Remember there are two types, hardware and software. Your first line of defense if usually your hardware firewall which is in most cases housed on your router. Home routers use network address translation (NAT) to filter and properly direct packets to a specific client (computer). Basically this makes it possible to have multiple clients on a network at one time without having to have multiple global IP addresses. Google > 184.108.40.206 (your house/isp) > 192.168.1.14 (your computer). By definition only, this is technically a firewall. Packet headers are analyzed and redirected. With no security settings in place this can go entirely unfiltered and you are susceptible to everything. Fortunately, firmware upgrades allow for your router to be upgraded and security settings can be put into place that allow for automatic dropping of packets that are flagged as a security risk. You may also change these settings yourself to allow incoming and outgoing packets to be sent and received. Think port forwarding and video games. This quickly crosses over into setting up a proxy but keeping things on topic.
The most common example of a software firewall is Windows Firewall. It basically works the same way except it's a program that runs while you are running windows (instead of always on as is when your router is on). Software firewalls tend to be more secure because they focus more on security rather than basic function. They can also cross reference with a server for the latest security threats and detect malicious software. Also, since this runs on our computer software firewalls can monitor specific programs individually to which are requesting access to the internet and which are attempting to accept incoming connections. Another added bonus vs a hardware firewall is that software firewalls can determine if another computer on the network is already infected and protect your computer from whatever it is that may be. The router might prevent it from getting in or out, but once something is in, it's in.
Assuming you don't fall victim to downloading something that is mislabeled, like a movie that is really a virus and not a movie at all; drive-by downloads happen when you don't know something is being downloaded. As far as language, it really depends on the vector used in order to trick the user into downloading the file. A realistic and common example would be falling victim to cross-site scripting (XSS) to a website with a hidden iframe of one pixel and tall and wide with a command written to connect to for the download to begin as a background process. Another common example is falling victim to both types. You download what you believe to an .mp3 of your favorite song (and it very well might be) however an additional line is added in for you to connect to a host server to download whatever malicious software desired when you play the song.
"The quieter you become, the more you are able to hear..."
"Drink all the booze, hack all the things."