Text Production, Ch Five Part 2

2. GRAMMAR MADE OPERATIONAL  

    2.1 GRAMMAR designates a vast, elaborated corpus of conventions about language forms and formats.¹ [The derivation of the word from Greek ‘grammatikos’, meaning ‘of letters’, suggests how strongly the notion was originally allied with writing rather than with speech (cf. V.3.1; Robinson, 1976: 174f).] Together with punctuation and spelling, grammar has been a traditionally focal concern of writing programs. Proficiency in these three areas is the conventional (though misguided) basis for judgements of literacy (I.2.8). Yet despite this obvious social relevance, the fundamental operation of these three sub-systems of language has widely remained obscure. They have been generally acquired more via force of example (dinned in) than via explicit, workable principles. Surveys indicate that traditional grammar instruction does not seem effective for today’s students (cf. Braddock, Lloyd-Jones, & Schoer, 1963; Postman & Weingartner, 1966; Memering, 1978; Mellon, 1981). But a less traditional approach might have a more significant effect on writing skills (cf. Kolin, 1981).

   2.2 Whereas spelling attracted little research until recently, and punctuation even less, grammar was the dominant concern of American ‘linguistics’ for decades (cf. II.1; II.3). However, the main priority of this research was not to improve literacy education. School grammar was decried by linguists (e.g. Roberts, 1958) as inaccurate, biased, and subjective. The surveys of usage that linguists undertook (e.g. Leonard [Ed.], 1932; Kurath, 1949) dispelled some myths about the language Americans really used, which was not always appreciated by officialdom. Meanwhile, the ‘linguistic’ grammars pursued an entirely different intention from school grammar. The descriptive, formalized approach gave little heed to the decision-making involved in the active use of grammar within communication. By discounting evaluation, linguistics largely lost sight of communicative motives for selecting grammatical options.

   2.3 Composition teachers were caught in yet another dilemma. They could either retain school grammar despite documented ineffectiveness and linguists’ censure; or, they could go in search of a new grammar not found either in tradition or in linguistic research. [RECENT NOTE 2004. I am finally undertaking the second of these in book form. Stay tuned.] Neither recourse was attractive, and forced compromises usually ensued. Some schools phased out grammar instruction; others tried teaching linguistics itself instead; and still others {233} adopted textbooks with an ornamental sprinkling of fashionable linguistic terminology (“kernels,” “right-blanching/left-branching sentences”,, etc.) for an otherwise traditional framework. Predictably, these tactics did little to improve the active use of grammar.

   2.4 The ideal qualities of a learners’ grammar are not hard to specify. First, it should be accurate, should reflect the organization of good prose by skilled writers. Second, it should be workable, i.e., should be stated and learned in such a way that the average student, regardless of background, can acquire and use it. Though hardly abstruse, these two standards are frequently not met by instructional materials that embark on a jumbled to define grammatical categories partly in terms of content, and partly in terms of structure. Picture a naive, unskilled write, trying to find out what “subjects” and “predicates” are. Here’s some obtuse advice from widely-used textbooks:

The subject tells who or what the sentence is about, and the predicate tells something about the subject. (Mills, 1979: 2)

A subject group ordinarily names persons, things, or ideas that perform the action or exist in the state described by the predicate. (1982: 117)

A subject is a noun or noun equivalent that performs an action or is in a particular state of being; it usually appears before the verb and determines the number (singular or plural) of the Verb. The verb signifies the action or state of being of the subject. The predicate is the verb and all the words related to it. (Corder, 1979: 463)

We are aware of the Noun-like quality of the subjects and the Verb-like quality of the predicate, whether or not we can explain them . If you are asked to divide a sentence into two parts, you will invariably divide it between “subject” and “predicate” (Moody, 1981: 310)

 The subject is ] whatever verb agrees with in person and number. [The] predicate [is] whatever follows the subject. (Williams, 1981b: 213ff)

 For all their diversity, these definitions have one thing in common: they hardly help a naive student locate the “subject” and “predicate” in a sentence. The content-based approach (e.g. Mills, Broderick) is too vague. “What the sentence is about” could be any of its content. not just what’s expressed in the “subject”; nor is “the predicate” the only thing that can “tell something about the subject”— whatever participial modifiers or relative clauses can too. The structure-based approach (Corder, Moody, Williams) is less vague, but only for people who already command a secure knowledge of grammatical terms, e.g., who can pick out “whatever the verb agrees with in person and number.” Besides, a “noun” doesn’t “perform an action,” but only {234} identifies the agent referred to by the noun. The implication that subject and predicate form two neat blocks in a row, as Moody and Williams appear to suggest, is not even true of common, simple sentences (cf. V.2.21). The “we”-people who “are aware of the noun-like quality of the subject and the verb-like quality of the predicate”, and the “you”-people who “invariably divide” a sentence “between subject and predicate,” are far more likely composition teachers than students (cf. V.2.6).

   2.5 The naive student who needs to identify a sentence is no better served. Again, there is a forced choice between the forbiddingly technical and the unworkably vague:  

Defined by form or pattern, a sentence is a basic unit of language, a communication in words, having as its core at least one independent finite verb with its subject. In addition to being a basic unit, the sentence is a natural one. It nearly always contains two pieces of information the listener is conditioned to expect from it: who or what is involved, and what does he, she, or it do or feel. [...] We speak of sentences in general as complete units, capable of standing alone without the support of supplementary comment. (Willson, 1980: 29)  

Either the students must know how to identify the “independent finite verb with its subject.” Or, they can puzzle over poorly definable notions like “a communication in words,” “a natural unit” (the really natural unit is the “stretch of text,” III.2.28), “pieces of information,” and “supplementary comment”—all of which can apply to utterances not formatted as sentences, e.g. fragments (V.2.32-37). Such definitions take the “sentence” as a given, rather than as one among several possible language formats that needs to be constructed and recognized. Therefore, students get no clear tactics to distinguish between a sentence and a non-sentence punctuated as one.

    2.6 The merits of individual textbooks are not the issue here. The point is that they all rely on terms quite natural to English teachers, but not at all transparent to the unskilled writers we are supposed to be helping. These books are addressed not to their real users, but to the people who decide what textbooks to buy (VI.3.26)—teachers who share the same ways of thinking as the textbook authors. Quite conceivably, many students can’t handle grammar precisely because textbooks expound its basic notions in terms students can’t understand or apply, like a ladder you can’t climb because the lower rungs are missing. Many writing handbooks don’t even attempt to define “subject,” “predicate,” or “sentence,” as if everybody must already have mastered these subtle notions—which is at present not even true for middle-class white students, much less for minorities. Or, textbook authors feel that such matters should be explained with condescension (cf. criticism in Lanham, 1974). In this light, traditional grammar instruction has to be ineffectual (V.2. 1).

     2.7 An accurate and workable grammar for the average learner needs a {235} different design. Grammatical categories must be made transparent enough that contemporary students can handle them without the formal and metaphysical expertise presupposed by most textbooks and reference works. If, as mentalist linguistics claims, speakers of English have acquired and “internalized” a grammar, grammatical categories can be expounded and understood in terms of what someone does with the language. Student speech manifests a considerable store of linguistic knowledge. The question is how to tap that knowledge and consciously apply it to the tasks of writing.

     2.8 These premises stipulate the design of a grammar not provided either by linguistics or by traditional schooling. The grammar has a stated purpose: supporting the goals of literacy education. Hence, the grammar must be tailored to the user. The naive student experienced in casual talk, but not in writing. Plainly, such a grammar could not be designed from some abstract “theory of competence” intended as a vehicle for “pure research.” Indeed, such theories fail exactly because they lack any recourse to application, and hence have no way to define their own object of inquiry. Nor could such a grammar be a vehicle for English teachers to converse with their colleagues via obscure, or condescending edification of the naive students that inhabit average classrooms. This approach defines its application, but not adequately enough to be a useful instructional tool.

    2.9 The first step is to explore how much grammatical knowledge is enacted by average speakers of English. A major problem posed by the English language is the fuzziness of word classes. A word like ‘last’, for example, can belong to any of the four classes of content words (nouns, verbs, adjectives, adverbs) without changing its form. Whereas many languages reshape a word to fit each class, English has dropped off most inflections and now shows many grammatical distinctions by patterns of phrase linearization. Thus, the technique is not to classify the words themselves, but the positions they occupy in utterances. Many linguists have remarked on the ease with which even young children form grammatical utterance patterns long before  they have any contact with school grammar.

   2.10 I shall describe one probe that illustrates these issues. A test was administered to about 150 school children from Gainesville, ranging in age from 8 to 17 years; later, a replication was done among rural school children in Maclenny, Florida, a tiny town near Jacksonville.” The following grid was presented:  

(302) The _____, _____, _____, _____, _____ man

 The _____, _____, _____, _____, _____ lady

 The _____, _____, _____, _____, _____ airplane

 The _____, _____, _____, _____, _____ word

 The _____, _____, _____, _____, _____ city  

{236} The children simply were told to “fill the blanks with adjectives that sound good.” I assumed that most of the children would be grammatically naive and have no explicit definition for the notion of “adjective.” Their teachers confirmed that grammar instruction had been uncommon, or at least not emphasized. However, the position between article and noun would, I reasoned, naturally be filled by most English speakers with words that are, ipso facto, adjectives.

   2.11 The task actually implies two other tasks we didn’t announce. The announced task was the easiest: to make sure that every blank contains an adjective. Here, most of the children were quite successful. Their spelling, which I had found on other tests to be, erm, uncertain (VI.40), occasionally obscured the data, e.g., whether ‘the experience man’ is a solecism in spelling or in grammar. Significantly, even made-up words were formed to look like adjectives, usually via their endings: ‘the bobochous lady’, ‘the mustacheous man’, and the ‘businesy’, ‘unordinating’, ‘love-obtained’, and ‘inexistencing’ ‘city’. The evidence was overwhelming that fourth-graders with known literacy problems nonetheless demonstrated an operational understanding of the grammatical category “adjective.” Less than 1 % of the data contradicted this finding, for example:  

(303) The big black ugly dub with ploted air city

(304) The big wite loud zoomed with anger airplane  

In fact, the author of (304) went back and inserted ‘plane’ in the proper slot after ‘loud’.

 2.12 A harder task was to make the adjectives plus their noun form a coherent configuration. The vast majority of the data also met this criterion. Unexpectedly (for me anyway), the children typically displayed negative attitudes about men, words, and cities, vs. positive ones about ladies and airplanes, e.g.:  

(305) The bow-legged, emptyheaded, knock-kneed, stupnosed, extremely grotesque man

The slim, willowy, graceful, courteous, attractive lady

The gigantic, metallic, bird-like, shining, silver airplane

The obscene, dirty, bad-mouth, fowl, 4-letter word

The bustling, overcrowded, grimy, crime-filled, smelly city  

Especially the ‘word’ was a target of aggression, perhaps because of teachers’ adverse comments on the utterances and papers of children. In acrimonious disregard for its holy origin proclaimed in the New Testament (John I: 1), the ‘word’ appeared as ‘misspelled’, ‘illegible’, ‘illiterate’, uncalled-for’, ‘silly’, ‘boring’, ‘trite’, ‘overused’, ‘tedious’, ‘exasperating’, ‘rambling’, ‘confusing’, ‘incomprehensible’, ‘outlandish’, and ‘meaningless’. Small wonder if children learn little from language instruction that so berates them. {237}

   2.13 The age of the children affected their tolerance for doubtful coherence. The youngest children ably created configurations that would do credit to a surrealist:  

(306) The big medal fat mean bad airplane

(307) The fat ghosty scarry long squared city

(308) The long hard to say blue fat little word

(309) The strong, surprising, shaking, long, scarey word

(310) The great pulsating, secreteing throbbing pink word  

Contradictory configurations were also found, though more rarely:  

(311) The handsome monstrous intelligent colossal stupid man

(312) The very little kind of big word

(313) The German mispelled mispronounced scientific French word  

The fact that the children had a firmer grip on the grammatical category “adjective” than on the demands for coherence seems important. It suggests that the tradition of defining grammatical notions in terms of content (V.2.4) may be both unnecessary and confusing. Children are more likely to learn what an “adjective” is from their language experience than from definitions like: “in general, adjectives modify, or in some way change the meaning of, nouns and pronouns,” ¹ [Pronouns? show me! – Maybe you mean ‘the living I’ or ‘the new you’?, which no student writer would ever write!], where “nouns” (you guessed it) “name persons, places, things, or ideas” (Moody, 1981: 346, 31 1).

   2.14 The hardest task was of course to get the five adjectives into an acceptable order— an issue still far from explained in theory (cf. IV.2.74). Adjective sequences, like most recursions in language, are normally no longer than three (IV.2.13, 70). As expected, the children were often uncertain about the best order. Sometimes they would draw arrows on the paper to rearrange their sequences. One child improved (314) by making it into (314a):  

(314) The blue and white American fast huge airplane

(314a) The huge fast blue and white American airplane  

Not all such changes brought a clear improvement, however:  

(315) The shiny metalic wonderous large flying airplane

(315a) The shiny wondrous large metalic machine airplane  

Even my research group and I had a hard time figuring out the best order for some sequences. It was much easier to rearrange (316) than (317):  

(316) The cargo, war, green, powerful, long airplane

(316a) The long, powerful, green war cargo airplane

(317) The large, noisy, crowded, fast white airplane

(317a) The large, fast, white, noisy, crowded airplane  

We had to break cases like (317) down into shorter versions we could agree

{238} about: ‘the large white airplane’, ‘the large, fast, white airplane’, and finally, ‘the large, fast, white, noisy, crowded airplane’ (where ‘noisy’ and ‘crowded’ could be interchanged).

 2.15 Later, we did a comparison study with whole sentences containing only three blanks for adjectives in each of two positions, e.g.:  

(318) The man used _____, _____ _____words

(319) The lady lives in the _____, _____ _____ city.  

The instructions were the same as before (V.2.10); yet this condition improved both the coherence and the order of the results. The children produced very few incoherent statements, and almost no disordered sequences. Greater coherence was encouraged by using the format of normal complete statements. Greater order came from limiting recursion to the customary threes (cf. IV.2.13, 70; IV3.37). These two factors elicited language knowledge that the less natural format of the earlier tests had failed to tap. Grammatical knowledge as such is generally fuzzy, but comes into sharp perspective when applied to a specific task such as formatting an everyday statement (cf. I.4.6).

    2.16 This straightforward demonstration may have useful implications. The further grammar is removed from natural communication, the more likely ordinary people are to lose control of it. Instruction that drills students on grammar alone is therefore unnecessarily arduous, and its transfer over to spontaneous writing is doubtful. The specialized drill-setting apparently interferes with the grammatical system in discourse processing, because the loss of context makes the grammar much less well-defined. Hence, learners need to reorganize their system of grammatical routines at the expense of those already devised and practiced in everyday communication. This reorganizing in turn draws a heavy load, so that performance is further degraded below one’s actual knowledge of the language (cf. I.2.8.2; I.3.3, 25). It is unrealistic to hope the effects of such drills will carry over when processing reverts to normal. More likely, the whole stressful reorganization of the system will be set aside as soon as the drill is over, especially by non-traditional learners (I.2.23.10). In contrast, familiar, natural uses help people make the best of their grammatical knowledge.

    2.17 Further research of this kind may refute a commonplace assumption of conventional linguistics since Saussure: that the “grammar of a language” is, or can be treated as, an idealized, uniform abstraction with clear borders. The postulate that language can be separated from language use (II.1.2; II.3.7) may prove to be so fundamentally misconceived that a large part of linguistic research cannot support the theoretical claims based on it. Grammar appears to be a body of fuzzy knowledge that assumes clear forms when put to everyday uses, but tapers off when applied to less familiar and controllable patterns (e.g., five modifiers in a row). If so, grammar doesn’t exist, except as a derived construct of grammarians, until it is used {239} in some language event (I.4.6, 10, 14; IV.1.1). As speakers of English, students have a kind of grammatical knowledge that is more secure, but less explicit, than the decontextualized grammar of textbooks or teachers. The former kind of knowledge is the only realistic basis for attaining the latter. Writing instruction need not (and cannot) teach English grammar from scratch; but it can and should identify from student papers those categories which everyday language experience has not clarified sufficiently for the demands of writing. Then, we can seek to harness that same experience such that these categories are manifested in simple actions the students already know how to perform. Of course, these actions, when done as exercises, depart from their natural spontaneous context of communication; but at least, they are closely derivable from such contexts and sufficiently habituated via daily experience that they should not encumber or distort processing as much as old-style grammar drills.

     2.18 The sentence is one hell of a problematic category. Many statements of everyday speech are not expressly formatted as a sentence: either something is missing, or their boundaries are fuzzy (cf. III.2.28, 30; IV3.45; V3.2, 8). [The determination of linguistics to make the sentence the obvious structural framework upon which all formal definitions depended has its counterpart in pedagogy, e.g. in a textbook nonsensicially opening with: “you speak in sentences” (Mills, 1979: 1}.¹ [See now my paper “Sentence first, verdict afterwards: On the long career of the sentence”, WORD 50, 1999, 1-31] .Here is a portion of a transcribed passage spoken by a freshman student. As usual, a short pause is indicated with “/,” and a long one with (IV.2.25); a hyphen marks where a word was broken off.  

(320) first I rented a steam cleaner // and um / I steam clean-ned / steam cleaned the roof by attaching a hose / and plugging the machine in the wall / didn’t take much and // um I can / you can th- the machine’s light enough that you can put it up on the roof // and uh // you walk around with the spray handle to the pressure cleaner or a steam cleaner whatever // and um / walk slowly back and forth on the on the roof / um / applying the pressure / you know to the tile on the roof to get off the loose dirt and paint  

The entire passage could be structurally analyzed as one long sentence linked with ‘and’ and ‘that’, and broken more by pauses and shifts than by distinct clausal boundaries. The content and intonation of the recording, however, suggest a division into several statements: renting; cleaning with the rented equipment; getting to the roof; moving around; taking off impurities. The phrase ‘didn’t take much’ reads like the predicate whose subject is ‘plugging the machine in the wall’; on tape, it comes across as a general comment on making the machines work. Thus, the phrasing hovers in a grammatical limbo, but is conceptually and pragmatically well-defined.

    2.19 Unlike speech, writing normally conforms to “the conventional expectation that a sentence begins with a capital letter and ends with a period, “ as Corder (1979: 203) prosaically observes; but he circumspectly goes on: {240}  

Fortunately, no one has successfully dictated what shall go in between. To be sure, most English sentences do contain a subject and a verb, and most make some kind of statement that can be read and understood by itself. But [ ...] both the content of your sentences and the form of your sentences are yours to determine.  

Most textbooks are less enlightened than Corder’s on this point—so much so that a paradox has arisen. Skilled writers have far more freedom from worrying about grammatically proper sentences than do unskilled writers, who could profit more if they worried less. Sentence errors are a frequent occasion for excessively punitive grading practices (cf. Harris, 1981). The sharp reactions toward “sentence errors” (fragments in particular) reported in Hairston’s (1981: 797) survey of professional people in 63 occupations are probably an after-effect of composition classes. In real life, such “errors” are most apt to become acute when needed as leverage in social or institutional conflicts (cf. I.2.1 1; V.2.3 1; VI.1. 3).

    2.20 Social factors make it all the more urgent to experiment with easy methods for tapping everyday language experience to define the sentence. One way to tell what something is, the mentalists had argued, was to “transform” it. Though such operations need not constitute normal text production, they could be selectively performed on already produced samples for strategic purposes (cf. II.3.15ff—the basic idea behind “sentence-combining” (II.3.17-47). The operations can be straightforwardly classed as addition, deletion, and conversion (cf. van Dijk, 1972, 1977). In a procedural methodology for grammar, learners could carry out one or more such operations and contrast the results with the original sample. The contrast should make the desired grammatical category transparent as a language event. Though the setting differs from everyday communication, the operations should be ones people commonly perform during the latter (V2.17).

   2.21 Imagine now some naive students learning to recognize some stretch of text as a sentence (cf. Beaugrande, 1982a). First, they are given two interlocking definitions: (1) “every sentence must have at least one independent clause”; and (2) “every clause must have at least one subject and predicate.” But instead of explaining these notions in terms of formal structures or vague content (V2.4f), I ask the students to do something they already know how to. They see simple examples such as:  

(321) Her father owns the market.

(322) Sometimes one of the dogs runs away.  

and convert them according to this strategy: {241}  

Make up a ‘who/what question about the statement made in the sentence. The PREDICATE of that sentence is all the words you used again in the ‘who/what’ question; the SUBJECT is the rest.  

The strategy is immediately and easily applied to simple examples. Students devising a ‘who’/’what’ question about the main statement conveyed by the sentence should most naturally replace the subject of the sentence (or of its main clause) with the ‘who’ or ‘what’. The question is then placed alongside the original sentence and the words that got used again are counted up. The others are reckoned as the subject.¹ [Sentence-modifiers that qualify the whole utterance act are not strictly attached to either subject nor predicate, e.g. ‘in my opinion’, ‘frankly’, etc. This nebulous status is shown by the ability of these modifiers to be either used or emitted in the ‘who/what’ question. Note also that the tendency to replace the subject with the ‘who’ or ‘what’ is only a preference easily modified in a context of expectations about what is interesting or important (cf. R. Posner, 1980).] For the two examples, the students obtain:  

(321) Her father owns the market.

(321a) Who owns the market?

(321b) predicate = owns the market subject = her father

(322) Sometimes one of the dogs runs away.

(322a) What sometimes runs away?

(322b) predicate = sometimes runs away subject = one of the dogs  

As we see from (322), having a part of the predicate before the subject is no obstacle: the question conversion moves such a part out of first position. Methods where sentences are sliced like salami, as in immediate constituent analysis (cf. II.1.13), or where subject and predicate are defined by position (cf. V.2.4), render these formats unnecessarily hard.

    2.22 This operation is rehearsed by the class, and possible mix-ups are clarified. The question must ask ‘who’ or ‘what’ do/does/did something, not ‘who’ or ‘what’ something is/was done to. The question words can be only ‘who’ or ‘what’, not ‘when’, ‘where’, ‘which’, etc. And, if there is a choice, the shortest question should be used, e.g. (323a), not (323b):  

(323) My friend at the college gave out the figures.

(323a) Who gave out the figures?

(323b) Who at the college gave out the figures?  

 Pre-tests and post-tests were conducted to measure the usefulness of this technique. On each test, the students had ten sentences in which they were to underline all parts of the subject once, and all parts of the predicate twice. On some test runs, I started by asking the students to indicate whether or not they “felt confident about grammar and sentences.” Many{242} answers did not agree with the evidence of the pre-tests. Some students expressed low confidence, but performed quite well; some thought themselves better prepared than the results suggested. Apparently, estimates of one’s own grammatical expertise reflect attitudes formed during positive or negative experiences in prior English classes—and are thus not very reliable predictors.

    2.23 I ran each session within a single 50-minute class period. Ten minutes went to each test, so that 30 remained for the entire presentation and the exercises based on a worksheet of samples. The pre-test was given before any grammatical explanations. Each sentence was scored as either right or wrong (since the definitions of “subject” and “predicate” depended on each other), so that the paper would score between 0 and 10 errors. In my first trials, the post-test showed improvements over 500%. To see if these results could be obtained by the average teacher, a group of teaching assistants at the University of Florida replicated the method with a larger sample. Four assistants gave the tests to a total of 96 students, mostly freshmen, again within a single class period. The total errors came to 602 on the pre-test and 121 on the post-test: an improvement of 495%, almost five times better. If we set aside the “special services” section for students with known literacy problems, the improvement was 898%. 73 students improved, 13 stayed the same, and 2 got worse. The statistical significance of our results is too obvious to need calculating.

    2.24 An equally simple teaching module is used to define the clause. To see whether a clause is independent, and hence a useable core for a sentence (V.2.21), a similar conversion operation is carried out. This time, the strategy is: Make up a ‘yes/no’ question about the statement made in the clause. If the question works, the clause is INDEPENDENT As it happens, only an independent clause makes a question that can sensibly be answered with ‘yes’ or ‘no’. For instance, speakers of English would make (324) into (324a), but hardly  (325) into (325a):  

(324) He works too hard.

(324a) Does he work too hard?

(325) Because he works too hard.

(325a) Because does he work too hard?  

 This technique catches the dependent clause punctuated as a sentence, e.g. (325)—a widespread type of fragment (V.2.36). Comma splices, on the other hand, make two yes/no questions (V.2.42).

    2.25 These techniques specify the minimum requirements for a clause or sentence, but do not yet describe the more elaborate formats of “compound” and “complex” sentences. Those require an understanding of how junctives (or “linking words,” as I say in class to sound less technical) ¹ [Traditionally, all junctives are imprecisely called “conjunctions,” despite the inclusion of disjunction and opposition. In IV.2.52, I offered a more exact terminology.] {243} “link up clauses-another formidable issue for naive writers. The most compact approach I found is to start with the definition:  

 A clause not preceded by any linking word is independent.  

Samples at this stage include simple sentences like (326), and complex sentences like (327) with the independent clause after the dependent one. Clauses are separated by commas to make the clause boundary obvious (a later punctuation module along the lines presented in IV.3 takes care of comma placement).  

 (326) The lady over there is taking her time

 (327) Whenever he works so hard, he looks exhausted.  

In an early pilot, I included a module specifically devoted to telling apart “simple,” “compound,” “complex,” or “compound-complex” sentences. Students disliked this terminology, and had trouble using it. They said it sounded confusing and technical, especially “complex,” which reminded some of “complicated” and others of a neurosis, and “compound-complex,” which seems to short-circuit opposite notions (or to compound the neurosis). I soon found that the terms could be discarded by working entirely with “independent vs. dependent clauses”—the constituents which make up these sentence types. The approach became more compact and, at the same time, dropped some undesirable terms from the traditional repertory (cf. V.2.28f).

    2.26 Next, the students memorize only the list of common linking words that do not create dependencies, rather than the much longer list of those that do. In current usage, the list contains four principal items, ‘and’, or’, ‘but’, and ‘so’; ‘for’ is fading away, and ‘either’, ‘neither’, and ‘nor’ go along with ‘or’, but have some peculiar effects. I therefore concentrate on the main list of four. The next definition is: Any linking word on the list of four makes the clause after it equal to the clause before it. If the clause to be classified is at the start of the sentence, the definition can’t be applied. If there is an independent clause before, as in (328), or a dependent clause before, as in (329) — the “yes/no” question helping to identify which is the case—the linking word makes the next clause into another of the same. This approach blocks the misconception that any clause introduced by ‘and’, ‘or’, or ‘but’ is necessarily independent.  

(328) The moon rises among the stars, and the town sinks into silence. [independent]

(329) When the moon rises, and the stars come out, the town sinks into silence. [dependent]  

The students examine a series of samples, circling independent clauses and underlining dependent ones. Common sentence patterns up to three clauses are practised. {244}

    2.27 These brief modules should at least suffice suggest to principles of the method (cf. Beaugrande, 1982a). First, the students learn to work with intricate grammatical categories by doing simple, familiar actions on language samples. Expertise in traditional grammar is not required. Second, attention centers on the categories most essential in going from speech to writing. Third, these operations are meticulously streamlined by the instructor before reaching the student. The trade-off at stake here—if the student’s task is to be easy, the instructor must work all the harder, and vice versa (VI.3.5)—must not be resolved at the student’s expense. Since the average teacher is already overloaded, textbooks and computer adjuncts must be developed that reflect this streamlining; and hide-bound publishers must be persuaded to distribute them (cf. VI.3.26). For all three reasons, simplicity, compactness, and brevity become top standards for designing instructional materials and methods. Increasing numbers of students need to get a grasp on grammar. But those with the greatest needs are the least likely to profit from drills in diagramming sentences and “naming parts of speech.” As we saw from the findings on adjectives (V.2.10-16), most word classes are anchored in everyday language experience. We can tap that experience just enough to clarify the issues that trouble untrained writers.

    2.28 Admittedly, instructional design from this standpoint is arduous. There is no principled way to know if the current version is the most compact one. Later, it may emerge that the work of two operations can be done with just one. For example, I devised the following means for finding the agreeing verb of the predicate, since agreement is a difficult issue for naive writers with certain dialects. Two steps, sometimes three, will do the job. First, insert a so-called denial word into the statement, namely, ‘doesn’t/don’t’, ‘didn’t’, or ‘won’t’. Second, the “agreeing verb” of the original statement is the one now located after the denial word.¹ [ Black English has some inverted forms where the denial word comes at the start of the utterance, but usually, I believe (and I’m just judging from my African American fiends), only if the subject is ‘nobody’, ‘no one’, or the like.] One example was:  

(330) Our boss wants to call a meeting.

(330a) Our boss doesn’t want to call a meeting.  

If you can’t insert a “denial word,” a third step is done. Insert ‘not’ or ‘-nt’, and the agreeing verb is the one before this insertion; or, if the statement already has a ‘not’ or ‘-n’t’, the verb is the one before that, and no insertion is necessary. This technique exploits the peculiar English constraint of attaching negation in declarative sentences to the finite verb via an auxiliary that is either already present or else added for the purpose. By using contractions as denial words, I preclude confusion with any other positions that ‘not’ by itself can occupy in a sentence.

    2.29 In another module, I was struggling with tense¹ [Greenbaum and Taylor’s (1981: 173) “paper- correcting” research found that 5 of the 27 English teachers they surveyed changed ‘was’ to ‘is’ in ‘I told them my daughter was now a doctor’, under the gormless impression that ‘now’ required the present tense.]. I tested insertions of time expressions like ‘right now’, ‘next year’, etc. This classic textbook {245} problem comes from the asymmetry between tense and time, so that both content and structure can be confusing or conflicting (cf. V.2.4f). [On the confusion between thoughts and sentences, cf. I.2.16.7; II.3.42; IV.2.18, 32f.] You explain past, present, and future tense in terms of when something happens, and then face about and admit that the present tense can be for both past or future time (Quirk et al., 1972: 86). Or, you stress changes in verb forms, and get stuck on verbs like ‘set’, ‘put’, ‘fit’, and so on, which don’t alter their forms for tense. I finally realized that the whole dilemma could be less effortfully resolved with the denial-word insertion I had previously devised for working on verb agreement. The “past” is the tense that takes ‘didn’t’; the present takes ‘doesn’t/don’t’, and the future takes ‘won’t’. These words are either already present or can be mentally inserted to decide the tense of any finite verb in a sentence. Sample sentences become manageable as soon as the student supplies any context. Some samples were:  

(331) Our kids set the table

(331a) Our kids don’t set the table. [present]

(332) Our team just set a new record.

(332a) Our team didn’t just set a new record. past]

(333) The sun will set before 7:00.

(333a) The sun won’t set before 7:00. [ future]  

No lectures on the nature of time and perspective are needed; nor is an unchanging verb form an embarrassing exception. A speaker of English naturally gets the desired results, simply because language experience guides the action that in turn defines the category (V.2.16f). In tests run by Barbara Stephenson and Michael McCoy, 45 college students made less than half as many errors on recognizing agreeing verbs and tenses after learning this technique than they had made before. 35 students improved, 5 stayed the same, and 5 got worse.

      2.30 Easy operations for identifying sentences and clauses are a good basis for clarifying and resolving common sentence problems, such as fragments, splices, and run-ons. These formats are popularly called “sentence errors,” though this classification has seldom been convincingly defended. The fragment is to be eschewed because it “doesn’t really say anything” and “leaves the reader wanting something more” (Glazier, 1981: 67, 44); and because it “does not communicate a grammatically complete thought” (Moody, 1981: 363). The comma splice is condemned on the grounds that “it makes two complete statements that should not be run together in one sentence” (Glazier, 1981: 44). These claims overlook the sensible communicative motives that lead to fragments and splices. The patterns are straightforward by-products of the asymmetry between statements and sentences {246} (V.2.18, 37). For the writer, the fragment does convey a “complete thought” of some kind (V.2.32ff); and the comma splice invariably joins components the writer feels belong together in one statement (V2.39ff). Besides, these patterns may be repaired with mechanical changes (e.g., replacing a comma with a semicolon) that have no effect on “thoughts.” Recently, textbooks have attained enough tolerance to discuss cases where fragments are motivated (Corder, 1979: 23Off; D’Angelo, 1979a: 573ff; Tibbetts & Tibbetts, 1979: 366; Willson, 1980: 201f; Mills, 1981: 241).

     2.31 Again, the composition teacher is caught in a dilemma between coercive rules and free expression, between language attitudes and language realities (cf. I.3.13). Severe persecution of fragments and splices intimidates students about producing any sentences; or alienates them when they find such formats in competent professional writing and conclude that the teacher is unreasonable or uninformed. In turn, anxieties and loss of motivation degrade writing performance (cf. I.2.10; III.2.12). On the other hand, some readers may be excessively intransigent toward “sentence errors” in important situations, above all when the writer is at a social disadvantage (cf. V.2.19). My own solution to the dilemma is a delicately balanced compromise. I explain to the students that certain “errors” can be used against them in a school or career setting; but my role is to diagnose and assist, not to evaluate and punish. I try to be a chronicler of public attitudes rather than their enforcer (cf. I.2.19; V.3.26). Accordingly, I have been probing the motives, causes, and effects of the troublesome sentence patterns. The sentence is a unit whose demarcation leads processing to predict certain components in standard formats (III.2.28). A “sentence error” is registered if a pattern either lacks some components (fragment), or has a poorly integrated surplus (splices and run-ons). Whether these deviations are disturbing depends on situation, text type, and reader focus. Like most conventions, sentence patterns can be creatively violated, provided the result leads to a convincing, insightful new order (cf. I.3.7, 10).

     2.32 The SENTENCE FRAGMENT can illustrate the approach advocated here (cf. Kline & Memering, 1977; Harris, 1981). Its creative use is common in good writing. A critic who disdained T.S. Eliot’s ‘Rhapsody on a Windy Night’ because it opens (334) and closes (335) with sentence fragments  

(334) Twelve o’clock.

(335) The last twist of the knife.  

would only be ridiculed. The context of the poem, and of many like it, calls for the juxtaposition of brief, fragmentary statements among which ‘all clear relations’ are ‘dissolved’ in ‘a crowd of twisted things’. These motives justify overriding the core-and-adjunct priciple in favor of listing. Prose can also deploy fragments to good effect, such as listing brief impressions or thoughts (Corder, 1979: 231; Willson, 1980: 202): {247}  

(336) There are other things that affect me the same way. Blue-and white striped sheets. Vermouth cassis. Some faded nightgowns which were new in 1959 or 1960. (Joan Didion, “Farewell to the Enchanted City”)

(337) Sam Clark’s Hardware Store. An air of frankly metallic enterprise. Guns and churns and barrels of nails and beautiful shiny butcher knives. (Sinclair Lewis, Main Street)  

Also, fragments can assign heaviness to word groups by making them autonomous stretches, rather than adjuncts of something else. All the resources that normally get distributed over a whole sentence are focused on just a few words. The heavy fragment can be a direct recurrence (IV.2.37) of an element from a preceding sentence (Corder, 1979: 23 1) (or less often, an anticipation, cf. IV2.58):  

(338) Many have never been taught the pleasure and pride in setting standards and then living up to them. Standards! (John Gardner, Excellence)  

Heavy fragments can suggest the importance and style of a telegram, as in the NCTE “tel-a-message”:  

(339) Contacted you last month about membership in National Council of Teachers of English (NCTE). But must reach out to you again. Response to current membership drive tremendous; English/language arts educators joining NCTE in record numbers. Association membership growing. [etc.]  

This quasi-telegram (sent letter rate) signed by the NCTE executive director contains 32 stretches punctuated as sentences: 20 fragments, 10 imperatives whose lack of a subject aligns them with fragments), and 2 complete sentences. The short statements and paragraphs draw attention (cf. IV.2.69; IV3.17) and evoke the hurry-up urgency of telegram situations (e.g., sudden requests for money). The NCTE leadership knows better than most textbook authors how fragments can very well convey “complete statements” or “thoughts” and “really say something” (cf. V.2.5, 30) — and get action.

    2.33 Purposeful sentence fragments usually correspond to perfectly good conceptual chunks (“thought units”) (cf. IV.2.18, 32f; V.2.30; V.3.8). This correspondence also contributes to fragments that report or imitate speech rhythms, since many non-sentence phrasings are uttered as separate units, with a pause before and after (cf. I.4.II.2; III.2.28; IV.2.17; V.2.18; V.3.7). Spontaneous answers to questions are often fragments (cf. Kline & Memering, 1977; D’Angelo, 1979a):  

(340) GRACIE: My poor brother Willie, he was held up last night.

GEORGE: Your brother was held up?

GRACIE: Yeah, by two men.

GEORGE: Where?

GRACIE. All the way home. (Burns, 1980: 70)  

{248} To make every answer a complete sentence (e.g., ‘My poor brother Willie was held up last night all the way home’) would destroy the effectiveness. For similar motives, writers use fragments in a conversational prose style, as in (341) and (342). Only one word in (342) is an actual quote of speech; the rest suggests internal speech (“Erlebte Rede”) of someone on a bad LSD trip trying to figure out what’s going on.  

 (341) I notice Harding is collapsed beside McMurphy and is laughing too. And Scanion from the bottom of the boat. At their own selves as well as at the rest of us. (Ken Kesey, One Flew Over the Cuckoo’s Nest)

(342) Jane in front of his face, a foot away, then way back over there on the sofa, then zooming up again, all of it rocketing back and forth in the hulking heat.”Sandy!”—somebody is in the house looking for him, Hagen? who is it?—seems Babbs wants him in the movie. (Tom Wolfe, The Electric Kool-Aid Acid Test)

Advertizements are fragmented for both reading ease and focused attention (cf. V2.32). Contrast (343) with (343a), or (344) with (344a) (after D’Angelo, 1979a: 571f):  

(343) Each is a masterpiece. Realistic, yet delicate.

(343a) Each is a realistic, yet delicate masterpiece.

(344) I guess that’s what makes a classic a classic. The ability to look completely different depending on how it’s used.

(344a) I guess the ability to look completely different depending on how it’s used is what makes a classic a classic.  

The reader is more likely to focus on the individual pieces of the message, especially if processing is casual (e.g., skimming a newspaper or a magazine). (344) elicits brief suspense with a dummy place-holder (‘that’) to make readers curious (cf. IV.2.63).

    2.34 Speech rhythms and conceptual chunking are typical causes for fragments in student writing. The issue is whether the fragments are motivated, like those in V2.32-33, or accidental and disruptive. We can explain to students why a fragment might be written and leave them to decide by themselves what dangers or benefits are involved in a given case. A naive student naturally marks off as a sentence what would constitute a separate stretch of speech, as in:  

(345) We wandered off looking for something to steal. Little kids’ toys in particular.  

Structurally, the most common fragment is forrnatted as an adjunct whose core is in an adjacent sentence, generally the preceding one (346-348) (Harris, 1981: 177), but sometimes the following one (349) {249}  

(346) You see I’m trying to avoid another scrambled egg breakfast. Basically because I hate them.

(347) Coming off the bench, Junior Mickey Masties performed superbly for the gators. Averaging in the double figures with consistent outside shooting.

(348) I like to wear a pants outfit. Maybe a pair of pleated slacks, a silk shirt, and a jacket to match.

(349) Once an emergency has been established. The trip down to Aigle is quite cheap.  

The fragmentation might result from time lags as conceptual or phrasal chunks are returned by memory search. In post-activation, the chunk comes too late and is expressed after the slot it would normally occupy (III.3.32; IV2.11; V 3.30f). Closely-knit phrasal chunks, such as a dependent clause (346), a long modifier (347), or an appositive list (348), might be processed with just enough delay that their relation to the foregoing structure is blurred and the latter is closed off as a sentence. Another explanation could be in terms of reactivation. The text producer can intentionally restart a suspended statement (cf. IV.2.50, 67; V.3.20f, 30). A fragment that repeats the last word of the sentence before it resembles a reactivation, e.g.:  

(350) We are waiting. Waiting for someone else to solve our energy problems. (D’Angelo, 1979a: 571)  

From the readers’ standpoint, fragments due to reactivation and post-activation would be easy to process via look-back.

    2.35 Some accidental fragments in student writing occur because their format roughly resembles a sentence. A non-agreeing verb form, e.g. a participle, may get confused with a main verb:  

 (351) These objects being the mountains.  

Long or complex adjuncts are easy to mistake for complete sentences:  

(352) Not even an extremely educated man will react positively to a wordy, tiresome opening statement. Because, as a whole, people do not feel ads are that important and will not take the time to read on if their interest is not caught right away.  

The student who wrote (352) was created a fragment with three subject-predicate cores in complicated subordinations. Complex modifiers pose a similar danger:  

(353) Captured in her spell-binding look and knowing every second he stood there risked her seeing him.  

The writer of (353) demonstrated considerable fluency and vocabulary, and made few mechanical errors. But his striving for syntactic variety and complexity led to fragments he wouldn’t have written if he’d been content with a simpler style (cf. II.3.46; V2.41).

    2.36 In V.2.24, I outlined a technique for recognizing independent {250} clauses by converting them into “yes/no” questions. A sentence fragment fails this test, either because it lacks the needed subject or verb, or because it hinges on a dependent linking word. For long modifiers with no verbal participle, both subject and verb would have to be supplied to make a question:  

(345a) Little kids’ toys in particular.

(345b) Did they look for little kids’ toys in particular?  

Modifiers with past participles need a subject and finite verb:  

 (353a) Captured in her spell-binding look.

 (353b) Was he captured in her spell-binding look?  

A modifier whose present participle was misjudged as a finite verb forces us to warn against dropping off the ‘-ing’ for the question:  

(351) These objects being the mountains.

(351a) Are these objects the mountains?  

Dependent clauses have both subject and verb, but don’t make sensible questions (V.2.24), e.g.: 

(352a) Because people do not feel ads are that important

(352b) Because don’t people feel ads are that important?

.Naturally, these conversions become harder when the fragments are quite complex, as in (352) and (353). Students whose desire for complexity engenders numerous fragments should simplify their style. Or, they can practice doing sentence-combining on the patterns of sentence + fragment, or fragment + sentence, until complexity becomes more tractable (cf. II.3.47) Realistic samples can readily be gathered from student papers.

     2.37 Sentence fragments are a less serious problem if student writers understand causes, motives, checking procedures, and remedies. Fragments naturally arise from asymmetry between statements and sentences (V2.30). Conceptual and pragmatic chunks need not emerge as complete syntactic chunks (V.2.18, 33). Thus, nothing is accomplished by telling students that fragments are devoid of meaning and purpose (V.2.30). Instead, the students should appreciate what tendencies lead to fragments, and what conditions justify or discourage their use in good writing. Then, workable techniques for finding and fixing fragments would be more efficient and reliable. Anxiety and inhibition are lower because the errors aren’t denounced or penalized, rather than explained (V.2.31). This approach can work, and has worked in several pilots I designed. Like any “error,” fragments are easiest to control if treated as natural events that can be detected and repaired whenever they might put the writer at a disadvantage. 2.38 Another by-product of chunking can be called the SPLICE: the joining of two independent clauses without a junctive or {251} appropriate punctuation in between. I coined this non-traditional term to subsume both simple juxtaposition (sometimes called a “run-on”) and the well-known “comma splice.” Just as a fragment occurs when a conceptual chunk does not map onto at least one independent clause, a splice occurs when such a chunk maps onto two independent clauses without proper means to combine them inside a sentence. The outcome reflects the fuzzy borders of spoken statements. Shaughnessy (1977: 18) suspects “a psychological resistance to the period-perhaps because it imposes an end on a unit the writer usually had difficulty beginning.” Such anxiety would drain away attention from sentence boundaries that don’t match statement boundaries. The plain splice has nothing at all between the clauses, e.g., after ‘starts’ in:  

(354) All basketballs game are started with a jump ball unless a technical foul is called before it starts then the team that shots the technical takes the ball out at the beginning of the game.  

Leaving a major pause boundary unmarked is fairly conspicuous and can create an unmanageably large chunk (cf. V.2.44). My samples suggest that plain splices are more common when a writer is young or inexperienced, or is trying to build a confusingly complex sentence like (354). Otherwise, fragments or comma splices, which at least mark pauses and boundaries, appear in greater proportions in both professional and student writing. Nor can I find any textbook willing to tolerate plain splices.

   2.39 COMMA SPLICES, where a comma alone joins independent clauses, are much more frequent because they are less noticeable and disruptive. The comma indicates a pause and identifies the boundary between two main cores (cf. IV.3.6). Moreover, a comma between clauses is standard if one of them is dependent. Naive writers uncertain about clause types would thus be very prone to making comma splices out of closely related statements, e.g.:  

(355) Melyin couldn’t get in, he didn’t look old enough.  

Readers also seem less disturbed. In Hairston’s (1981: 797) survey, comma splices were not judged as serious a failing as fragments (cf. V2.19). Professional writers working for a conversational style produce comma splices writers working for a conversational style produce comma splices freely, e.g.:  

(356) Today marriage is old-fashioned, it’s like getting your spats cleaned. (Burns, 1980: 5)  

Willson’s textbook (1980: 209f) describes comma splices in an “antithesis,” and in a pattern of ‘it isn’t this, it’s that’ as “legitimate”:  

(357) It was more than an annoyance, it was a pang. (Winston Churchill)

(358) To allow the Madhi to enter Khartoum would not merely mean the return of the whole Sudan to barbarism, it would be a menace to the safety of Egypt itself. (Lytton Strachey)  

Though noted writers like Churchill and Strachey presumably command the sentence, most textbooks, even those willing to admit justified fragments (V.2.30), either don’t mention these splices or else condemn them. D’Angelo (1979a: 567) calls the ‘it isn’t this, it’s that’ pattern “unacceptable” (as does Moody, 1981: 372):  

(359) The process isn’t really hard, it just takes patience.  

Whereas Willson (1980: 209) allows lists of three independent clauses—in which listing should apply more than core-and-adjunct — such as: 

(360) The shrubs were leafy, the walks had been carefully raked, and the fountain shone in the sunlight. 

Moody (1981: 372) rejects the same construction and wants a period in the place of the first comma. My colleagues in the English department disagreed whether a pattern with ‘so/such’ in one clause counts as a comma splice, e.g.:  

   (361) I was so mad, I just left.  

The point of contention is whether a ‘that’ must be inserted, or whether it is optional, as it is at the start of an indirect statement (362), and of a relative clause depending on a noun acting as the object of a verb or preposition in that clause (363) (Quirk et al., 1972: 788, 215):  

(362) He said (that) they might be wrong.

(363) This is a man (that) you should know  

There is no obvious reason to insist on the ‘that’ in formats like (361) only, unless we’re nervous about comma splices.

     2.40 As I argued for fragments, comma splices should be remedied in view of their causes and motives. One cause already noted is the relatedness of the two statements. The second statement usually gives support or elaboration to the first, as in these student passages:  

(364) The journalism class was a very good one, we all worked together and got along with the advisor.

(365) The school didn’t financially support the paper, all costs were raised by the journalism class. 

My tapes of students reading their papers aloud showed a shorter pause at the comma than at the period in such cases. Another cause is the confusion between clause-linking junctives vs. adverbials which, though conceptually and pragmatically similar, do not have the same syntactic function. ‘And’ is confused with ‘also’, ‘moreover’, ‘too’, ‘besides’, or ‘in addition’; ‘but’ with ‘however’, ‘only’, ‘still’, ‘nevertheless’, or ‘all the same’; ‘or’ with ‘otherwise’; and ‘so’ with ‘therefore’, ‘thus’, ‘hence’, or ‘consequently’. The result is technically a splice, e.g.: {253}  

(366) Becoming an attorney is often difficult, however, it is a very rewarding occupation.

(367) The letter had to arrive the next day, otherwise it would be too late.

(368) A stage can swallow up one person, therefore, the actor’s objective should be to fill up the entire stage.  

To the grammatically naive student, such a pattern seems unobjectionable. The privileged status of junctives over transition words of comparable meaning is a subtle distinction easy to miss.

    2.41 Like fragments, splices are harder to detect if the student works for complex phrasing (V.2.35). For example, a dependent clause may be put between two independent ones, so that either one of its boundaries or the other is technically a splice:  

(369) They talked for a while on their pasts, even though they had geographic differences they shared many common interests.  

This sentence came from the same student whose elaborate phrasing entrained him in sentence fragments like (353) in V.2.35. Reported speech can also obscure sentence formats. A declarative quotation is usually set off from the main verb by a comma. But usage is unclear about having two such boundaries in one sentence; we find that format in the writings of Lewis Carroll (1960: 100), who was fastidious about punctuation,¹ [He even wrote ‘sha’n’t’, ‘wo’n’t’, etc., with an apostrophe for each position where letters were omitted.] e.g. the Dormouse’s reply:  

(370) “I wasn’t asleep,” it said in a hoarse, feeble voice, “I heard every word you fellows were saying.”  

Yet some textbooks (such as Willson, 1980: 210) class that usage too under comma splices. This diagnosis is less contestable if the writer shifts from direct to indirect quotation. (371) might be more objectionable than (371a):  

(371) “A casual escape,” Keith said, he didn’t see any reason to breed rumors.

(371a) “A casual escape,” Keith said, “I don’t see any reason to breed rumors.”  

   2.42 Ordinary comma splices are easy to detect because, being two independent clauses, they yield two ‘yes/no’ questions (V.2.24). (365) gives us (365a) rather than (365b):  

(365) The school didn’t financially support the paper, all costs were raised by the journalism class.

(365a) Didn’t the school financially support the paper? Were all costs raised by the journalism class?

(365b) Didn’t the school financially support the paper, all costs were raised by the journalism class?  

{254} Or, the components of the comma splice can be recognized as independent clauses because they are not introduced by a linking word (cf. V.2.25).

    2.43 The conceptual and pragmatic relatedness of the two spliced parts counsels against the old schoolroom remedy of replacing the comma with a period. The writer intended the parts to go together in one statement (V.2.30, 38). One student explained, “I felt it wasn’t clear yet, so I added something.” A semicolon would keep the statement unified, but might be noticeably overused by unskilled writers, thanks to its the dual function for pausing and look-ahead (half a comma, half a colon, IV.3.26) (cf. Shaughnessy, 1977: 33f). The most satisfying solution is a junctive that makes explicit the conceptual relationship the writer intended, e.g.:  

(364a) The journalism class was very good, because we all worked together and got along with the advisor.