Natural human communication is multimodal. We pair speech with gestures, use emoji with text, and combine writing with images from doodles to comics to advertising. This communication is structurally complex, especially in visual narratives where grammatical structures organize both the sequential text (syntax) and the sequential images (narrative). Such complexity poses a challenge to linguistic models that focus on single modalities. Multimodality also clearly uses patterned “templates” akin to the productive lexicon of language, be it in comics, cartoons, memes, advertisements, and more. I will outline a model of the language faculty for addressing this complexity embedded within Jackendoff’s (2002) Parallel Architecture of language. Different types of human expression — speech, gesture, drawings, and their multimodal interactions — arise as emergent activation states out of this broader cognitive architecture, reimagining what constitutes the language system.