The answer lies in the fact of how many times the signal has to be diverted. A telephone call to the USA or Australia may take a fibre-optic cable route or a satalite connection, depending on what the telecom system finds open at that precise time. The signal takes the most direct path available. Fibre-optic cable takes the least distance, as it travels under the sea. That is why there is not usually an 'echo'. If a signal has to bounce up to a satellite first, it travels 76 000 kilometres before it reaches the earth station again. That causes a lapse of up to a second. In a "live' satellite linkup discussion, the signal may have to be 'bounced' back and forth by up to three satellites, depending where the two parties talking to each other are. So a delay of up to 3 seconds or more develops from the person talking, to when it is received at the studio. The interviewer has to take this into account. That is why the person being interviewed will usually signal the anchorman that his sentence is finished by calling their name. That way the interviewer knows he can ask his next question. Of course, after the signal comes into the studio, it has to be sent out again, so the words you actually hear at home may have been spoken up to 5 seconds before.
One may also experience a delay or an echo when speaking on a mobile phone. Your signal is first relayed to the service provider's main distibution centre. Then it is sent back to the phone you are speaking to. If you are in a remote part of the country from the main distribution centre then it may take a second before your signal comes back. This can make your voice echo in your own ear.