Typos in comments

compiler/deSugar/DsBinds.lhs

@@ -694,7 +694,7 @@ drop_dicts drops dictionary bindings on the LHS where possible.
          will be simple NonRec bindings.  We don't handle recursive
          dictionaries!
 
-    NB3: In the common case of a non-overloaded, but perhpas-polymorphic
+    NB3: In the common case of a non-overloaded, but perhaps-polymorphic
          specialisation, we don't need to bind *any* dictionaries for use
          in the RHS. For example (Trac #8331)
              {-# SPECIALIZE INLINE useAbstractMonad :: ReaderST s Int #-}

compiler/rename/RnPat.lhs

@@ -527,7 +527,7 @@ rnHsRecFields ctxt mk_arg (HsRecFields { rec_flds = flds, rec_dotdot = dotdot })
        ; dotdot_flds <- rn_dotdot dotdot mb_con flds1
 
        -- Check for an empty record update  e {}
-       -- NB: don't complain about e { .. }, becuase rn_dotdot has done that already
+       -- NB: don't complain about e { .. }, because rn_dotdot has done that already
        ; case ctxt of
            HsRecFieldUpd | Nothing <- dotdot
                          , null flds

compiler/simplCore/CSE.lhs

@@ -108,9 +108,9 @@ INLINE or NOINLINE. (Examples from Roman Leshchinskiy.)  Consider
         bar = foo yes
 
 We do not expect the rule to fire.  But if we do CSE, then we risk
-getting yes=no, and the rule does fire.  Actually, it won't becuase
+getting yes=no, and the rule does fire.  Actually, it won't because
 NOINLINE means that 'yes' will never be inlined, not even if we have
-yes=no.  So that's fine (now; perhpas in the olden days, yes=no would
+yes=no.  So that's fine (now; perhaps in the olden days, yes=no would
 have substituted even if 'yes' was NOINLINE.
 
 But we do need to take care.  Consider

compiler/typecheck/TcEnv.lhs

@@ -426,7 +426,7 @@ Note especially that
  * It does not extend the local RdrEnv (tcl_rdr), because the things are
    already in the GlobalRdrEnv.  Extending the local RdrEnv isn't terrible,
    but it means there is an entry for the same Name in both global and local
-   RdrEnvs, and that lead to duplicate "perhpas you meant..." suggestions
+   RdrEnvs, and that lead to duplicate "perhaps you meant..." suggestions
    (e.g. T5564).
 
    We don't bother with the tcl_th_bndrs environment either.

docs/users_guide/glasgow_exts.xml

@@ -5358,7 +5358,7 @@ is extended thus: defaulting applies when all the unresolved constraints involve
 </para></listitem>
 </itemizedlist>
 So, for example, the expression <literal>length "foo"</literal> will give rise
-to an ambiguous use of <literal>IsString a0</literal> which, becuase of the above
+to an ambiguous use of <literal>IsString a0</literal> which, because of the above
 rules, will default to <literal>String</literal>.
 </para>
 <para>