The Science of
How Kadam Works

The most memorable vocabulary learning happens through encountering words in meaningful context. When a learner notices an unfamiliar word mid-story, the surrounding narrative provides clues that bind the word to memory far more effectively than a definition alone. Words learned this way are retained longer and transfer more readily to new contexts.

Krashen's Input Hypothesis argues that language is acquired when learners are exposed to comprehensible input before being asked to produce it. Therefore asking someone to form sentences using a structure they have never encountered causes them to only guess. Exposure first creates the internal model that production can then draw on. Therefore in Kadam, every quiz follows from a story.

Research has overturned the older view that the first language needs to be suppressed. In fact, it is a resource, as it can serve as a fully formed conceptual framework onto which new linguistic forms can be mapped. Translation also forces precision. A learner who vaguely understands a word must commit to an exact equivalent therefore revealing gaps that paraphrase tasks tend to conceal.

Not all error correction produces learning. Simply marking an answer wrong and moving on is ineffective. Metalinguistic feedback directs the learner's attention to the specific linguistic property that caused the error, without giving the correct answer outright. The learner must still do the cognitive work of repairing their output, and that effortful repair is precisely what drives restructuring of the learner's internal grammar.

Evaluating one's own performance requires accessing and appraising one's knowledge state, which is the kind of reflective monitoring that distinguishes skilled learners from rote memorisers. That rating then feeds directly into review scheduling where words rated hard return sooner, and words rated easy are deferred.

Metacognition refers to a learner's awareness and regulation of their own cognitive processes. In language learning, its most direct application is error review through stepping back from the flow of practice to examine, deliberately, where knowledge breaks down.

Baddeley's working memory model identifies the phonological loop as the system responsible for holding and rehearsing sound-based information. Papagno and colleagues showed that blocking this loop significantly impairs foreign vocabulary learning while leaving other cognitive tasks unaffected. When a learner reads aloud or shadows an audio recording, the sound of a word, its rhythm and stress, become additional encoding channels alongside the visual form.

When a speaker processes a sentence with a particular grammatical structure, that structure becomes temporarily more accessible and more likely to be reused in production, even with entirely different vocabulary. Kadam uses this by highlighting unfamiliar constructions during the story reading phase. When the learner then moves to a sentence-formation task, the recently encountered structure is still active in memory, making production feel more natural.

Researchers distinguish breadth (knowing many words at a surface level) from depth (knowing a word's full range of collocations, and register). At advanced levels, depth predicts reading comprehension more strongly than breadth. In Kadam's Word Vault, a word begins with its core sense. As proficiency rises from A1 to B2, additional senses are unlocked from dictionary sources and introduced in context.