Prepare for the MLAT by training the specific operation behind each subtest: induce number-system rules from examples, map sounds to symbols consistently, decode altered English words through roots and context, identify grammatical functions rather than word forms, and practice retaining invented word pairs despite distraction. Distinguish aptitude from proficiency testing before choosing any drill, because the two reward opposite preparation.
Aptitude versus proficiency: the distinction that decides your preparation
The MLAT measures capacity to learn language material quickly; CEFR-aligned proficiency tests measure what you can already do in a real language. Confusing these two leads you to prepare for entirely the wrong task.
The Modern Language Aptitude Test, developed by John Carroll and Stanley Sapon and distributed by Second Language Testing, Inc., presents invented language material and English-based puzzles rather than any target language. It is designed to indicate how readily a person can learn a new language. For current administrative facts such as formats and ordering, treat the publisher's site at 2lti.com as the controlling reference rather than secondary summaries.
Contrast that with proficiency assessment as described on the Berlitz testing page: CEFR-aligned tests that evaluate speaking, writing, listening, and reading in more than thirty real languages, reporting what candidates can do. A proficiency candidate reviews the target language itself; an MLAT candidate sharpens learning operations on unfamiliar material. If your drills look like language classes, redirect them toward the subtest operations described below.
| Dimension | Aptitude test (MLAT) | Proficiency test (CEFR-aligned) |
|---|---|---|
| What it reports | Likely ease of learning a new language | Current skill in a specific language |
| Material | Invented words, artificial number systems, English puzzles | Real texts, audio, speaking and writing tasks |
| Target language knowledge | Not required | Central to every item |
| Useful preparation | Rule induction, phonetic matching, function spotting, memory under interference | Practicing the four skills in the actual language |
| Reference scale | Aptitude norming by the publisher | CEFR levels describing can-do performance |
Number Learning: inducing an unfamiliar number system from examples
This task plays audio examples of numbers in an invented language and asks you to infer how the system combines parts, then apply the rule to values you have never heard.
The difficulty is structural: a novel number system may be additive, subtractive, or ordered differently from English, and the audio gives you worked instances, not a rule statement. Train by listening to small sets of invented numbers, writing a candidate rule, and testing it against the next example before it is revealed. Speed of hypothesis revision matters as much as the first guess.
Worked scenario: suppose a tape gives 'nad' = 3, 'kiv' = 10, and 'nad-kiv' = 7. A plausible mistake is reading 'nad-kiv' additively as 13 and answering future items with addition. The better decision is to test the pattern: if three-before-ten yields seven, the system subtracts, so 'kiv-nad' should yield thirteen and 'niv-kiv' with niv = 2 should yield eight. Why it matters: one untested assumption silently corrupts every later item in the set, while a single check example exposes it.
Phonetic Script: matching sounds to symbols without English-letter habits
You learn a short phonetic alphabet and then identify which English words contain which symbols, or how symbols reappear in new transcriptions. The trap is reading symbols as ordinary letters.
The skill is consistent symbol-to-sound mapping: once a symbol is defined, it keeps its sound regardless of English spelling conventions. Build this with minimal-pair drills — transcribe pairs like ship versus sheep using the assigned symbols, then transcribe unfamiliar dictated syllables. Check yourself by reading your transcription back to a partner's key, noting every place your English-reading habits overrode the assigned sound value.
A useful exercise: assign yourself six invented symbols with non-intuitive sound values, transcribe ten dictated nonsense syllables, and decode a partner's transcription of the same syllables. Expected observations: your first errors cluster where a symbol resembles an English letter with a different sound, and errors drop sharply once you verbalize each symbol's sound before writing it. If a symbol still triggers its letter-sound reading after two rounds, slow down and re-anchor it with a keyword word containing that sound.
Spelling Clues: recovering English vocabulary from degraded forms
This part presents English words in altered or puzzle-like spelling and asks for meaning or matching. It rewards deep vocabulary organized by roots and context, not list memorization.
Work on two levers. First, morphology: knowing that 'spec' relates to seeing lets you reconstruct altered forms of inspect, spectator, and conspicuous even when surface spelling is distorted. Second, context inference: practice reading deliberately corrupted sentences and deciding what the damaged word must mean from its grammatical slot and neighbors. Compare your inferences with a dictionary and record which root or cue actually produced the right answer.
A concrete drill: take a paragraph, alter one letter and delete one vowel from every multisyllable word, then reconstruct it against the original. Observation to expect: words you can parse by root reconstruct instantly, while rare words force reliance on context alone. That split is the point — the drill teaches you which of the two strategies to deploy when an item resists decoding, instead of stalling on the surface form.
Words in Sentences: grammatical function, not word form
Items underline a word in a sentence and ask you to match its grammatical function — subject, object, modifier, and so on — to underlined words in other sentences. Function, not appearance, is the target.
The skill is identifying what role a word plays in its sentence regardless of its part of speech or position. Train with a substitution test: replace the underlined word with a known exemplar of each function and see which substitution preserves the sentence. Practice across sentences where the same surface form serves different roles, so your judgment depends on structure rather than the dictionary entry.
Worked scenario: in 'The generous teacher gave the surprised class a long assignment,' the target is 'surprised.' A plausible mistake is labeling it a verb because it ends like a past tense. The better decision is the substitution test: 'the hungry class' works, so 'surprised' is a modifier of 'class,' not an action. Why it matters: MLAT-style matching asks which other sentence contains a word doing the same job; answering by form sends you to look-alike verbs that do entirely different work.
Paired Associates: retaining invented word pairs under interference
You briefly learn vocabulary pairs in an invented language and must recall them later. The challenge is keeping several similar-sounding pairs distinct while new pairs keep arriving.
The core obstacle is interference: invented pairs often overlap in sound and structure, so a weakly encoded item gets overwritten by the next one. Build two habits. First, encode each pair with a distinctive image or sound hook on first exposure rather than silent repetition. Second, practice staggered review — recall earlier pairs after learning each new batch — so retrieval, not mere exposure, does the strengthening.
Exercise with a self-check rubric: create ten invented pairs where six share a syllable, learn them to one perfect recall, then learn five decoy pairs and recall the original ten. Milestones to observe: after the decoys, eight or more originals recalled with no syllable swaps is a strong learning indicator; swapping words that share a syllable identifies interference as your specific weakness; dropping pairs learned first suggests your earliest encodings were too shallow. These observations guide which drill to repeat — they are learning milestones, not predictions of any score.
An adaptable practice sequence and concrete readiness checks
Train subtest operations in rotation rather than one at a time, because the MLAT's parts reward different skills. A four-week rotating sequence with weekly self-scoring keeps every operation warm.
Week one, diagnose: run one short self-made drill of each type — number induction, symbol mapping, degraded-word decoding, function identification, pair retention — and record which two operations feel weakest. Weeks two and three, weight roughly two sessions toward the weak operations and one each toward the stronger ones, always on invented material you have not seen. Week four, run mixed mini-sets that alternate operations within a single sitting, which mirrors the demand to switch rule systems quickly.
Readiness checks before you finish: you can state the inferred rule of an invented number system and verify it on one held-out example; you transcribe dictated nonsense syllables with symbols faster than you can read them as letters; you label grammatical functions using the substitution test without reverting to word form; your pair-recall milestone from the previous section holds after distraction. If any check fails, return to that subtest's drill for a focused week.
- Rotate all five operations weekly; do not block-practice only your favorite subtest.
- Always test a rule hypothesis on one new example before committing to it.
- Score self-drills against your own previous attempts, treating the numbers as learning milestones only.
- Keep every practice set on invented or altered material so the rule-induction habit, not memory, does the work.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.