The underutilized semicolon

The semicolon, long in decline, may be in renewal, but it remains rare in its central use: replacing a comma and coordinating conjunction to unite two independent clauses. Current pedagogy tells us that the semicolon represents a bigger break than the conjunction and a smaller break than period. Could this account have caused the semicolon's decline, as we increasingly relied on written accounts for rules of mechanics? Can writers really classify the degree of logical connection of clauses into as many as three distinct categories? Most people have a hard enough time with only two. Plain-language exponents of short sentences instruct to break off a sentence when it becomes "too" long. Does anyone expect ordinary people or even skilled writers to divide degree of connection between connected thoughts into three neat, objective compartments? Even if the compartments existed in an objective sense, our mental powers may be too weak to distinguish them.

Guided by finding an interval greater than a comma but less than a period, writers reasonably choose not to apply voluntarily an intermediate standard as ephemeral as clear-and-convincing evidence, a standard so unintuitive people apply it only when serving on a jury under court order. If the semicolon had vitality before people relied so heavily on formal explication, people must have used a different criterion, either in addition to or in place of the intermediate-pause account. The real vitality of the semicolon — the best way to use it — depends on the need for a neutral connector rather than an intermediate pause because to show a connection between ideas facially we rely primarily on two coordinating conjunctions, but and and. But combines logical conjunction with contrariety. When you encounter but, the writer implies that hearing the first clause would ordinarily decrease your expectation of hearing the second. It seems natural to assume equivalence between and and logical conjunction: but stripped of its contrariety. So assumed, and is the generic connector, but the negative-expectation connector, and no positive-expectation connector exists. And may not be quite as positive as but is negative, but and is positive, not generic. Otherwise we would have to choose more carefully between and and but whenever but applies.

The semicolon is the language's neutral connector. Use it to connect ideas directly without implying either positive or negative expectation of what follows.

(Next entry I'll discuss the main circumstance in which expectation neutrality is important.)

The semicolon and expectation’s equipoise

(Second and final entry in the semicolon series.)

Freer semicolon usage would help stem the promiscuous spread of artificial connectors, an assault on Concision. It would also contribute to Clarity by simplifying the representation of expectation neutrality, since and and but represent positive and negative expectancy of the following independent clause; but you may wonder when expectations are ever precisely neutral. The most important circumstance where a writer wants to represent the expectation as neutral occurs when the first clause expresses both a positive expectation and a negative expectation; the writer wants to avoid seeming to prejudge which is the stronger.

Oliver Wendell Holmes Jr. used the semicolon with great precision in his famous epigram, where the first clause expressed positive and negative expectations:

We do not inquire what the legislature meant; we ask only what the statute means.

Holmes renounces a purposivist approach to statutory interpretation but intends the implication to elicit surprise, greatest when a correlation obtains neither positively nor negatively. Holmes can create a sense of surprise because the renunciation carries two implications or expectancies: to interpret the text instead of interpreting the legislature's will or to renounce interpretation altogether. Logically, Holmes might have said instead: "We do not inquire what the legislature meant; we ask only what the legislature should have said."

A legal writer who uses semicolons with unusual effectiveness is Judge Richard A. Posner in The Economic Analysis of Law:

Generally, specific performance (ordering the party who breaks his contract to perform, on penalty of being held in contempt of court if he does not) will not be ordered as a remedy for breach of contract; the promisee will have to make do with damages, as suggested in the Holmes dictum quoted earlier.

(The Economic Analysis of Law, p. 117 [reference to Holmes coincidental].)

Here you can see the same pattern as in the quote from Holmes. The policy-of-denying-specific-performance clause activates two opposed expectancies: that another remedy will replace it or, alternatively, that none will; a different remedy or no remedy at all.

Revising Orwell: Initial Conjunctions as Light Adverbs

Using the adverbial connective however (see "Banish Stock Transitional Expressions") is not an outright stylistic weakness but a device useful, even necessary, in speech, hence in fiction, a device, though often excessive, that persists because of over close identification of competent speech and writing. Characteristics of speech demanding artificial connectors are its short sentences and explicit signposts, so nonfiction writing calling for short sentences and signposts can warrant using adverbial connectives. Writing requires short sentences and signposts when it demands hyperclarity, as for legal briefs in our overcrowded courts.

Some statistics help in grasping the size of sentence-length differences. Many essayists write sentences averaging 22 to 28 words. Bryan Garner recommends sentences in legal briefs average 20 words. Plain-language advocates typically call for sentences averaging 15 to 20 words. Many excellent fiction writers seem to average around 18 words per sentence. (Hat tip to StevenBerlinJohnson.com, where you can find some surprising statistics and observations on writers' sentence lengths.)

If clarity demands only moderately shorter sentences, as in brief writing, occasionally starting a sentence with a coordinating conjunction, such as and or but, is a less distorting method of artificial connection than adverbial connectives, such as however, moreover, nevertheless, and in addition, terms depicting relations between referring clauses, not referenced objects. To appreciate why initial conjunctions are less distorting than adverbial connectives, it helps to understand the connective paradox: when they begin sentences, these so-called conjunctions function as adverbial connectives. A coordinating conjunction at the beginning of a sentence doesn't tie the two sentences together, sometimes advertised: it doesn't function like a semicolon to establish a relationship stronger than separate sentences but weaker than conjoined coordinate clauses. The reader, having read the sentence complete, can't undo the perception, so but no longer functions conjunctively. But takes the meaning of however and comes to exemplify the same lexical category. But always has an adverbial component; otherwise, it couldn't assert contrariety. Dropping the conjunctive component by using but at a sentence's beginning leaves only a light-adverb remainder.

When exceptional circumstances don't limit sentence length, the writer can be more precise by striking the initial conjunction, combining the sentences to turn the initial conjunction into an ordinary conjunction, or rewriting. George Orwell disagreed. Here's an example including an initial conjunction, but, and an adverbial connective (or stock transitional phrase), on the other hand, from George Orwell's essay "Politics and the English Language":

A newly invented metaphor assists thought by evoking a visual image, while on the other hand a metaphor which is technically 'dead' (e. g. iron resolution) has in effect reverted to being an ordinary word and can generally be used without loss of vividness. But in between these two classes there is a huge dump of worn-out metaphors which have lost all evocative power and are merely used because they save people the trouble of inventing phrases for themselves.

I would rewrite the passage by eliminating the transitional phrase, using but to connect clauses within one sentence, and replacing the compound predicate reverted... and ... used with a simple predicate and adjective phrase, thus:

A newly invented metaphor assists thought by evoking a visual image, while a metaphor which is technically 'dead' (e. g. iron resolution) has in effect reverted to being an ordinary word, generally usable without loss of vividness, but in between these two classes there is a huge dump of worn-out metaphors which have lost all evocative power and are merely used because they save people the trouble of inventing phrases for themselves.

The transitional phrase on the other hand helped convey that Orwell mentioned the newly invented and the dead as opposite statuses for metaphors. Joining the clause about the worn-out metaphors with the previous sentence, containing the contrast, makes the same point about opposite statuses without using the transitional phrase on the other hand. The clause about the worn-out metaphors introduces the opposites by locating worn-out metaphors between them. The sentence is longer than any of Orwell's, but the passage gains precision using fewer words.

The new topic/stress principle: Topic is concrete, stress abstract

Recall the sequence (concrete to abstract, near to far):

conversation – informal prose – formal prose – poetry

Formal prose is a style of writing that evolved—primarily to serve abstract matters, where clarity is the central virtue—to exploit the specific virtues of written discourse. It is far-mode clarity that is most prized in this style; near-mode is essential but subordinate. To subordinate near to far, formal prose uses a characteristic device at various structural levels: new matter is introduced in far mode and developed in near mode.

At the sentence level, this is accomplished by an application of topic/stress segmentation: the stress—which introduces important new information—is abstract; the topic—which recapitulates old information—is concrete.

Definitions and principles

In formal prose (which includes the most effective legal-brief writing), the topic (usually the sentence subject) announces what the sentence is about, often through association with previous information. The stress position (I’ll continue using the term despite the technical misnomer) refers to the last word or words before a period, colon, semicolon, and sometimes a dash; it contains important new information. (We know this about topic and stress mainly due to the work of Joseph Williams and George Gopen.)

Construal-level theory links concreteness to psychological proximity and abstractness to psychological distance. In sentence processing, the topic’s position is near and the stress’s position is far, and formal prose not only honors the topic and stress positions, their contents are typically concrete and abstract respectively, to correspond with their near and far locations in the sentence. New information is first presented abstractly in the stress position and then developed concretely by being recapitulated in a more specific form in the topics of subsequent sentences.

A counter-example?

The reader expects the topic to be concrete and the stress abstract, and each receives greatest emphasis when they satisfy the expectation. This observation answers a counterexample offered by Wayne Schiess, purporting to show that the topic is more important than the stress:

To me, number one emphasizes President Bush more. 

(1) President Bush made mistakes.

(2) Mistakes were made by President Bush.

Addressing Wayne’s argument fills a lacuna in topic/stress theory: what determines the stress-position’s size? Although “President Bush” constitutes a terminal phrase in number two, that phrase—referencing a near-mode concrete particular rather than a far-mode disposition—isn’t well suited to receive stress. The reader expands the stress position to encompass a suitable abstraction, which it finds in the sentence’s predicate, “were made,” which the sentence emphasizes.

(Generalizations like this new topic/stress principle are often best used to sharpen intuition rather than to replace it. I don’t think it would have occurred to me that number two emphasizes the predicate without its aid, but once I’ve applied the principle, the intuition perseveres.)

Rewriting the “writing rules”

The new topic/stress principle grounds, consolidates, and corrects several established “writing rules.”

Avoid nominalization is one-sided over-reaction; nominalization creates far-mode abstractions, commonly suitable in the stress but not, such as to supplement an excessively abstract verb, in the topic’s vicinity.

Favor agents as subjects is a simplistic rendition of formal-prose’s preference for concrete topics.

Concrete examples should precede new abstractions describes a practice in the (informal, near-mode) plain style, unsuited for formal prose.

Writers should exploit all punctuation marks: Reflections on misguided campaigns to reduce punctuation types

Omissive punctuation practices on Twitter convince some observers the apostrophe is superfluous, but a recent essay in New Republic news magazine dismissed the rumors of imminent apostrophe extinction. The author, however, wasn’t exactly happy about the apostrophe’s endurance, cautioning only that leaving it out will continue to “look funny” in formal writing. 

Considering the absence of consensus about which to exterminate, the impulse to kill some disliked punctuation is surprisingly strong. George Orwell thought the semicolon unnecessary and resolved to avoid it; some writers demand abolition of the dash; competent legal writers have opposed hyphenation of many compound adjectives, claiming they’re unsightly and often unnecessary; it’s been claimed that the comma was invented or perpetuated because publishers benefit from their supposedly unnecessary consumption of space; and I’ve condemned the virgule.

This false economy of punctuation types isn’t rational, since an abundance of types for marking syntactic distinctions means greater ease for readers. We have few punctuation types not because of the uselessness of marking additional syntactic distinctions but because of the difficulties of socially coordinating on a new punctuation type, which must be commonly understood and highly practiced. Contrived punctuation doesn’t stick: it requires too great an adoption rate before it gathers momentum. Emoticons (like the smiley) may seem an exception, but they prove the rule: they augment lexicon rather than representing syntax. Lexicon accrues more rapidly than punctuation types not only because we need many more semantic distinctions but also because we more readily learn the meaning of new semantic than syntactic signs. (Acquiring words is in the genes, but writing and its punctuation are parts of culture.)

Why do some writers wish for fewer types of punctuation? One reason is that overuse and misuse often turn them against the whole type. I formed a prejudice against the virgule (/) when enduring an employer who expressed any conjoined or disjoined legal claims with a weaseling and/or. Exposure to some bad freestyle blogging incites people against the dash, and the irritating misuse of the apostrophe to create plurals of names could be enough to alienate some writers.

Another source of animus against punctuation variety is that writers often don’t understand how punctuation helps readers. On discovering that they can understand text without a certain punctuation type, they conclude that it’s unnecessary (the main argument against the apostrophe in the New Republic piece), but punctuation serves primarily to enhance cognitive fluency, not to render text intelligible or disambiguate expressions.

Finally, pedagogy’s emphasis on signaling literacy and competence through correct grammar and mechanics leads some writers to view punctuation marks as occasions for error rather than as promoters of cognitive ease. Fewer distinctions mean less embarrassment.

Using all the available punctuation marks is part of exploiting the full expressive power of written language. But keep in mind what does not follow: if variety (in punctuation types) is a spice of life, heaviness (of punctuation tokens) is a drag.

The colon: When the explanation is more important than what’s explained

Each punctuation mark serves a core function, and usage should follow the core function whenever the rules governing that function are applicable. Disputed Issues has considered the core functions of several punctuation marks: The comma sets off nonrestrictive elements; the semicolon neutrally connects independent clauses; and the dash emphasizes matter tangential in its immediate context. Following the core functions means eschewing rules unrelated to the core function unless the core function is unrelated to the construction. To take the comma, usage guides sometimes state the rule that a comma doesn't set off an adverbial clause at the end of a sentence, but the restrictive - nonrestrictive distinction the writer should apply eviscerates the rule.
The core function is the main function for ordinary discursive text. The colon has a variety of uses, such as exemplification by lists, but the central discursive use of the colon is to substitute for a word like because to create a clause more central than the independent clause to which it would be subordinated. From the opposite end of the grammatical telescope, the colon demotes the independent clause to a parenthetical role.
Here's an example:

Density is audience relative: the optimal density for experts is higher than for novices; but density's audience relativity isn't as great as you might think.

Grammatically, an adverbial clause could substitute for the clause following the colon:

Density is audience relative because the optimal density for experts is higher than for novices; but density's audience relativity isn't as great as you might think.

The colon serves better than the adverb, since the matter in the because clause is more important than what precedes, which only creates a transition through a more general proposition; the more important propositions shouldn't ordinarily be subordinated to the less important. The clause following the colon becomes independent when the colon is substituted, but this happenstance doesn't affect the colon's usage; a subordinate clause can follow the colon, and the independent clause's significance would remain parenthetical.

Constructing sentences for precise emphasis: The fundamental principle of advanced writing

A legal-writing authority advises:

 View your reader as a companionable friend—someone with a warm sense of humor and a love of simple directness. Write like you're actually talking to that friend, but talking with enough leisure to frame your thoughts concisely and interestingly. John R. Trimble, Writing with Style 73 (2d ed. 2000). (HT: Bryan Garner, Usage Tip of the Day, November 12, 2013.)

Some writers hail Trimble’s advice as profound, while others ignore it as meaningless, but I hold it is quite wrong. Legal-brief writing (like other efforts at exerting intellectual influence) differs from conversation not just in degree: influential intellectual writing differs from conversation in its guiding formal virtue. Whereas good conversation is (or seems) spontaneous, good writing is clear.

One way the difference manifests is that competent writers force important new ideas to the sentence’s end. The last word or tight phrase preceding the point of syntactic closure (period, semicolon, or colon) is termed by Joseph M. Williams (Style: Toward Clarity and Grace) the stress position; and according to another student of sentence structure, George D. Gopen (A new approach to legal writing), failure to exploit the stress position is legal-writers’ single greatest formal weakness: out of hundreds of lawyers Gopen has trained, the stress position was properly used by a handful. Proper use of the stress position is at the threshold of competent writing, but misuse of the stress position doesn’t always sound bad. Locating trivia in the stress position produces limp sentences, but often lawyers fill the stress position with misleading substantive language. When a document contains sentences with misleading emphases, readers—due to conflicting cues about what’s important—find the document’s meaning hazy.

The stress position isn’t unique to written English; spoken English sentences end in higher pitch, but in spoken English, stress position is subordinate to nonverbal cues. It is also subordinate to standard word order, which conversation usually follows because reorganized sentences sound contrived, violating the conversational norm favoring spontaneity. Take as an example the previous paragraph’s final sentence, which trades moderate disfluency for high clarity:

When a document contains sentences with misleading emphases, readers—due to conflicting cues about what’s important—find its meaning hazy.

This is too contrived for good conversation; without the engineered word order, we might say:

Readers find a document’s meaning hazy, due to conflicting cues about what’s important, when it contains sentences with misleading emphases.

The talk version beats the clear version in cognitive fluency (and in apparent spontaneity), but it loses in clarity (partly) because of its misuse of the stress position. Hazy meaning is the sentence’s key contribution, whereas the talk version stresses misleading emphases, an idea previously introduced. Stress position isn’t the only way reorganized sentence structure departs from talk, but Gopen’s experience indicates that, in legal writing, it’s the most ignored. Exploiting the stress position requires sentences differing from talk.

Haziness takes a toll on all argumentative writing; in abstract endeavors, it detracts from thought itself. With clarity being much about emphasis, reorganizing sentence structure is a medium through which clear writing deepens thought. (“Plain-talk writing” is inherently inimical to clear thought.)

In the next entries, I’ll discuss how and why the importance of Williams and Gopen’s discovery of the stress position is almost invariably missed by writing authorities. Resistance to exploiting the stress position will be seen rooted in a misguided attachment to the pragmatics of talk. We will also see that clear writing’s difference from talk has implications for … you guessed it, the comma. It supplies the last big piece to the comma puzzle.