Clause depth, not word count: how a sentence complexity analyzer reads your syntax

Most readability tools treat a sentence as a bag of words: count the words, count the syllables, divide, scale. That works surprisingly well as a rough gauge, but it misses the thing experienced editors actually react to. A 30-word sentence built from two clean clauses reads easily. A 30-word sentence with four nested subordinate clauses reads like a legal contract. Same length, very different difficulty — and a syllable-counting formula scores them identically.

The Sentence Complexity Analyzer measures the dimension those formulas ignore: syntactic structure. Instead of asking "how long is this sentence?", it asks "how many ideas are stacked inside it, and how deep do they nest?" This piece explains what it counts, what the numbers mean, and how to act on them.

Clause depth: the headline number

The tool's central metric is average clause depth. The idea behind it is simple: a sentence with no subordinate clauses has a depth of 1 — it makes one assertion and stops. Every subordinate clause you add nests another idea inside it, raising the depth. In practice the analyzer estimates depth by counting subordinating words — because, although, since, while, whereas, if, unless, when, where, that, which, who, and their relatives — and adding one for each. A sentence with two subordinators reports a depth of 3.

It is a deliberately mechanical proxy, not a full grammatical parse, so read it as a trend rather than a verdict. But the trend is reliable: a document whose average clause depth sits near 1.2 is plain and direct; one that averages 2.5 or more is asking readers to hold several suspended ideas in working memory at once. That is where comprehension quietly drops.

Simple, compound, complex

The analyzer also sorts every sentence into one of three classical buckets, using the same clause signals:

• Simple — no subordinators and no coordinators (and, but, or, nor, for, yet, so). One clause, one idea.
• Compound — coordinators but no subordinators. Two or more equal-weight clauses joined side by side ("I wrote the draft and she edited it").
• Complex — at least one subordinator. One clause is grammatically dependent on another ("She edited the draft because the deadline moved").

The mix matters more than any single sentence. Prose that is almost entirely simple sentences reads as choppy and childish; prose that is almost entirely complex reads as dense and exhausting. Good writing alternates — and seeing the actual distribution for your draft is far more useful than guessing.

Passive voice and coordination ratio

Two supporting metrics round out the picture. The passive count uses a deliberately conservative heuristic: a form of to be (is, was, were, been, being…) followed within three words by a past participle — a word ending in -ed or -en. It will miss a few exotic constructions and occasionally over-count, but on ordinary prose it catches the passives that bury your subject ("the report was reviewed" instead of "we reviewed the report").

The coordination ratio compares coordinating words to the total of coordinators plus subordinators. A high ratio means you tend to chain ideas as equals ("and… and… but"); a low ratio means you tend to subordinate one idea beneath another. Neither is wrong, but knowing your default helps you correct it. Habitual coordinators produce run-ons; habitual subordinators produce nesting.

Where the tool tells you to cut

The analyzer flags specific sentences as split candidates, and the rule it uses is worth knowing so you can trust the suggestions: a sentence is flagged when it runs longer than 35 words or contains three or more subordinating clauses. Those two triggers catch the two distinct ways a sentence becomes hard — sheer length, and structural nesting — and the longest, deepest sentence in your text is surfaced separately as the single worst offender.

When you get a flag, the fix is almost always the same: find the subordinate clause carrying its own complete idea and promote it to its own sentence. The connective tissue ("…, which meant that…", "…, although it should be noted…") is usually where the difficulty hides, and it is usually disposable. Splitting one deep sentence into two shallow ones drops the paragraph's average clause depth without losing a word of meaning.

Reading it alongside a readability score

Clause depth is a complement to, not a replacement for, the familiar grade-level formulas. A syllable-based score like the Flesch Reading Ease or the Gunning Fog Index tells you whether your words and sentence lengths suit your audience; the complexity analyzer tells you whether your syntax does. The two disagree in informative ways. A passage of short words arranged into deeply nested clauses will score "easy" on Flesch and "hard" here — and the complexity tool is the one that matches how it actually reads.

To see your own draft's structure, paste it into the Sentence Complexity Analyzer. It reports your average clause depth, the simple/compound/complex breakdown, your passive rate, and a ranked list of the sentences most worth splitting — turning a vague feeling that a paragraph is "heavy" into a precise map of exactly which sentences to untangle first.