Halliday & Matthiessen (1999: 614):
Children make the transition from protolanguage to language typically in the second half of the second year of life. […]; we may assume that in general terms they are recapitulating the phylogenetic evolution of language, although of course we can only speculate about the way that evolution took place (it is important to say explicitly that all human languages known today are equally far removed from that phase in our semiotic history). During that stage they learn to construe elements and figures, and in this way "semanticise" both the construction of experience and the enactment of interpersonal relations. In terms of the grammar, they learn to form groups and clauses, and to select systemic options simultaneously in transitivity and in mood.
 Since the 'phylogenetic evolution of language' includes the evolution of Modern English from Anglo-Saxon, this is obviously not true in the terms that it is stated. (The ontogenesis of Modern English does not recapitulate the evolution of Modern English from Anglo-Saxon.) The notion that ontogenesis recapitulates phylogenesis derives from the 'Biogenetic Law' of Ernst Hæckel, in the field of biology; Thain & Hickman (1994: 67):
Notorious view propounded by Ernst Hæckel in about 1860 (a more explicit formulation of his mentor Muller's view) that during an animal's development it passes through ancestral adult stages ('ontogenesis is a brief and rapid recapitulation of phylogenesis'). Much of the evidence for this derived from the work of embryologist Karl von Bær. It is now accepted that embryos often pass through stages resembling related embryonic, rather than adult, forms.
 If different languages have evolved at different rates, then this is not strictly true; though it is true enough to undermine racist claims that speakers of the most conservative languages are somehow inferior, which is presumably the reason for including the parenthetical disclaimer.