Update README.md

grammar changes

Fdic/README.md

@@ -27,7 +27,7 @@ FrequencyDictionaryIO.writeFdic(dictionary, "en-80k.fdic")
 ```
 
 ## Performance
-Performance varies greatly depending on the combination of machine and dictionary being decoded. But `fdic` always
+Performance varies greatly depending on the combination of machine and dictionary being decoded. But `fdic` is always
 superior in both size and speed to both a plain text dictionary, and a GZIPed text dictionary.
 
 Some Machine/Dictionary combinations show more modest speed improvements, but considering simply GZIPing would have
@@ -111,15 +111,15 @@ Integer permission to represent the frequency for `is`.
 Pretty quickly the terms require fewer than 8 characters to represent their frequency, so representing it as a binary
 number begins to increase the overall size of each entry.
 
-To solve this I came up with `Variable Length Longs` which only take as many bytes as necisary, the smallest numbers
+To solve this I came up with `Variable Length Longs` which only take as many bytes as necessary, the smallest numbers
 requiring just 1 byte when encoded.
 
 ### How
 This is achieved by using the **Most Significant Bit** (_MSB_) as a `Continuation Flag` when parsing. This means that each
 encoded byte only represents 7 bits of data, meaning in worst case scenarios we could use up to 10 bytes to represent
 an 8 byte Long. In this use case it never occurs, as term frequencies are never that large.
 
-When we begin reading a `vlong` feild, we mask out the continuation bit, take the 7 data bits, shift them depending on
+When we begin reading a `vlong` field, we mask out the continuation bit, take the 7 data bits, shift them depending on
 how many bytes we've read so far for this vlong, and add that to an accumulator Long.
 
 ```mermaid