Hi Mohammad, I will attempt to answer your questions. Regarding the length of the ENSO record, we focus on the period after 1950 because this is the period when we have the most reliable observations. We do have sea surface temperature data before 1950, primarily by ships of opportunity, and we do have datasets that use statistical methods to fill the data gaps and reconstruct the temperature data much farther back, generally to the late 1800s. So, we can (and do) analyze ENSO events before 1950 with these datasets, but there is considerably more uncertainty in the data. If you check out Fig. 3 of this paper, you can get an idea of how sea surface temperature data coverage has changed with time. 

You're right that we also can reconstruct ENSO records much farther back using various types of paleoclimate proxy records such as tree rings and corals. Guest author Dr. Kim Cobb wrote about this topic in this post. I actually was part of a study that used tree ring data to reconstruct ENSO data back 700 years (found here). Of course, it's important to realize that data uncertainties are even higher for these data than for the historical period. 

Regarding any comparisons to 1974-75, I am not aware of any detailed comparison that has been done (I have not yet done it). Seeing some similarities would be expected, given that a La Nina influence is present in both cases. The key question is whether there is something distinct about the 1974-75 pattern (or the whole 1973-76 evolution) from the typical La Nina pattern that can inform us about what to expect in the coming winter. Answering that question cannot be done with observational data alone, as the sample size is too small and the chaotic climate variability too large, so we would need some detailed modeling study to complement the observational data. But it's something to consider!