Empty Components

Two historical accidents leave shapes inside characters that do no work. A scribe adds an arbitrary stroke so one glyph writing two words can split in two; the stroke says only “this one is different.” Or form drift grinds a once-functional part into wreckage signaling nothing about the word’s sound or meaning. Both routes yield empty components: pure form, carrying no information about the spoken word behind the character — a fourth category beside the form, meaning, and sound components that serve writing’s three attributes.

Absence of function removes the natural retrieval cues, so empty parts are hard to remember and invite false etymologies. The repair has two moves: attach a causal one-liner to each empty component — “mark splitting this character from its twin,” “corroded bow-and-arrow” — and render the part as glass inside the mnemonic, storing the non-function in the image itself. Neither move requires paleography; both end the search for a hook that was never there.

Marks That Split One Glyph into Two Characters

• One glyph, several words. Early script let one character write several words; context carried the load until the split had to become visible. One route adds a semantic component: 月 (ròu, “meat; body”) joined 北 (běi, “north”; originally writing “back”) to give 背 (bèi, “back”). The other adds an arbitrary mark, empty by construction.
• One stroke, one job. 夫 (fū, “adult male”) is the standing-adult picture of 大 (dà, “big”) plus one stroke that announces a different character and has no connection to 一 (yī, “one”).
• Marks ride dual-duty bases. In 少 (shǎo, “few”), 小 (xiǎo, “small”) does meaning and sound work at once; only the slash is empty.
• 口 is the workhorse mark. 口 (kǒu, “mouth”) separates 高 (gāo, “tall”) from 京 (jīng, “capital”) and originally sat where 香 (xiāng, “fragrant”) now shows 日 (rì, “sun”).
• Early marks hide kinship. 高 and 京 began as the same tall-building drawing — tall buildings stood only in cities, whence the “capital” sense — but the early 口 sent them down separate tracks; the modern pair barely resemble each other.
• A mark can moonlight as the phonetic. In 句 (jù, “phrase”; its original “crooked” now written 勾 gōu), 口 disambiguates and supplies the sound; the 勹-shaped leftover is empty residue of 丩 (jiū, “entangled ropes”). In 千 (qiān, “thousand”), 亻 — variant of 人 (rén, “person”) — is a real phonetic Mandarin no longer betrays; the crossbar is the empty mark.

Corruption: Drift That Breaks Function

• The damage test. Form change counts as corruption only when it damages sound or meaning work; drift that leaves both legible is ordinary evolution.
• The bow that became a body. 射 (shè, “to shoot”) began as a drawn bow and arrow; a hand — now written 寸 (cùn) — joined to stress the draw; across a sequence of ancient forms, the bow-and-arrow collapsed into a 身 (shēn, “body”) lookalike. That 身 is residue; the traditional pulling-the-string-toward-your-body story reanalyzes residue as semantics.
• Corruption by merger. 春 (chūn, “springtime”) was 艸 (cǎo, “grass”) + 日 + 屯 (tún, “seedling”), 屯 the phonetic via the zh-/ch-/sh- ↔ d-/t- sound-series trade, until 屯 and 艸 fused into the empty shape 𡗗.
• Same shape, different graves. 𡗗 recurs in 奏 (zòu, “play music”), 泰 (tài, “peaceful”), 奉 (fèng, “to offer”), 秦 (qín, “Qin dynasty”), and 舂 (chōng, “pound grain”) — corruption every time, different ingredients each time except 秦 and 舂. Identical shapes prove nothing about origin.
• Buried phonetics. 外 (wài, “outside”) carried 月 (yuè, “moon”) for sound, now 夕 (xī, “evening”); 布 (bù, “cloth”) carried 父 (fù, “father”), now ナ; 早 (zǎo, “morning”) carried 棗 (zǎo, “jujube”), now 十 (shí, “ten”).
• Damage and demotion are separate calls. The corrupted top of 表 (biǎo, “exterior of clothing”) still counts as the phonetic it was, 毛 (máo, “hair”); the corrupted 夕 of 外 is filed empty. Corruption often empties a component, never automatically.

Glass in the Memory Image

• Build on real structure only. Slicing 偷 (tōu, “to steal”) into a person on a roof at night with a knife invents structure the character never had and trains the wrong decomposition. Ground every mnemonic in functional components.
• Two legitimate cue sources. A character’s traditional explanation under an explicit historically-false flag — the 射 bowstring story is vivid because it pretends the residue means something — and your own imagination.
• Render the empty part transparent. For 射, picture an archer statue: drawing hand solid, body glass. Transparency makes “carries no information” a perceptual property of the image; recalling the picture recalls which parts to ignore.
• Cap the depth. Universities in Taiwan and China teach corruption as its own graduate paleography subject; a learner needs the one-liner plus the glass image, pulling in full histories only when they aid retention.

Related Pages

• How Chinese Characters Work - cluster hub; where empty components sit in the reading order
• Three Attributes of a Character - the form/meaning/sound triad that empty components fall outside of
• Surface vs Deep Structure - functional-component decomposition; empty parts are where surface shape misleads most
• Sound Components - the function that buried phonetics like the 月 of 外 once performed
• Sound Series - the initial-correspondence families that make 屯 a credible phonetic for 春
• Meaning Trees and Original Meanings - the original-sense reasoning and “now written X” notation used throughout
• Rules of Effective Memorization - the domain-general memory rules the glass convention instantiates

Sources

• Outlier Linguistics, Chinese Character Masterclass — commercial course; lesson PDFs kept locally outside this repository. https://www.outlier-linguistics.com/

Open Questions

• Strokes added for visual balance are sometimes described as a third origin of empty components; only marks and corruption are established here. Is decorative regularization a real third route or a subtype of corruption?
• What criterion moves a corrupted component from “still the phonetic” (the 毛 of 表) to “empty” (the 夕 of 外) — residual recognizability, or case-by-case judgment?
• Does the glass convention stay discriminating once dozens of memory images each contain a transparent element, or does it need variants at scale?