TOEIC Link Reading — Lexical Inference From Morphology and Collocation Under Unknown Words

Every reader, at every level, meets words they do not know. The difference between a mid-band and a high-band performance is not the size of the vocabulary — no vocabulary is large enough to guarantee full coverage of a business passage — but what the reader does at the moment an unknown word appears. The weak response is to stop, re-read the sentence hoping the word will resolve on a second pass, and lose both time and the thread of the passage. The strong response is to infer: to fix a working meaning for the word from the evidence around it and keep reading. Lexical inference is a trainable skill built on three evidence sources, and a reader who combines them can read past an unknown word so smoothly that comprehension barely dips.

The reason inference works is that words in real text are heavily constrained by their surroundings. A word does not float free; it carries the marks of its internal structure, it keeps company with a predictable set of neighbors, and it sits in a sentence whose logic narrows what it could mean. Each of these is a source of evidence, and while any one alone may be weak, together they usually pin the meaning close enough to answer the question. The trained reader does not need the exact dictionary definition — only a meaning precise enough for the task at hand, arrived at in the second or two that separates fluent reading from a stall.

Evidence source one: morphology

The internal structure of a word is often its most reliable clue. English builds meaning out of roots, prefixes, and suffixes, and a reader who can decompose an unfamiliar word can frequently recover its rough sense from its parts. A word beginning with in- or un- or dis- is likely a negation; a -tion or -ment ending marks a noun, usually an action or its result; an -ize ending marks a verb of making-or-becoming. Meeting reconfigurable cold, the reader who sees re- (again), config (arrange), and -able (capable of) can construct "able to be arranged again" without ever having learned the word. Morphological decomposition will not always give a precise meaning, but it very often gives the part of speech and the polarity — whether the word is positive or negative, an action or a thing — and that is frequently enough to eliminate wrong answers and confirm the right one.

The limit of morphology is that it can mislead when a word's parts no longer add up to its modern meaning. But on the business and semi-technical vocabulary the TOEIC Link module favors, morphology is transparent far more often than it is treacherous, and it is the fastest of the three evidence sources because it requires no re-reading — the clue is inside the word itself.

Evidence source two: collocation

Words keep predictable company. Certain verbs go with certain nouns, certain adjectives with certain things, and the pattern is stable enough that the neighbors of an unknown word often reveal it. If a passage says a proposal was unanimously ___, the blank is almost certainly a word for approval or rejection, because those are what a body does unanimously. If a report describes a figure that ___ sharply, the unknown word is a verb of rising or falling, because those are what figures do sharply. Collocation works because language is full of high-frequency partnerships, and recognizing the partner tells you the category the unknown word must belong to even when its exact identity is hidden.

Collocation is especially powerful for the confusable, near-synonym vocabulary that the module tests, because it distinguishes words that morphology cannot. Two words may share a root and still differ in the company they keep, and the neighbor is what selects between them. Building sensitivity to these partnerships is closely related to the confusable-pair discrimination the Part 5 vocabulary items demand, and the same collocational instinct that fixes an unknown word in a reading passage is what separates a tenant from a tenet or a principal from a principle when the surface forms are close. Reading widely in business English, and noticing which words habitually travel together, is how this instinct is built.

Evidence source three: local context and sentence logic

The third source is the logic of the sentence itself — the connectives, contrasts, and parallels that constrain what a word could mean. A word after but or however is likely to oppose what came before; a word after because or therefore stands in a causal relation to it; a word in a list is probably similar in kind to its list-mates. If a sentence says "the launch was not delayed; it was merely ___," the not… merely structure signals that the unknown word is a milder alternative to "delayed" — perhaps "postponed slightly" or "rescheduled." The sentence's grammar has done most of the work of defining the word before you even consider its parts. This source is the slowest of the three because it requires holding the sentence's structure in mind, but it is also the most precise, because it uses the specific claim being made rather than general patterns.

The three sources are strongest in combination. Morphology gives polarity and part of speech; collocation gives category; sentence logic gives the specific shade. A reader who runs all three in the second an unknown word appears usually arrives at a meaning precise enough to proceed — and crucially, keeps moving. Stalling on an unknown word is doubly costly, because it not only spends time but lets the earlier facts of the passage decay while you stare, a capacity problem examined in working-memory load management and chunking. Inference protects comprehension precisely by keeping the reading rate steady.

Why speed depends on not sounding words out

Lexical inference also interacts with reading rate. The reader who meets an unknown word and slows to sound it out internally, syllable by syllable, pays a large time cost for no comprehension gain — pronunciation does not reveal meaning. The efficient move is to accept the word as an unresolved token, infer its rough sense from the three sources, and continue at pace. This is part of the broader habit of reading for meaning rather than for sound, developed in subvocalization reduction and reading-rate calibration; an unknown word is exactly the moment when the temptation to subvocalize is strongest and the payoff is lowest.

A four-week protocol

Week one — decompose. Work untimed with a passage set. Every time you meet a word you are unsure of, write its parts and what each contributes, then guess a meaning from morphology alone before checking. The goal is a reflex for seeing structure inside words.

Week two — read the neighbors. Continue untimed, but now, before decomposing, note the words immediately before and after the unknown one and ask what category they force it into. Compare your collocation guess with your morphology guess; where they agree, your confidence should be high.

Week three — use the sentence. Add the third source. For each unknown word, identify the connective or structure governing its sentence and let it narrow the meaning. You should now be triangulating from all three sources and noticing which one carries the most weight for a given word.

Week four — at pace. Reintroduce TOEIC timing. The target is to fix a working meaning and keep moving within a second or two, without re-reading. Track how often an unknown word actually blocks a correct answer; for most trained readers the answer, by the end of the month, is almost never — because a meaning precise enough for the question is nearly always recoverable from the evidence already on the page.

The habit worth keeping

The durable result of this training is a changed relationship with unknown words. They stop being emergencies that halt reading and become ordinary events the reader resolves in stride, using evidence that is always present because language always supplies it. No test-taker can memorize every word a business passage might contain, but every trained reader can infer the ones they have not met — and that ability, more than raw vocabulary size, is what lets comprehension hold steady across a full reading module.