It's straightforward to show that this "pause" in the LT is purely a product of the 1997-98 El Nino -- that is, of natural variability. It would not exist without it, if everything else stays the same.
In the following graph I plot what I call the "reverse trend" -- the trend from any year on the x-axis to today. So, for example, the graph shows that the trend from 1983 to 2015 is 0.12°C/decade, and from 2003 to today is 0.04°C/decade. (There are error bars, but the point made here doesn't depend on them so I'll leave them off.)
The red line is the actual measured temperature anomalies. (I used annual averages to keep this simple.) The blue line assumes simply there was no big El Nino spike and that 1998's anomaly was the average of 1997's and 1999's.
As you can see, the "pause" for the actual data -- the red line's dip to slight negativity in 1998 -- depends closely on choosing 1998 as one's starting point. That's precisely the definition of a "cherry pick" -- choosing the starting point to give the result one wants, regardless of its scientific legitimacy or whether it is long enough to represent climate, and not natural variability.
The blue line, with no 1997-98 El Nino, shows no pause at all at that point. In fact, it's never negative for any starting point before 2015.
As expected, the trend over climatically representative intervals, such as the 30 year period 1985-2015, is essentially identical to that which includes the natural variability of that big '97-98 El Nino. El Ninos and La Ninas balance out over the long-term. (They're not the only natural factors that can have an influence, of course -- volcanoes and changes in solar irradiance do too -- but no need to include them here to make the point.)
And shorter trends, of about 10 years or less, aren't statistically significant, and usually not even close to being statistically significant.
The lower troposphere "pause" over 18 yrs X months is purely an artifact of a big El Nino -- that is, of natural variability.
Note added 9:00 pm: A commenter asked a good question: what about the other El Ninos that have occurred since? The average of the monthly MEI (Multivariate Enso Index), an ENSO proxy, from Jan 1999 to present is 0.04, very close to neutral. (For example, 1997's MEI was 1.58. 2015's is 1.59.
In other words, La Ninas and El Ninos have tended to cancel out since 1999. So it's not a bad assumption to take the data as it is, excepting only the 1997-98 El Nino. But it's not a perfect assumption, as one would still need to account for ENSOs before 1999 -- the average MEI from January 1979 to December 1996 is 0.48.