Using word searches and context to reject spurious word matches
English grammar, vocabulary, and context and sensibility restrictions greatly narrow the number of possible words at any one position. Thus the Ramey message needs to be read linguistically, not simplistically letter by letter. For exactly this reason, debunkers typically avoid the linguistic considerations and focus on letter-by-letter ambiguity. But linguistic analysis can often eliminate the ambiguity and return only a few, and maybe only one likely word. This is standard cryptographic practice. (An example of using linguistic analysis in cryptography can be found in this New York Times article. Another example is reconstructing messages by listening to the sounds of dot matrix printers.)
This is illustrated in the following two graphics of the critical "VICTIMS" word. Why is it very likely VICTIMS and not some other 7-letter English word?
This is actually one of the clearer words in the message. Only the fourth letter is so light and poorly formed that it is impossible to match with anything. We can use the fairly distinct letters in the word to greatly narrow down the number of possible words from the thousands of possible 7-letter English words. Here the search letters VI_ _I_S are used and put through the search engine at onelook.com to find possible word matches in the English language, including possible acronyms. We also know from grammar of the short English words around this word ("THE ??????? OF THE") that this must be a noun.
The graphic shows the 10 possible noun matches with these search letters. When historical context and sensibility restrictions are applied, the only match that makes any sense is "VICTIMS". Would a military message about Roswell contain words like VIRGINS, VIOLINS, VIROIDS, etc.? How many people even recognize highly obscure words like VIBRIOS, VILNIUS, VIGLIUS, or VINNIES? (The slang and regional "VINNIES", what people in Seattle call St. Vincent DePauls thrift store, didn't even exist back in 1947.)
If we expand the search by throwing out the more problemmatical search letter "S", several dozen other possible matches turn up, but only one might make any semantic sense: VIEWING. All other matches yield nonsense.
The first letter would also be strongly matched by "Y". The second graphic below shows a similar word search using only the "Y" and the two "I"'s. There are far fewer possible 7-letter "Y" words and here NONE of them make any possible sense. I have also done searches using a less probable "E" for the second letter and also end up with nonsense words like YESHIVA. Again one has to read the message using linguistic constraints, not letter-by-letter.
The same is true for other consensus words in the Ramey memo, like FORT WORTH, TEX., THE "DISC", and WEATHER BALLOONS. In isolation, letters in these words are often unclear and ambiguous, but people don't read letter-by-letter but in words and groups of words. This plus applying historical context and linguistic constraints make these the only possible words for these positions. (Click here for a demonstration of how good we are at completing words from fragmentary letters and context.)
Also see companion graphic for comparison with four other proposed mundane words for "VICTIMS" that might make sense (at least in isolation) and judge for yourself which one best matches the actual letters by eye.