Since several people made the same point about word segmentation, I obviously didn't express myself very well. What I was trying to say is that, assuming you s

Since several people made the same point about word segmentation, I obviously didn't express myself very well. What I was trying to say is that, assuming you speak the language, you will recognize "dog" a separate ‘ thing’ in a sentence. If I point to a German Shepherd asking what that is, anyone who speaks English can reply "dog". It can't be shorter; it could be longer but then you're adding information ("big dog"). Even illiterate people know "dog" as a distinct ‘ thing’. Since even spoken words are distinct concepts, with a beginning and end, and having boundaries (even if the speech waveform looks continuous), it seems natural to show the boundaries (eg., by placing spaces) if you going to starting writing down the words. Well, we know the spaces weren't added in early writing, by why not? Wikipedia's reason #1 is that adding inaudible spaces would be illogical. I'm not convinced that that was the main reason. The other reasons seem more plausible.