An intuitive way to think about this is to consider a problem of arranging 0's and 1's.
Say you have n balls to distributed amongst k people with the same conditions as above. That each person should get atleast $1$ ball and the number of balls with one person can be at max $l$. Therefore the problem is as follows:
You have $n$ 1's and $k-1$ zeros to arrange in such a way that the zeros don't lie in the ends and zeros are not adjacent. This ensures the minimum 1 ball per person condition. The sum of the of the 1's between zeros is the number of balls for that person.
Example:
$$111011110110\cdots$$
Therefore $x_1=3$, $x_2=4$, $x_3=2$ and so on. There are $n-1$ holes to be filled amongst the $n$ 1's by $k-1$ zeros. And this can be done in ${n-1}\choose{k-1}$ ways.
And about each person having a maximum of $l-1$ balls condition can be enforced by subtracting the cases in which one person has $l$ or more balls. This is easy in this case because only 2 people can have 11 balls at a time$(11+11+3=25)$.