Terms and connectors vs natural language search - when to use each

Short answer: Terms-and-connectors (Boolean) searching gives you exact control. You decide which words must appear and how close together. Natural-language searching ranks results by relevance to a plain-English question. Use connectors when you know the words courts use, or need every case with a phrase. Use natural language when you're exploring an issue, or describing facts that judges may word differently.

How terms and connectors work

You combine search terms with connectors that set the logical and positional rules:

• AND / OR / NOT decide which terms must, may or must not appear.
• Proximity connectors require terms in the same sentence, the same paragraph, or within n words.
• Expanders match word endings or single characters.
• Quotation marks match an exact phrase.

Westlaw and Lexis syntax side by side

Free tools differ. Google Scholar supports phrases, OR and a minus sign but no proximity. CourtListener supports Boolean operators, phrases and proximity searches (for example "slip floor"~10). Check each tool's help page before relying on a connector.

An example

You want cases where a store had notice of a spilled liquid:

(spill! OR liquid OR wet) /p (notice /s (constructive OR actual)) & (store OR supermarket)

That finds paragraphs that discuss a spill and notice together, which is much tighter than a search for the words anywhere in the opinion.

Where connectors shine

• Exhaustive searches: "every case in this court using the phrase mode of operation".
• Terms of art with fixed wording: res ipsa loquitur, qualified immunity, statute section numbers.
• Checking a negative: if a tight Boolean search finds nothing, that tells you something real.
• Reproducibility: the same query gives the same set of results, which you can document.

Where natural language and fact-based search shine

• Early research, when you don't yet know the vocabulary courts use.
• Fact patterns: a judge may write "a puddle near the produce display", not "spill". Relevance ranking and meaning-based (semantic) search find cases worded differently from your query.
• Speed: a good relevance-ranked list surfaces the leading cases in the first results.

The trade-off: natural-language ranking is opaque. It won't guarantee that every matching case is returned, and the order can change as the system changes.

A combined workflow

1. Start with a plain-English or fact-based search to find a few strong cases.
2. Read them and note the exact words the courts use.
3. Build a terms-and-connectors search with those words to sweep for anything missed.
4. Use the citing references of your best cases to fill gaps.

Frequently asked questions

Is Boolean searching outdated?

No. It remains the most precise and reproducible way to search, and courts and librarians still teach it. Most researchers use both methods.

What is the difference between /s and /p?

/s (Westlaw) or w/s (Lexis) requires both terms in the same sentence. /p or w/p requires the same paragraph, which is looser and catches more cases.

Why did my terms-and-connectors search return nothing?

Common causes: too many AND conditions, a proximity connector that's too tight, or a word the courts don't use. Loosen one condition at a time.

Does natural language search find every relevant case?

Not necessarily. It ranks the results it judges most relevant. For exhaustive coverage, back it up with a Boolean search and citing references.

Sources

• Thomson Reuters, Westlaw search connectors and expanders help pages
• LexisNexis, Lexis+ terms and connectors help pages
• Google Scholar Help, search tips (scholar.google.com)
• Free Law Project, CourtListener search help (courtlistener.com/help/search-operators/)