One view of language origins sees it as ancient and selection-driven; the other as recent and emergent. Such disagreement occurs because language is ephemeral, detectable only by indirect proxies. Because internalized language and symbolic thought are tightly linked, the best archaeological proxies for language are symbolic objects. Nothing indicates convincingly that any hominid behaved symbolically prior to Homo sapiens, which originated 200 kyr ago but started behaving symbolically only 100 kyr later. Most probably the necessary neural underpinnings arose exaptively in the extensive developmental reorganization that gave rise to anatomically distinctive Homo sapiens, and were recruited subsequently via a necessarily behavioral stimulus. This was most plausibly the spontaneous invention of externalized language, in an isolate of Homo sapiens in Africa, that initiated a feedback process between externalized structured language and internalized language/organized thought. These subsequently spread in tandem throughout a species already biologically predisposed for them. Despite its qualitatively remarkable result, this exaptive process would have been perfectly routine and unremarkable in terms of evolutionary mechanism.