There are probably other reasons as well. For a very long time IBM treated the PC as a toy business; from their perspective, the "real" money was made on mainframes and mini computers. And to some degree they were correct- IBM was making so much money that a career at IBM was so guaranteed that IBM didn't pay unemployment insurance because they never laid anyone off.

I'm not saying they didn't make money on PCs, I'm just saying that the genious IBM Boca Raton team that created the PC (RIP team lead Don Estridge) was initially considered secondary to the company's core business. From IBM Corporate's perspective, the PCs primary value was as a smart terminal that a customer might use to connect to their mainframes or system 36 or AS/400s, or something that a customer might want to run a spreadsheet on.

So back to the point- How does this apply to your observation of PL/1 vs Pascal? Pascal was very popular on the PC (in one form or another). My school taught UCSD Pascal (required for my mechanical engineering major), but hanging out in the computer lab, I very quickly ditched it for Turbo Pascal, which came with a glorious IDE and could be bought for $49, which was not pricey even in the mid 1980s. PL/1, considered a "big iron" language was probably available on a PC, but I can guarantee you it wouldn't have cost $49 bucks.

Turbo Pascal, Quick Basic, C, and surprisingly dBase went on to dominate the PC development landscape through the mid-late 80s, and those "toy" PCs had very sharp elbows when it came to carving out market share.

And then there's the performance. In an article from that era that compared Cobol to PL/1, Cobol was up to six times faster. Now let's take that with a grain of salt: language wars then were no less passionate then than language wars now. PL/1 was the new kid on the block and Cobol was the language that institutions, particularly large financial institutions, ran their business on. Cobol proponents had every reason to want to make the newcomer look bad.