AI Secrets of the Best Briefs

Curious what top brief-writers do differently from the rest of us? This AI-powered study shares concrete, practical solutions.

The Best Briefs What AI Can Teach Us About That “Short and to the Point” Feeling

ROSS GUBERMAN

The author is the president of Legal Writing Pro, the author of Point Made: How to Write Like the Nation’s Top Advocates , and the creator of the legal-editing tool BriefCatch.

Imagine the best brief-writer you know. You can feel free to imag- ine yourself. Now give your gut answer to these questions: Does that lawyer write shorter sentences than average? Use the passive voice less often? Include more analogies? Use fewer adverbs? Discuss more case law? Write fewer words? Use fresher language? Perhaps the answers seem obvious. But what comes first, the perception of “great legal writing” or the answers to those questions? You can tackle the challenge of defining “great legal writing” in many ways. I recently surveyed thousands of judges to get their take on briefs, for example, and I will keep soliciting and sharing similar insights. But I wanted to try something different. Besides asking judges what they think distinguishes the good brief-writers from the bad, why not identify a group of exceptional brief-writers and then use artificial intelligence to figure out what they do differently from the rest of us? After all, judges might all agree that short sentences are hot while the word “clearly” is not, but wouldn’t it be great to see data backing that up? That is, unless that “short and to the point” feeling is really a proxy for something more meaningful but harder to pin down. Here’s what I did: I created two universes of briefs and mo- tions to help develop the five BriefCatch scores I’ve devised for legal documents.

The first set consisted of tens of thousands of pages of motions and briefs signed by dozens of top-rated lawyers. To remove my opinions from the equation, I relied mainly on Chambers and Partners’ rating of top litigators and appellate advocates. For di- versity, I did add briefs from the Solicitor General’s Office across several administrations, briefs that had won Green Bag awards for “exemplary legal writing,” and briefs that judges had singled out as exceptional, either publicly in opinions or privately. The second set: the same number of randomly selected mo- tions and briefs of similar types. It’s fair to question my selection method as arbitrary or elitist. If it were arbitrary, though, we wouldn’t have found so many significant differences in writing between the two sets of briefs. The same goes for the objection that “these bigwigs didn’t really write these briefs themselves.” I worry about elitism, too. But to believe that the selection method colored the results, you’d have to believe that equally good briefs from other lawyers are “good” in a vastly different way from the ones we did look at. And that the writing choices of the top performers in our study reflect their credentials more than their writing. You could also ask why I didn’t focus on who prevails in court rather than on reputation. The truth is that I’ve done that, too, and our number crunching has yielded intriguing insights on predicting which party will win. But writing savvy

Published in Litigation, Volume 46, Number 4, Summer 2020. © 2020 by the American Bar Association. Reproduced with permission. All rights reserved. This information or any portion thereof may not be copied or disseminated in any form or by any means or stored in an electronic database or retrieval system without the express written consent of the American Bar Association. 1

is only a sliver of what dictates outcomes. What’s more, lawyers stuck with bad law, bad facts, or both are less likely to win and more likely to feel forced into writing choices that could muddy the pool of losing briefs. Think of how Supreme Court justices often write more vividly in dissents than when they have to win over skeptical colleagues. For the data analysis, I retained part of a team that developed an algorithm to predict which Supreme Court justice wrote an opinion. Here Are Some Ground Rules First, artificial intelligence and machine learning are fantastic tools, but you need human expertise too. A simple example: The phrases “default judgment” and “trust assets” appeared much more often in the random set than in the heavy-hitter set, but those differences reflect the subject matter of the litigation. You also need expertise to ferret out the take- aways behind the data. What’s going on in the lawyers’ minds that draws them to the best writing choices? And if great legal writing is more than just “I know it when I see it” or even just “Write short and to the point,” what do the data suggest you can do to make readers happier—and more persuadable? Second, relative rates are what matters. Almost any word, phrase, or device can work sometimes. So if you look at the right-hand columns in the tables below and ask, “Isn’t such-and-such OK when you . . .?” the answer is probably “yes.” Again, it’s all relative. Third, gaming the system doesn’t work. If you take an in- effective brief and sprinkle in some “good” words and devices below, it will be just ineffective. The analysis applies only to ac- tual filed work product that grapples with real facts and real law. Finally, the analysis doesn’t consider quoted language. Now, on to what artificial intelligence and I discovered together. Let’s begin with some conclusions that might surprise you. Myth busters, you might call them. On average, the top brief-writers, both trial and appellate . . .

• do NOT write shorter sentences. • do NOT write shorter paragraphs.

• do NOT file shorter briefs or motions. (In the October 2018 Supreme Court term, for example, members of the elite Supreme Court bar who ultimately prevailed on the merits submitted longer merits briefs on average than any other group.)

Illustration by J.F. Podevin

• do NOT fare better on traditional readability metrics like the Flesch index or grade-level score. • do NOT vary their sentence length more than other lawyers do (though they are likelier to follow an unusually long sen- tence with an unusually short one). • do NOT use the word “that” at a lower rate—in fact, they do so at a higher rate. • do NOT use shorter words (but see below). • do NOT use more active voice (but see below). • do NOT use fewer adjectives and adverbs (but see below). Before you denounce the data as sacrilege, consider that what readers experience as overly long sentences could be the occa- sional gargantuan sentence (writers like Chief Justice Roberts draft those, too) floating amid a turgid document. And when read- ers admire short, crisp, declarative sentences, they’re probably not counting words, computing means, and diagramming syntax. Clarity and conciseness are still worthy goals, of course. It’s just that if you believe in data and want to induce that “short and to the point” sensation, the best brief-writers have more nuanced and productive ways to break the mold. Let’s start with substance. Language analysis can yield insights into what kinds of arguments lawyers make and how, not just into what words they use to do so:

Also bridging style and substance: data on headings, defined terms, parts of speech, types of words, and even punctuation.

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . Use slashes. Use semicolons to join two clauses.

TABLE 2

Punctuation Use em dashes to expand on a point. Use colons to explain a point. Conventions Include headings (though they’re no more likely to include subheadings).

Define parties in more than one way, as in “Defendant”

or “Employer”. Use acronyms.

Parts of Speech

Use pronouns like “it,” “she,” and “they” (but not “he” or “him”). Include the relative pronoun “that” after a verb. Use verbs. Use adjectives and adverbs (but see below).

Use prepositions, especially “at,” “by,” and “of.”

Now let’s turn to pure writing style—what you can most read- ily adapt. Here’s how I broke down the style data:

• Analyzed the relative rates of unigrams (a single word or punctuation mark) along with bigrams and trigrams (combi- nations of two or three). • Looked for meaningful differences between the two sets. • Looked for patterns in those differences and then grouped them into nine coherent factors. Here are the nine factors. Factor 1: Speed Up The legal-writing punditry has focused so much on fighting “legalese” (see below) that it sometimes neglects the many bu- reaucratic, heavy-handed, or cumbersome terms that belong to modern English but can drag down a document. To that end, this factor consists of 176 correct-but-ponderous terms that are less common in top-notch briefs.

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . .

TABLE 1

Write “the Court must.”

Reasoning Use language like “for

example” or “for instance” to introduce examples. Address the court in the second person: “Consider,” “Suppose,” and so forth. Use language like “it is true that” or “to be sure” to concede a point. Use a parenthetical for a single-sentence quotation. Quote from authorities within sentences rather than as a full sentence or a block quote. Include language like “some courts” or “many courts” to synthesize case law.

Have a high ratio of cases cited per word. Use “ id. ” Use See also or other signals introducing similar cases on the same point.

Case Law

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . .

TABLE 3

Cumbersome Sentence Structure Follow a semicolon with a conjunctive adverb like “however” or “therefore.” Start sentences with wind-up, throat-clearing phrases like “It is apparent that.” Wordy Phrases Use wordy expressions like “with respect to,” “the fact that,” “are obligated to,” “exists to,” and “in the event that.” Long Words Use long words that have shorter alternatives like “absence,” “characteristic,” “initiated,” and “regarding.”

Factor 2: Verb Surge Powering up this factor: 720 verbs that top brief-writers use more often—and another 90 that they use less often. The verbs fall into three patterns.

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

TABLE 6

Punchy Words

Use vivid, conversational words like “afield,” “array,” “beforehand,” “bulwark,” “chance,” “core,” “gap,” “signs,” and “track.” Start sentences with crisp openers like “But,” “Few,” “Let,” and “Only.” Use elegant phrases like “and thus to,” “far more than,” “in turn,” “let alone,” “need not,” “nor did the,” “the same way,” and “to do so.”

Punchy Sentence Openers

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . .

TABLE 4

Punchy Phrases

Length Quality Use vivid verbs like

Use vague verbs like “indicate.”

“alter,” “erodes,” “falter,” “hoodwink,” “override,” “pinpoints,” “refutes,” and “stymies.”

Factor 4: Punchiness and Pizzazz Like all professional writing, legal writing can be spare and con- cise, yet flat and dull. To that end, we’ve identified another 428 terms—loosely de- fined as “punchy”—that you’re more likely to see in the top brief- writers’ work. What unites these 428 terms? Language you’d read in elegant essays or hear in compelling speech. See Table 6 above. Factor 5: Lawyered Down I was especially curious to see how “legalese” fared. It’s easy to lament. It’s harder to define. And it’s avoided more in theory than in practice. We did find meaningful differences in use rates for 165 terms that I’ll divide into two categories: pure legalese, as in “hereto- fore,” on the one hand, and “normal” words and phrases—like “regarding” or “pursuant to”—that many lawyers and judges sim- ply love too much, on the other. Factor 6: A Slight Modification Bans on adjectives and adverbs are as popular as they are unwork- able. The adjective “disguised” matters in “she used a disguised voice,” just as the adverb “rarely” is key to “courts rarely require.” That said, we did identify 100 modifiers that the best writers use less often than average—and another 100 that they use more often. You want “quality” modifiers, so to speak. See Table 8 on the next page. TABLE 7 RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . Pure Legalese Use language like “aforementioned,” “foregoing reasons,” “forthwith,” “herein,” and “instant case.” Lawyerisms Use language like “contingent upon,” “i.e.,” “namely,” “prior to,” “proximity,” “pursuant to,” “slippery slope,” “to the extent (that),” and “with respect to.”

Diction Use familiar verbs like “expect,” “mimics,” “signaled,” and “try.”

Use bureaucratic or pretentious verbs like “anticipate,” “effectuate,” and “impacting.”

Factor 3: Passive-Aggressive Rates of the passive voice itself do not differentiate the two sets very well. On the other hand, many of those who decry the pas- sive voice can’t really define it, and I’ve learned that, like the phrase “short and simple,” “passive voice” is more a feeling the reader has than the product of linguistic analysis. Some passive constructions are as popular with top brief- writers as they are with the rest of us: “achieved by,” “undermined by,” and “represented by,” to name a few. Yet others are much less common in the great-briefs set, as you’ll see below. Because the real issue with the passive voice is that it makes it harder to see what’s happening, I include over- used nominalizations in this factor, too. All told, this factor includes 177 cases in which the better brief- writers get active.

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . Use passive constructions when the focus should be on the actor: “employed by,” “relied upon by,” “caused by,” “permitted by,” “noted by,” and “produced by.” Use nominalizations rather than active verbs, like “achievement,” “alteration,” “compilation,” and “modification.”

TABLE 5

Passive Constructions

Nominalizations Use active verbs like “achieves,” “alters,” “compiles,” and “modifies.”

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . Use language like “actual,” “fanciful,” “inexcusable,” “infamous,” and “optimal.” Use language like “alternatively,” “comparably,” “contemporaneously,” “inordinately,” and “unequivocally.”

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . . Start sentences with language like “As for,” “After all,” “If,” “Indeed,” and “Those.” Start sentences with language like “At the same time,” “Put another way,” and “To begin with.” Use language like “and so,” “by extension,” “for that reason,” “likewise,” “more to the point,” and “not only because.” Include numbered lists with language like “Second,” “Third,” and “Fourth.” Use language like “than any,” “than that,” and “than those.” Use language like “days later,” “weeks later,” and “months later.”

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . .

TABLE 8

TABLE 10

Adjectives Use language like “candid,” “inapt,”

Sentence Openers: Lightness

Start sentences with language like “Consequently,” “Regarding,” and “Subsequently.”

“mistaken,” “rare,” “tiny,” and “unsettled.” Use language like “elsewhere,” “partly,” “precisely,” “sooner,” “thoroughly,” and “worse.”

Adverbs

Sentence Openers: Logical Precision

Start sentences with language like “Additionally,” “Also,” and “Furthermore.”

Midsentence Logical Moves

Use language like “and, therefore.”

Factor 7: Tone Police How about tone? Are better brief-writers more civil, more logical, less aggressive in making their points, as so many judges contend? The answer: Yes, but the distinctions are slight. Legal writing professors can take a victory lap over the data on “clearly” (we ignore “clearly erroneous”). But only a short lap: “Clearly” appears slightly less often in top briefs, though still far more often than many realize—including in the briefs of many appellate stars who decry “clearly” in public. We include two types of tone differences below.

Numbered Lists

Comparisons

Time References

Include complete dates.

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . .

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . Use language like “clearly,” “completely,” “drastically,” “utterly,” and “wholly.” Use language like “blatantly,” “deceptively,” “disingenuous,” “draconian,” “egregious,” and “flagrantly.”

Sentence Adverbs

Start sentences with language like “More specifically,” “Notably,” and “Significantly.”

TABLE 9

Intensifiers Use language like “entirely.”

Judgmental Modifiers

Use language like “incorrect,” “mistaken,” and “wrong.”

TOP BRIEF-WRITERS ARE MORE LIKELY TO . . . End sentences with words like “before,” “change,” “course,” “enough,” and “like.”

RANDOM BRIEF-WRITERS ARE MORE LIKELY TO . . . End sentences with words like “entirety,” “exclusion,” “inapplicable,” “justified,” and “thereafter.”

TABLE 11

Final Word of Sentence

Factor 8: Gushing Flow The greatest difference of all relates to internal logic. How well does the lawyer massage disparate points into a cohesive whole? How well does the lawyer create order out of chaos? How well does the lawyer push the reader forward? How well does the lawyer avoid needless interruptions? See Table 10 above right. Factor 9: Lighter Than Air Because our analysis included punctuation and capitalizations, we could crunch data on how sentences start and end. I was curious to see a pattern I hadn’t noticed before: Perhaps because they have a good ear and want to end sentences elegantly, the best brief-writers are much likelier to end sentences crisply. See Table 11 to the right. Speaking of endings, let’s close with an image of you as the data-backed ideal brief-writer. Your secret is not that you re- cite “Concise. Clear. Organized.” before you go to bed while your

colleagues don’t. But it’s not an ephemeral art, either. You discuss fewer cases for the same points while interspersing more pithy quotes from the ones you do cite. You add the occasional dash or colon to elaborate or explain. You’re not afraid to concede a point outright, and you try to synthesize as much case law as you can. On the wording front, you strike wordy phrases. You freshen up your draft with punchy language. You replace dull verbs with vivid ones. Rethink your modifiers. Apply a light touch to sen- tence openers and endings. And add headings, numbered lists, and logical connectors. I forgot to mention: And do all that while getting the law right, the record mastered, and the deadlines met. It’s not easy. But it’s still much more science than art. q

Page 1 Page 2 Page 3 Page 4 Page 5

Made with FlippingBook - professional solution for displaying marketing and sales documents online