So why were modem speakers such a persistent feature, and "fixture" of the time?
Three basic reasons:
- Adding a simple amplifier and a speaker is the most easy way to handle unexpected situations
- In many countries/networks having a speaker active while establishing a connection was mandatory to make it legal/get a validation
- Adding it was as well the least cost intensive way to handle complex situations.
Keep in mind, the kind of modems most people used were not for dedicated lines, like today's cable or DSL installations, but operating over regular dial up. To start with, anything could happen - starting from some angry mom picking up the phone were her son has hooked up his BBS, wrong numbers, all the way to signal tones outside the spec the modem was built for (for example ring tones in various countries used different rhythms and frequencies).
So having a speaker was a quite good debugging tool - after all, whatever happened on a phone line would be within 3000 Hz, clearly noticeable by humans listening.
This is also the reason why phone regulators/companies required them to be installed as well. They wanted to minimize interruption of third parties (due to mishandled/directed calls) as well as having the user able to detect special situations on his own, without calling for service. In fact, in many cases it was required that each and every modem had to have a handset (or a complete phone) connected, so the operator can pick it up and talk to any (unintended) party on the other side to apologize and/or first talk to the other side before switching to data. In reality most private users ignored that part (as well).
These are the reasons why feeding the speaker, until a valid connection was established, was the default configuration - though, it wasn't really enforced. If you didn't like it, it could be configured away - at least with most modems.
Last but for sure not least: modem producers liked it as well. After all, with a speaker they didn't have to invest much money in detecting all kinds of situations. Let the user handle it like he was used to doing with a regular phone call.