The article contains quite a few examples, though regrettably most are not supported by sound files. In one case there is a reference to some work by John Oswald, who made reversed recordings of William Burroughs in the 1970’s, but if you want to find the recordings you have to look in the WP article on Palindrome in the section headed ‘In speech’. I have listened to the recordings and they just sound like unintelligible mush to me (try these examples). The Palindrome article is a much better piece of work than the Phonetic palindromes one, I think.
There is one overwhelming flaw in the whole article, however, and that is the unquestioning use of the concept of utterances “sounding the same”. Anyone who has experimented with synthetic speech knows that after you have played your bit of sound material to yourself many times, it can be clearly heard as saying what you intended it to sound like, but if you ask someone else to say what they hear, the answer is usually not what you intended. This is a form of auditory illusion. Psychologists have worked on such illusions for a long time. The “Verbal Transformation Effect” is perhaps the best-known: hearing the same thing over and over again often results in successive perception of a whole series of words, some phonetically quite dissimilar (I have been a subject in a VTE experiment, and found the effect quite disorienting). There is a Wikipedia article on Semantic satiation that describes a somewhat similar illusion.
My feeling is that if you equip yourself with some software that can play a sound recording backwards, and you then start listening to any old scrap of speech material played forwards and backwards, pretty soon you will convince yourself that you are hearing something different from what you heard at the outset. And if you are looking for bits of speech that sound the same whether played forward or backwards, you will soon convince yourself that you have found some.
Unless this article can be backed up by scientific evidence, I think it should be taken out, or moved into the Palindrome article with a lot of improvements.