4.1 Extracting Rules From Examples

Turn examples into rules

The Defense Language Aptitude Battery (DLAB) is a U.S. Department of Defense aptitude test of roughly 126 multiple-choice questions, scored on a scale topping out at 164 points (lowered from 176 in 2016), taken in about two hours. Public sources list category thresholds the services use for school selection: Category I 95 (e.g., Spanish, French), Category II 100 (German, Indonesian), Category III 105 (Russian, Arabic, Thai), and Category IV 110 (Mandarin, Korean, Japanese, Pashto). It measures your potential to learn a language, not prior knowledge of one.

That framing drives every drill in this chapter. The task is never to recall Spanish or Arabic vocabulary. The task is to inspect unfamiliar examples, infer the rule that connects form to meaning, and apply it to a new item under time pressure.

The compare-isolate-state-test routine

Use a four-step routine on every grammar item:

1. Compare several example pairs side by side.
2. Isolate exactly what form changes when the meaning changes.
3. State the rule in plain language with placeholders.
4. Test the rule against a fresh practice-style item before you commit.

This sequence keeps you from guessing based on how English would phrase the sentence.

A worked mini-system

Consider an original drill. The examples say lom naka means red stone and lom naki means red stones. Only one thing changed: the noun's final vowel moved from a to i. A defensible rule is: final i on this noun marks plural in these examples. It is useful because it predicts that tor naki involves more than one of something, not a single item.

Notice the precision. If you state the rule loosely as i means more, you may misfire later when a verb such as tor-i turns out to mark past time instead of quantity. Over-broad rules are the single most common self-inflicted error on aptitude grammar items.

Hunt for paired meanings

Artificial-language drills usually present two sentences that differ by exactly one meaning element: one versus many, present versus past, actor versus object, or positive versus negative. When only one form changes alongside that one meaning, that form is a strong rule candidate. When several forms change at once, withhold judgment and wait for another example pair to separate the variables.

Write rules with placeholders so they transfer to new vocabulary: N for noun, V for verb, ADJ for modifier, NEG for negation. A rule such as N ADJ (modifier follows noun) is far easier to reapply than the memorized fragment stone red.

Stay inside public facts

Ethical, effective prep stays inside what is publicly known. The two-hour length, 126-question count, and category thresholds are public. None of that is an official grammar blueprint, and the examples in this chapter are original drills for skill-building, not claims about protected test items. Your edge comes from speed and discipline at rule extraction, not from leaked content.

Why the routine beats raw memory

The DLAB pace is unforgiving: roughly 126 items in about two hours leaves under a minute per question. You cannot memorize a vocabulary list, because there is none to memorize; the invented words change from item to item. What carries over is the procedure. A test-taker who has drilled compare-isolate-state-test handles a new mini-language the same way every time, while a test-taker relying on intuition re-invents an approach for each item and burns time.

Consider how the same routine scales. Given kel mora = big tree and kel morsa = big trees, you compare (only the noun-ish word changed), isolate (the inserted -s- before the final vowel), state (infixed s marks plural here), and test (does kel folsa predict more than one of fol? yes). The pattern is identical to the earlier lom naka drill even though the mechanism, an infix rather than a suffix, is different. That transfer is the entire point.

Common traps in rule extraction

Three traps recur in self-study and explain most missed items:

• The one-example leap. A single pair cannot separate variables. If both a prefix and a suffix change at once, you need a second pair that changes only one of them. Committing after one example is the leading cause of wrong rules.
• The English smuggle. Learners quietly translate the invented sentence into English, reason about the English version, then translate back, importing English order and agreement along the way. Stay in the invented system.
• The familiar-shape bias. If an invented word resembles a real one (say mora near more), learners assume a meaning. Resemblance is coincidence in a designed aptitude item; trust only the glosses you are given.

Name the trap you fell into after each miss. Leapt from one example or smuggled English order is a coaching note you can act on; got it wrong is not.

Rule-extraction checklist

• Compare two or more example pairs before deciding anything.
• Isolate the single form that tracks the single meaning change.
• State the rule in plain language with placeholders (N, V, ADJ, NEG).
• Test the rule against a new item; if it fails, narrow it.
• After a miss, name the trap: one-example leap, English smuggle, or familiar-shape bias.