Determining Input Values for Results of Convolution

How can I get back the discreet values that made each value in a convolution without using the original convolution file. I am looking to count occurrence of quadruplets of letters and spaces in a large piece of text. I can associate each discreet value with its occurrence, however, I do not know how to get the four discreet values that made the convolved value to begin with. here is my code thus far:

A = fileread('text_file');

B = abs(A)';

x = 27;

C = [x^0;x^1;x^2;x^3];

Convolution = conv(B,C);

Pruned = Convolution(numel(C):(numel(Convolution)-2));

N = histc(Pruned, [0:1:max(Pruned)]);

Location = find(N);

M = N(Location);