Vjylku

From Wikipedia we have:

The Proto-Turkic language is the linguistic reconstruction of the
common ancestor of the Turkic languages that was spoken by the
Proto-Turks before their divergence into the various Turkic peoples.
Proto-Turkic separated into Oghur (western) and Common Turkic
(eastern) branches. One estimate postulates Proto-Turkic to have been
spoken 2,500 years ago in East Asia.

• Which language did the ancestors of Proto-Turks speak?


• If they spoke scythian (an Iranian language) then why did they shift
  to proto-Turkic?


• Notice that I'm not questioning assimilation done by a nomadic group
  to other groups, I'm questioning about the very first beginning of a
  new language family. A new language family does not pop up through Spontaneous generation in vacuum. How do linguists and anthropologists explain it?

As for the title question, the answer would be "many languages, including proto-Chinese". Focusing on the question in the body, the language spoken by the historical ancestors of proto-Turks, there are two main options. One is that they spoke "pre-proto-Turkic", that is, an undocumented language whose properties are not presently recoverable. As for the name of that language, just as we don't know what ethnonym the proto-Indo-Europeans had for themselves (or the proto-Turks), we don't know what name the 10-generation before ancestors of the Turks or Indo-Europeans had for themselves. I should point out that there has been a theory that Mongolian, Tungusic and Turkic are themselves derives from an Altaic proto-language, but that hypothesis is now rejected by most linguists working in the area. Still, it could turn out that there actually is some unproven relationship between Turkic and some other language group.

An alternative would be that the individuals who constitute the majority of the speakers of the language reconstructable as "proto-Turkic" descended from people who switched languages from something else to proto-Turkic. For example, perhaps those people were Scythians, but for some political reason switched language to proto-Turkic. This has happened many times in world history, for example the ancestors of current Arabic speakers in a number of countries originally spoke Nilo-Saharan or Berber languages (among others), but for political reasons adopted Arabic. But this is simply an imaginable alternative scenario, lacking in compelling historical support in the present case.

So the answer would be "pre-proto-Turkic", which means "whatever language preceded proto-Turkic". No new language family was started, instead, "proto-Turkic" refers to the limit of our ability to reconstruct languages. It may be sensible to talk of new families when dealing with daughter languages of a credible proto-language, such as Germanic or Italic in relation to Indo-European where we can say that the properties of PIE changed one way to give Proto-Germanic vs. another way to give proto-Italic. Until we find evidence of some lost sister of proto-Turkic, we have no reason to speak of "developing a new language family".