I believe the problem is a combination of the following (and likely a few others):
1. Borland was not really interested in making the language genuinely better. They were interested in tacking on features they thought would be profitable even if those features introduced problems/ambiguities in the language.
Well this is the investments bit that I already named. But I don't think it would have mattered much. Language is only small part of the package, and I think even that most of the reasons were not technical.
Borland was already in decline while MSVC was vastly inferior. Before say VS2005, the Delphi offering was still way more productive than the .NET 1.1 or MFC solutions that VS packed.
Something that gradually changed with VS 2005 and .NET 2 for the .NETties, but MS push for .NET meant another handful of years before MS would focus on native again (not a picknick for C++ MFC app programmers either)
2. Borland didn't promote the standardization of the language. On the contrary, they bastardized it for financial gain. They turned Pascal into a competitor to VB (that's why BOOL sets true to -1, to be compatible with VB - such a "useful" feature.)
It was the dominant convention when the winapi was invented. VB only copied that. C99 only made inroads way after 2003, and even later on non open source unix, when the systems we are talking in this discussion were already established for a decade or longer.
Back in the days of Turbo Pascal, Pascal was a worthy competitor to C (at least on MS-DOS), that is definitely not true today (on any platform.) FPC is a much better choice as a replacement for C/C++ than Delphi.
Same problem IMHO, no big backer. Even smaller even. But nobody said FPC should go after the corporate market.
Standardizations are IMHO vastly, vastly overrated. Successful standards only happen in exception circumstances, when there is an oligopoly of large vendors, and/or some large customers (like US government/military) force languages on vendors. And basically that means C and C++, and the rest failed or is window dressing at best.
Also qualitywise most of the those are not worth the paper they were written on.
The only thing is that maybe Borland would have developed with a longer futureproof outlook if there was a standard. But you don't need a standard for that, just some management sanity.
3. They took way too long to introduce "basic" features such as support for 64bit integers, unsigned 32bit types, initialized local variables, 64bit code generation and many others. Not to mention that they found nothing better to do than force some of their choices down their customer's throats e.g, you will eat our Unicode chars whether you like it or not, if you don't you have to go through all of your code and change the pchars to pansichars - no doubt everyone looked forward to that.
x86_64 only started emergency at the end of the decisive period 2000-2005, the unicode stuff much more. But by then the influx of new customers had already dried up, and they kept everything compatible in the hope to seduce people to upgrade. However that backfired (not enough innovation) so with D2009 they had to do something drastic
4. While MS was lowering the price of VS (to the point that it is now basically free for those who don't need support), Borland steadily increased the price of Delphi, nowadays I find it surprising that there are people who pay over $1000.00 for that language.
True, but that is a result of MS not needing the money like Borland did. Getting cheaper would only have motivated MS to make it wholly free, and what are you then going to do as paid-for company?
Unfortunately, all of those problems and, there are more, reflected on Pascal as a language. That's very unfortunate.
The problem is that the whole reflection is an amalgam of sentiments. From old Unix hurts to teenagers were forced to do 16-bit pascal when 32-bit C++ was more l33t.
I already answered the standards bit. I don't buy it for one bit, and I think language reasons are the least important ones anyway, let alone trivialities like standards (otherwise VB6 would never have been the most popular language in the world).
The absence of standards is just a stick C/C++ use in their argumentation, because that is one thing that is ok beyond discussion in that world. If the situation would be the other way around, we'd make fun of all the funny little characters in C and C++ syntax, and then conclude that any language without a module system is stone age and was declared dead in 1995 too.