Bottom-up processing is the construction of meaning from the smallest units of language upward: individual sounds become words, words combine into phrases, phrases build into sentences, and sentences accumulate into overall meaning. The listener or reader decodes the raw linguistic signal piece by piece, using knowledge of phonology, vocabulary, and syntax to assemble comprehension. It is the complement to Top-down Processing, which works in the opposite direction, from background knowledge and context down to the details.
A listener engaged in bottom-up processing performs a rapid chain of operations:
In practice, these stages overlap and interact. The challenge for learners is that each stage must happen in real time, and a failure at any stage cascades: if you cannot segment the sound stream, you cannot recognise the words; if you cannot recognise the words, you cannot parse the syntax.
Connected Speech is the primary obstacle to bottom-up processing in listening. In natural speech, sounds are linked ("pick it up" → /pɪkɪtʌp/), elided ("next week" → /nekswi:k/), assimilated ("ten bags" → /tembægz/), and reduced (unstressed function words become weak forms: "can" → /kən/, "was" → /wəz/). A learner who learned the word "comfortable" as four syllables will not recognise /kʌmftəbəl/ in speech. This gap between the learner's stored form and the actual spoken form is one of the most significant sources of listening difficulty.
In reading, bottom-up processing involves:
For proficient readers, word recognition is so fast and automatic that it feels effortless; the "bottom-up" machinery runs without conscious effort, freeing attention for higher-level comprehension. For learners reading in a second language, word recognition is slower and more effortful, consuming working memory that is then unavailable for meaning construction. This is why learners can sometimes decode every word in a sentence and still not understand it because their working memory was exhausted by decoding.
Neither bottom-up nor top-down processing is sufficient alone. Stanovich's (1980) interactive-compensatory model proposes that the two directions work simultaneously, with strength in one compensating for weakness in the other. A reader with strong vocabulary (good bottom-up) can comprehend a text on an unfamiliar topic. A reader with strong background knowledge (good top-down) can comprehend a text with some unfamiliar vocabulary. The strongest readers and listeners have both.
For teachers, the practical consequence is clear: develop both. A lesson that only activates schema (top-down) leaves learners who cannot decode the input still struggling. A lesson that only drills phonemes or vocabulary (bottom-up) leaves learners unable to use context and prediction to process language efficiently. The most effective receptive skills teaching integrates both processing directions within the same lesson.
Bottom-up processing is the counterpart to Top-down Processing; together they form the interactive model of comprehension at the heart of Receptive Skills teaching. In listening, Connected Speech features are the primary challenge to bottom-up decoding. Phoneme perception is the foundational unit of bottom-up listening processing. The specific strategies associated with each processing direction differ across Reading Subskills and Listening Subskills.