This paper is a top-down analysis of the non-protein-coding, canine genome. We demonstrate by use of the y-text-finder method, that the non-protein-coding genome contains lots of hidden y-texts, both short and long, proving that the non-protein-coding genome is the opposite of junk. They are written by means of a y-language of about 28 million y-words separated by stop codons and spelled by nucleotide letters A, C, G, and T. We use the Canis Lupus Familiaris reference genome, Roslin Institute, 2020, from which we select the non-protein-coding part. We show that 70-80 percent of chromosomal y-words are specific for the canine non-protein-coding chromosome, and we show how many y-words any non-protein-coding chromosome shares with any other non-protein-coding chromosome. We demonstrate the peculiar way by which the dog utilizes the nucleotide word-length of y-words to build up its y-language, moreover in a way it shares with the human non-protein-coding genome. In a large table we demonstrate how 18,398 Zipf-qualified y-texts/narratives are distributed over the 40 non-protein-coding chromosomes. 3,812 of these texts/narratives are alpha-qualified and similar in form to human novels. In the last table we compare selected, corresponding characteristics of the human and the canine non-protein-coding genome.