19 Jul 2007

The College Board is large; it contains multitudes

Submitted by Karl Hagen
This is part 3 of a series analyzing the College Board's view of English grammar. In part 1 I reviewed the general standards for the SAT writing test and argued that they were vague. In part 2, I began to reverse-engineer the grammatical framework used by the test makers and found numerous errors in the explanatory answers that the College Board provides with its practice tests. This installment is mainly intended to discuss inconsistencies in the explanations, but in preparing that discussion, I have run across yet more errors in analysis, which I will document first.

See part 1 for an explanation of how I reference particular questions.

Errors (continued)

Phrases and clauses confused

In SG-5.10.11, "because people have lost faith in the dollar" is called a phrase. It is actually a subordinate clause.

In SG-5.10.14, "Many of the instruments used in early operations of the United States Army Signal Corp" is called a clause. It is, in fact, a noun phrase.

Infinitives said to have tense

Choice (D) involves errors in verb tense and parallelism. The present tense of the infinitive "to originate" is not logical because the action of the sentence occurred in the past and before the action of the main verb, "is believed." (OC-4.10.4)

This is a confusion based on traditional labels for infinitive constructions, in which forms like "to originate" are called the present infinitive and forms like "to have originated" are called the perfect infinitive. (Warriner & Griffith, 1965, p. 153). Like present and past participles, these are forms without tense. A finite form of a verb is one that has tense. Infinitives are, as their name indicates, not finite forms.

Now on to the inconsistencies.


English grammar is taught in many different ways, and depending on the account one follows, there can be vast differences in terminology and analysis from one textbook to another.

From the point of view of a test developer, much of this variety is irrelevant in the day-to-day writing of questions. Different grammar books may disagree about how many tenses English has, but those differences in analysis often have little to do with acceptability judgments. In other words, two people might have widely different explanations of the structure of a sentence but still agree as to whether or not the sentence is acceptable.

To a student struggling to understand an explanation, however, variations in terminology and analysis create significant problems. Most students have negligible exposure to formal grammar, and so an answer key addressed to students should adopt a consistent and straightforward vocabulary to explain the problems. Moreover, the analysis must be consistent within the chosen grammatical framework.

Variant Terminology

The following are instances of inconsistent terminology. These do not constitute actual errors of analysis, but they do risk potential confusion for students because things are not named consistently.

Verb sequences such as "had received" are typically called the past perfect, but this structure is occasionally called the pluperfect (OC-1.7.12, OC-1.10.4, OC-2.6.21, OC-2.6.35) or the "completed past" (SG-6.4.3, OC-2.6.31).

"Main clause" and "independent clause" are used interchangeably, as are "subordinate clause" and "dependent clause." There are too many instances of both to cite, but the independent/dependent pair is used about three times as often as main/subordinate. There is no consistency in the use. In other words, it is not the case that some writers prefer the independent/dependent labels and others the main/subordinate labels. Usage is sometimes mixed within the same question (see, for example, OC-2.6.9).

Different levels of detail

Overall, the explanations use basic technical vocabulary to describe sentence structure. Terms like "subordinate clause" or "gerund" are common. Although that choice does make things more difficult for the student who is completely unfamiliar with grammatical terminology, it is understandable. It is excruciatingly awkward to discuss issues of grammar without a technical vocabulary, and it would be ridiculous to avoid all grammatical terms. Certainly, there are times when a precise technical explanation would only confuse a student, yet the level of detail used varies from question to question in ways that are not particularly helpful.

For example, in SG-2.7.6, "demanding assignments" is called a "phrase consisting of an adjective and a noun" in the explanation for choice E. This wordiness is particularly unaccountable as the same phrase was simply and accurately called a noun phrase elsewhere in the explanation to the same problem.

Some questions refer to verbals, others to participial phrases or infinitive phrases. Verbal is the traditional superordinate term to encompass all the non-finite verb forms, but in the context of these explanations is unnecessary, and in some cases misleading. In questions involving parallelism, for example, the basic principle followed by the College Board can be summarized with the dictum, "use the same grammatical form." Hence we see explanations like the following:

The phrase consisting of an adjective and a noun, "demanding assignments," is not parallel with the verbal phrase "working at part-time jobs." (SG-2.7.6).

But by labeling the participial phrase a "verbal phrase," students may infer that two verbal phrases would be parallel. Yet infinitive phrases are not parallel with participial phrases, although they are both verbal phrases, and when such a mismatch occurs, the more specific terms are used in some questions (SG-6.4.10). In others, "verb phrase" is used for participial phrases (SG-5.10.11). In still others, the nature of the phrases in question is left unspecified (SG-2.7.22; SG-4.6.18).

None of these terms are incorrectly used, but they do not form a consistent method of explanation.

'Main verb' applied inconsistently

For the College Board, the term main verb appears to mean the concatenated string of all verbs (including auxiliaries) in the predicate of a main clause. In other words, verb is used, as in Warriner and Griffith (1965), to refer to a multi-word sequence, and main indicates the verb's position in the main clause.

The requirement that the main verb lie in the main clause is implied throughout the explanations and stated explicitly a few times. For example:

Numismatics, or coin collecting, becoming popular in the United States around 1857, when the replacement of the large cent by the new flying-eagle cent led enthusiasts to start collecting the earlier coin. (OC-5.6.1, text of choice B substituted for underlined section)

The error in the sentence is explained this way:

Choice (B) creates a sentence fragment. The participle "becoming" cannot serve as a main verb. The other verb, "led," is the verb for the subordinating clause beginning with "when" and cannot serve as the main verb. Therefore, a main verb is needed to complete the action of the sentence.

Despite this clear definition, the term main verb is applied several times to the verbs of subordinate clauses (See SG-3.6.34, SG-7.3.3, SG-7.10.12). In most cases, the verb is not explicitly identified as being in a subordinate clause, and so these could be merely instances of error, mistaking the verb in the main clause for one in a subordinate clause, but OC-5.6.19 contains the following explanation: "The participle 'eating' cannot serve as a main verb for the subject of this subordinate clause"

This usage reflects a different way that the term main verb is used in some grammar textbooks, for example Morenberg (2005). Under this definition, the main verb is the lexical verb and any related auxiliary verbs no matter what sort of clause they appear in. Elsewhere in the explanations, the term complete verb is frequently used to mean the same thing.

In a handful of instances, main verb is used in still another way, to mean the lexical verb, that is, a verb that is not an auxiliary verb:

The past-tense helping verb 'did' appropriately expresses an action that was completed in the past, and the adverb "not" is properly placed before the main verb "make." (OC-4.5.23)

The passive voice of the verb is appropriately formed with the past participle of a main verb ("spread") and a form of "be." (OC-4.5.26)

This usage follows that of Quirk et al. (1985), and is one that has been widely adopted by other grammar books.

The inconsistency between this use of the term and the definition used elsewhere in the explanations is particularly evident in OC-4.5.23, where the explanation is deeply flawed, perhaps resulting from the author's confusion over the different meanings of main verb. The sentence in question is the following:

Unlike her best friend Margie, making the varsity soccer team as a freshman, Jill did not make the team until her junior year. (OC-4.5.23)

And this is the explanation of the error:

The error in this sentence occurs at (B), where the improper use of a gerund ("making") results in a sentence fragment. There is no main verb to indicate that Margie is performing the action described in the first part of the sentence. The pronoun "who," combined with the past-tense verb "made," is needed ("Unlike…Margie, who made…")

This explanation appears to be the result of formulaic thinking (no main verb = sentence fragment), but that logic depends on the main verb being in the main clause. The sentence is not a fragment, as it has a well-formed main clause ("Jill did not make..."). The problem comes from the a structure that appears to contain two opening modifiers, the second of which has no clear attachment to the rest of the sentence. There is also no reason to call making a gerund. In the traditional scheme, it makes better sense analyzed as a participle since there's no plausible analysis under which the phrase "making...freshman" fills a slot prototypically occupied by a noun phrase. In this case, the writer of the explanation seems to be combining two incompatible definitions of main verb without any awareness of how they differ.

Tense of modal verbs described inconsistently

The conditional present perfect tense of "could have maintained" is not consistent with the simple present tense of the subordinate clause ("are"). (OC-5.6.12)

Elsewhere in the explanations, could is either explicitly called a past-tense verb (for example, SG-1.10.6) or its tense goes unnamed (for example, SG-4.6.28).

Many grammar books do not specify the tense of modal auxiliaries. Some, indeed, explicitly claim that modal auxiliaries have no tense (Celce-Murcia & Larsen-Freeman, 1999, p. 137), so a certain vagueness about the tense of these verbs would be understandable, but the explanations show no consistent practice.

Similar confusion is found with would.

The exact tense of would is often avoided. For example, "would seek" is said to have "the appropriate tense" for a particular sentence (SG-4.6.12). It is explicitly called a past-tense form in SG-6.4.21, OC-5.6.26, OC-6.10.14.

Elsewhere, would is often labeled a conditional, without specifying a particular tense (OC-4.10.1). But the wording in other explanations implies that "conditional" is a tense:

It avoids the error of the original by using the correct verb tense to indicate a condition ("would spend") (SG-7.10.7).

And later in the same explanation, "The future 'will' should be the conditional 'would.'"

In SG-3.10.1, "would insure" is labeled a "conditional future."
In OC-4.10.1, "would have been governed" is called "present perfect tense."

Other tenses described inconsistently

The College Board normally describes tense in a twelve-term system. So we frequently find terms like the "past progressive tense" or the "present perfect progressive tense." Relative to this scheme, there are a number of incorrect labels of tense (although the tense is correct if verb tense is analyzed the way linguists do).

In SG-4.6.3, "is turning" is called present tense. By the College Board's standards, it should be the "present progressive tense." The same problem also occurs in SG-5.4.7

In SG-4.6.11, "were...being disrupted" is called the past tense. By the College Board's standards, it is the past progressive tense.

In SG-4.10.1, "would have ensured" and "had ensured" are both called past tense. According to the scheme the College Board uses, they should be called conditional past perfect and past perfect respectively.

In the next installment, I will consider one of the most pervasive inconsistencies of all in these explanations: the treatment of constituency, particularly what the College Board means by a "phrase."