Append to a dict of lists with a dict comprehension

I'd use filter:

>>> words = ['abcd', 'abdef', 'eft', 'egg', 'uck', 'ice']
>>> index = {k.lower() : list(filter(lambda x:x[0].lower() == k.lower(),words)) for k in 'aeiou'}
>>> index
{'a': ['abcd', 'abdef'], 'i': ['ice'], 'e': ['eft', 'egg'], 'u': ['uck'], 'o': []}

It is not possible (at least easily or directly) with a dict comprehension.

It is possible, but potentially abusive of the syntax, with a set or list comprehension:

# your code:    
d={}
for word in words:
   if word[0].lower() in 'aeiou':
       d.setdefault(word[0].lower(),[]).append(word)        

# a side effect set comprehension:  
index={}   
r={index.setdefault(word[0].lower(),[]).append(word) for word in words 
        if word[0].lower() in 'aeiou'}     

print r
print [(k, len(d[k])) for k in sorted(d.keys())]  
print [(k, len(index[k])) for k in sorted(index.keys())]

Prints:

set([None])
[('a', 17094), ('e', 8734), ('i', 8797), ('o', 7847), ('u', 16385)]
[('a', 17094), ('e', 8734), ('i', 8797), ('o', 7847), ('u', 16385)]

The set comprehension produces a set with the results of the setdefault() method after iterating over the words list. The sum total of set([None]) in this case. It also produces your desired side effect of producing your dict of lists.

It is not as readable (IMHO) as the straight looping construct and should be avoided (IMHO). It is no shorter and probably not materially faster. This is more interesting trivia about Python than useful -- IMHO... Maybe to win a bet?

No - dict comprehensions are designed to generate non-overlapping keys with each iteration; they don't support aggregation. For this particular use case, a loop is the proper way to accomplish the task efficiently (in linear time).