I would allow players to make the spell cover any shape they wanted, provided that shape is simply defined, contiguous, has no "islands" or "doughnut holes", and no part of it is further than 60ft from the point they touch while casting the spell. By "simply defined", I mean simple shapes (cube, sphere, etc), or anything that can be expressed simply and obviously, like "inside the house" or "the area under the bridge".
I came to this ruling because it leaves maximum scope for the players to do what they want, while minimising the number of ways that it can become complicated, boring or overpowered.
Analysing the rules
Generally, where things are vague in 5e, the best way to approach it is thus:
- Do other rules have anything related to say?
- How would players generally be expected to use it?
- Which interpretation causes fewer complications?
- Which interpretation protects against abuses or ways to bypass balancing mechanisms (action economy, spell slots, etc)?
Related rules
The section on spellcasting doesn't mention how to handle spells that don't use one of the listed shapes. However, the alarm spell can be cast to cover "an area within range that is no larger than a 20-foot cube", which means there is precedent for choosing your own shape of area, within certain size constraints.
Expected use
I'd expect players to use hallow to protect areas like the inside of a place they're staying in - a house, a castle, a cave, whatever. This could be any shape, really, and as a DM there's no reason to want to spoil that for them.
Complications
Circles don't tessellate, and the players may wish to cover larger areas. Handling the bits of a circle that protrude out through the walls of a square house may be tricky (if you rule that this happens).
Balancing
The 24-hour casting time and 1000GP cost protect against serious abuses, and with those in place, I can't see a particular way that one shape is "overpowered" compared to another.
Ruling
Given these observations, I see no problem in letting players choose any shape they like. Of course, the spell says "you touch a point, and infuse an area around it", so no part of that shape can be more than 60ft from the point they're touching. This means that to cover a corridor that's 120ft long, you'd have to cast the spell from the middle. For the sake of keeping things simple, I'd probably also say that the area can't have "islands" in it - for example, you couldn't hallow a doughnut shape; you'd have to cover the bit in the middle. (This also prevents you from using hallow to trap things in the middle, which is the only thing I can see that might be open to abuse.) Likewise, you can't use one casting of hallow to cover two separate but unconnected areas, even if both of them fall within the spell's range. The hallowed area must be contiguous.
Reasoning
Letting the player pick the shape covers their intended uses, while also preventing situations that are either fiddly to manage or detrimental to the game's balance. In other words, it streamlines the playing of the game without blocking anything that's fun or adding anything that makes it less fun - which is the mark of a good rule.
For some examples, consider the following scenarios:
- The players want to hallow the inside of a small house, without having gaps at the corners or weird curved segments of hallowed ground outside it. Letting them choose the shape lets you divide it simply into "inside = hallowed" and "outside = not hallowed".
- The players want to hallow a larger area. Letting them pick the shape lets them choose a shape that tessellates. If you're playing on a square grid, they can hallow cubes; if you play on a hexagonal grid, they can hallow hexes. If they want, they can pick other shapes that fit the terrain, like "up to the edge of the cliff" or "120ft of road".
- The players want to hallow a small building, but the building is shaped such that they can't see all of it at once (perhaps three sides of a square). Letting them pick the shape lets them hallow it all in one casting provided that it's small enough. Given two buildings of equal floor area where one can be hallowed in one casting but the other takes several because of its shape is needlessly complicated for little gain in entertainment value.
In general, the larger the fight, and the more tactical you wish to allow your players to be, the more helpful a battle map will be.
But, here are some suggestions that I and another DM friend of mine have used in the past, when Theater of the Mind made more sense.
Make a small DM map.
It doesn't have to be to scale, and it doesn't have to be perfect, but if you sketch out a rough schematic of the fight for your use, you will find it much easier to keep track of things. Then you can add things like entangle to this schematic.
Bonus points for this idea because it can be done on the fly, and you don't feel any pressure to draw well, or have the map be fancy, because you are the only one who will ever see it. You can use X's for the bad guys, and O's for the good guys (with the first letter of their name inside the O so you know who's who).
Group your enemies.
Keep your monsters in 2-3 member squads, and have each squad have the same initiative, attack bonus, AC, ST's, HP, etc... This way, you have fewer monster locations to keep track of. This basically makes them act like a bigger monster, who takes damage from AoE spells multiple times, and reduces the # of attacks they have at regular intervals (as one of the creatures within the squad dies).
Be warned: this makes the combat more swingy than having each monster go on their own initiative. If all of your monsters go at the same time, or small groups go togeather, they will do lots of damage at once. If they manage to go before the players, they may do more damage than they normally would, as the PCs may have been able to take out one of them before they went.
On the flip side, if they go after the players, the PCs may get to do more damage, and kill more of them than would have happened if you had them all go on different initiatives.
Not a huge deal, but something to keep in mind.
have your players announce intentions rather than actions.
Since players have less information in TotM than they do a battlemap, you are highly encouraged (here and elsewhere), to track intentions along with the spell used, rather than just actions. If your players want to block monsters A-C from closing in on their squishy party members, and they want to used entangle to do that, then have them say "I cast entangle to block monsters A-D from moving towards PC's X and Y", rather than saying "I cast entangle in front of monsters A-D". This way, you don't have to remember the location of the spell, only that Monsters A-D can't approach PCs X and Y without making the necessary saving throws, taking damage, etc...
This will reduce the amount you have to remember, and make it easier to keep track of the fight.
My personal suggestion is the DM map, but that is because I enjoy tactical combat as a player. So as a DM, I tend to feed that to my players. I run my games with facing rules, lots of cover, flanking, lots of AoE/control spells, etc... But then again, I'm the type of DM who is going to gravitate towards a battlemap for anything more than a 3v3 skirmish.
Best Answer
Doing this mechanically RAW, is probably impractical:
Plant growth is a third level spell on both the Druid and Ranger spell list, so available at 5th level, at the earliest. A character casting this spell repeatedly, might be able to create the effect that you want to achieve.
As you can see, there are two ways it can be cast, with effects that I have assumed are exclusive to each method of casting (i.e. the 'eight hour' casting doesn't additionally produce the effects of the 'one action' casting).
You asked for a means of creating 'near impenetrable vegetation'. Plants that slow movement by three-quarters, seem to meet that criteria.
This spell also occurs instantly, and has no stated conditions (unlike Entangle) by which the vegetation later withers. So the limiting factor is how many times you'd need to cast the spell to cover your desired area.
You stated that you'd like this area to be 'a few miles in radius', so let's start by calculating how many castings of the spell would be needed per square mile.
Multiple Druids and or Rangers working together, could obviously achieve this effect more quickly. However, any high level casters you add to your game add extra complication to it.
An easier and perhaps more reasonable solution would be to adjust your expectation of how large this area needs to be. What are the motivations of the characters that are spending all day, every day casting this same spell over again?
If you go with the eight hour casting of Plant Growth rather than the one action casting you can affect a much wider area each time. A circle of half mile radius is equal to 0.79 square miles. Therefore it would be easy for one relatively low level caster to affect an area 'a few miles in radius' over a period of months.
This method of casting does not produce difficult terrain. So, if that's what you want when you say 'near impenetrable vegetation' then this solution will not work for you. As DM, however, you could rule that the 'enriched' growth of the plants still makes them a good place for someone nefarious to hide. If all you need is a good hiding place then this could be enough.
But why not forget about RAW, and homebrew a solution?
There are lots of ways that you could achieve the effect that you want that would be narratively satisfying, without being bound by a RAW approach.
Any of the things that you've mentioned in your question ('spells, items, gates, otherworldly entities') could be a potential avenue for a solution to your issue, and narratively exciting for your PCs at the same time.
How something in your world has happened is up to you as a DM, but to get there you don't always need to be rigidly bound by RAW.
It's perfectly acceptable to use the Plant Growth spell as a jumping off point for your narrative inspiration, without having to worry about the players wanting to be able to replicate the exact situation themselves. NPCs may be able to do things that players can't and your PCs will almost certainly have skills that your NPCs don't.