Jill, if you are looking to create a new map that shows the change in soil type, you'll have to choose the output type: vector or raster. If you choose vector, you will need to convert the raster layer to a vector layer that we'll call V1. You will then have two vector layers V1 and V2 (the latter being the original vector layer). Next, you will want to recode the attribute values for one of the layers by multiplying those values by ten or hundred depending on the original value range. For example, if the original values are 1,2 and 3, your new values will be 10,20 and 30. Remember that you are doing this for just one layer--V1 for example. Next, you will merge the two layers. This may create undesirable "slivers" which results from boundaries that should overlap but that do not. There are many ways to deal with slivers, a search for that term on this forum will point you to techniques that deal with them. Finally, you will add a new field to your new merged layer where you will calculate the sum of V1 and V2 attributes (whose values will be added to the new layer during the merging process). This will create values such as 11,12,13,21,... etc. A value of 11 for instance indicates no change in soil type. A value of 12 indicates a change from 1 to 2, etc...
If you choose the raster route, you will want to recode the values from one of the two rasters. You can use map algebra for this (e.g. [R1] * 10) which will create a new raster we'll call R10. Using another map algebra calculation, you will "add" the rasters R10 and R2 ([R10] + [R2]). The resulting raster will have the same new values as those for the aforementioned merged vector layer. In fact, you can accomplish this in just one step: [R1] * 10 + [R2].
Now from here, I'm not sure how you want to quantify the distribution of these attributes. Are you seeking a measure of dispersion/clustering of direction type, or just an area summation of each directional attribute value?
