Lesson one in Joel's test should be: Do you regularly critically revisit the software team quality tests?
While I like Joel and the list is a good list, it's only a starting point and without proper knowledge of how to apply any part that that test tells you to apply, you won't get far by just introducing them to your team. Worse, it may do more harm than good.
With a current client, much of that test would've failed when I started out there. Explaining the managers that quality is something measurable in less bugs, better understanding of process and better understanding of delays (!) and cheaper on a very short notice helped me introduce many of those concepts.
It was and always has been hard to introduce any tools that cost money (do the math: a programmer may cost a company $6000 a month all in, that extra tool for $300 a year per seat that saves you 1 hour a day, somehow managers just can't calculate, they'll still consider it too expensive). It's a silly truth and in the end the company will not only loose time, but also the programmers.
But you have to start somewhere. Don't overwhelm the team, but slowly introduce them to new concepts. Source control (a good one, do not, I repeat, DO NOT use Visual Source Safe), feature/debug tracking and quiet working conditions will get you more than halfway, are cheap or free and are easy to introduce. You can (ab)use the feature tracking system for documentation, specifications, bug tracking and planning. Together with rollback facilities of source control this will get you a large part of the x/12 score.
Don't (yet) attempt anything that delays the process or "feels like" a lot of work before any result is seen. These are things like a release schedule, specifications like FD and TD and (very) expensive tools that require knowledge that you currently don't have (interpreting the output from a good memory profiler is an art on itself, for instance). Introduce new concepts slowly and you should be fine.