Why not just show a concrete infinite set of linearly-independent square-integrable functions? (Although, come to think of it, we must know something more about thedomainof those square-integrable functions. If that domain can beanymeasure space, then the proposition might not even be true in general.)