added parameter to the NLP directory and joining jar arguments with more standard os.path.join

[d1ma]

• Now you can specify the directory containing the Stanford parser
• When looking for the jar, you currently do basepath + filename, but it's better to use os.path.join