Unlike other computer languages ``where you have to practically know the whole language before you can use any of it usefully,'' Wall says that new Perl programmers can learn the subset of Perl they need to solve a particular problem without having to learn the rest of the language. Because of this property, Wall suggests Perl might be a good introductory language for beginning computer science students. While Wall hopes the new philosophy will help Perl will remain stable for a while, he predicts the continued development of new programming languages, particularly those that make programming accessible to ordinary people. ``I think to the extent that ordinary people are going to be programming, they're not going to be programming in C++ -- because you practically have to have a degree in computer science to do that.'' However, he says that simply adding more English syntax to programming languages is not likely to make programming easier. ``When they first came out with Cobol they said `this is a very English-like language,' but I think they confused the process of just throwing a few English phrases into the language and ending their sentences with periods with some of the deeper aspects of natural language. Our computer languages will start functioning more and more like natural languages, but there is a large amount of misunderstanding in the current computer science community about how natural languages actually work.''

And Wall, a linguist by training, is certainly qualified to talk about natural languages. In fact, he says his background in linguistics and his observations about natural language have influenced his programming language development work. ``Linguists realize the limitations of language and realize that when you are trying to solve a problem in a natural language you sort of whack at it until you get the thing into the shape that you want it to be,'' he explains. ``I think people program similarly. They rough things out and hack things in. I think a lot of computer scientists believe programmers are omniscient and can figure out before hand all the design criteria, get all the ducks in a row -- and then magically the program just writes itself because it will be obvious. I don't think real people think that way. I don't think they talk that way, or write that way.''

Wall also credits his linguistic background for his views on ambiguity in programming languages. ``My view on local ambiguity is that it's okay because people deal with it well. On the level of phrases and sentences people are very good at figuring out exactly what's going on -- so I think a computer language ought to be allowed to have ambiguity at that level.'' He adds that humans use various natural language clues to disambiguate sentences.