence in 1,1 and 2 could just as easily be seen as one macro-line and all labeled "1." Line 2b is part of the material that goes with the verb "means." Line 2c is a predicate (verb phrase) whose subject is in 2b. Line 2d is also loosely understandable as a clause conjoined to "a team" ("to organize a team and students interested in . . ."), and, thus, along with "team," it is a direct object to the verb "organize" in 2c. Line 2e recasts and adds to 2c. Finally, 2f is a complement to the verb "to get" in 2e (the Processing and organizing language 117 syntax here is: "to get (verb) the curriculum (object) up and running and working (complement)"). This is a perfect example of how loosely integrated sentences often are in speech. None the less, the syntactic resources of the language are used to link micro-lines together and thereby to indicate some clues as to how the hearer can integrate and link up information across intonation units (micro-lines). In many respects the speaker often discovers or modifies some of these links as she is speaking. For various reasons, having to do with personality and social and institutional relationships, it turns out the speaker of the text above did not want to be the person responsible for running meetings in the future or even the rest of this first meeting. Thus, having said that she is trying to organize "well what we're going to do in these meetings," she, then, recasts this throughout all of 2 as trying to organize not meetings, but "what it means" for all the participants to "try to organize" (themselves as) a team to get certain work done. Of course, "what it means" does not really fit semantically with the verb "organize" in Id, despite the fact that it is recasting, and, thus, loosely taking on the role of the direct object of this verb in le. This is a good example of how syntax, meaning, and organization are an emergent phenomena "on line" as we speak and interact with each other in real time. There is a good deal more in the details of this text (e.g. "takingup a part of coordinating this project," rather than just "coordinating this project," or "try to organize a team," rather than "organize a team") through which we could uncover the workings of individual, social, and institutional factors, or which we could relate to what we may know or suspect about such factors from other sources of evidence. 6.9 Tools of inquiry Lines, macro-lines, stanzas, and macrostructure are important because they represent how speakers marry structure and meaning. They show us how speakers carve up or organize their meanings. At the same time, the way in which we analysts break up a text in terms of these units represents our hypothesis about how meaning is shaped in the text. It depicts our analysis of the patterning of meaning in the text. As such, these units are also our tools of inquiry. We ask ourselves where we think lines, macro-lines, stanzas, and macro-structural units exist in the text, based on intonational, syntactic, and discourse features in the language we are analyzing, and what we know about the speaker's possible meanings, from whatever other sources (e.g. the larger context, other texts, interviews, ethnographic information, etc.). We make these structural decisions based partly on our emerging ideas about the overall themes and meaning of the text. We then use the 118 Processing and organizing language structures (e.g. lines and stanzas) that are emerging in our analysis to look more deeply into the text and make new guesses about themes and meaning. We may come to think that some of the units we have demarcated are wrong, based on a deeper inspection of the intonation, syntax, and discourse features of the text, as well as on the basis of the deeper meanings we are coming to believe and argue that the text has. In the end, a line and stanza representation of a text, like the one given for the seven-year-old's story about her puppy (pp. 110-12), simultaneously serves two functions. First, it represents what we believe are the patterns in terms of which the speaker has shaped her meanings "on line" as she spoke. Second, it represents a picture of our analysis, that is, of the meanings we are attributing to the text. As analysts, we must tie back to this representation all the situated meanings, themes, images, perspectives, and cultural models we are attributing to the text and its context. 7 An example of discourse analysis 7.1 Interview data as an example This chapter will deal with data in an attempt to exemplify some of the tools of inquiry discussed in this book. As I pointed out in Chapter 5, actual discourse analyses will rarely, if ever, fully realize the ideal model sketched there. Real analyses concentrate more on some of the building tasks we have discussed than on others; they use some tools of inquiry more thoroughly than they do others. Since discourse analysis, like all science, is a social enterprise, we hope and trust the gaps in our own work will be filled in by others. In tliis chapter, I do not attempt any full discourse analysis. Furthermore, I do not want to suggest that there is any "lock step" method to be followed in doing a discourse analysis. Thus, I use data here simply to give some examples relevant to a number of points raised in earlier chapters. I hope that my remarks are suggestive in helping others to think about their own data. The data I use here comes from extended interviews with middle-school teenagers conducted by my research team. Our interviews take a specific form. In the first part, we ask teenagers questions about their lives, homes, communities, interests, and schools. We call this the "life part" of the interview. In the second part, the teens are asked to offer more "academic-like" explanations and opinions about societal issues such as racism and sexism. We call this the "society part" of the interview. In addition, we "shadow" the teenagers in their lives in school, at home, in their communities, and with their peer groups, as well as collect data about those schools, homes, and communities. Each teenager is interviewed by a different research assistant on our project who is familiar with the teenager and his or her environment. The teens all view the interviewer as a school-based (indeed, college-based) person. And, in fact, we are interested in whether, and how, each teenager will accommodate to this identity. We have also interviewed, in a similar way, some of the teenagers' teachers and some university academics to see how they talk about similar issues.