In a lecture, I've been told that one of the signs that homo habilis and homo erectus couldn't speak was that they probably lacked the distinction between oral and pharyngeal cavity. But do we really use it, apart from the nasal consonants?
This bit of the book Developmental Neuropsychiatry: Fundamentals describes the difference between the oral cavity in humans and other animals. It doesn't refer to Homo habilis in this context, but it says:
As a result of this anatomical pattern, the range of sounds that an animal can make is limited because of the pharyngeal cavity, which is necessary of sound amplification, is small.
Still, this lower position of the larynx produces a larger pharyngeal space above the vocal chords, which makes possible a greater range of sound modification and becomes the key to the production of articulare speech.
Which suggests that the pharyngeal cavity has a more general use in speech than just making nasal consonants.
This article goes into some detail about Human speech anatomy and its evolution; what it says about the pharyngeal cavity is less about the cavity itself than about the position of the tongue, which allows the separation of that cavity into different sections, leading to different wavelengths being produced. They further say that apes have their tongue entirely in their oral cavity, and are as a result unable to make vowels other than schwas (the basic vowel you make when you're making no effort whatsoever) (they do mention one exception but that species apparently doesn't use its pharynx to make vowels like we do).
The article makes another very interesting point that is relevant to your question, I quote:
Speech communication would be possible without quantal vowels [note from me: the sounds it was talking about that we can make but monkeys can't]. Indeed, there would have been no selective advantage for retaining whatever mutations led to the evolution of the human supralaryngeal vocal tract unless some form of speech had already been part of hominid culture.
That is an excellent point, which is that we need to separate the concept of "speech" from "fully modern speech". Things evolve gradually, and there will have been a continuum from simple communication and vocalizations that our latest common ancestor with other apes likely did, to modern speech with its sounds, grammar, and anatomical and cognitive abilities it implies. And since we no longer have examples of anything on that continuum, we don't know which parts of it we'd be more or less likely to think of as "speech".