The smallest unit of 'Spoken language' is

Linguistics describes a language in layers, and each layer has its own basic building block. The sound layer (phonology) deals with the contrastive speech sounds a language uses to keep words apart; the word-structure layer (morphology) deals with the meaning-bearing pieces words are built from; meaning and sentence structure are studied as fields of analysis (semantics and syntax) rather than as building blocks.

The stem asks for the basic building block of language as it is spoken, so the relevant layer is the sound layer. A phoneme is a contrastive speech sound of a language: the spoken word cat is made of the three phonemes /k/, /æ/ and /t/, and swapping just one of them (/k/ for /b/) yields a different word, bat. No smaller stretch of speech is used by a language to distinguish one word from another, so the phoneme is where the analysis of spoken language bottoms out.

Cross-check against the other terms:

• Morpheme - a meaning-bearing piece of a word (un-, kind, -ness in unkindness). Every morpheme is itself spoken as a string of phonemes, so it sits one layer above the sound layer.

• Semantics - the branch of linguistics that studies meaning: word senses, sentence meaning and interpretation. It names an area of study.

• Syntax - the system of rules by which words combine into phrases and clauses. It also names a level of description rather than a unit of speech.

Hence the smallest unit of spoken language is the phoneme.