diff --git a/README.md b/README.md
index d111abe3a94a78d48f301fd10ba4e376da6c275a..ce871d082605e224716d1ec67f0409f640603d72 100644
--- a/README.md
+++ b/README.md
@@ -3,14 +3,15 @@ performance is critical is in terms of a collection of n+1 files containing
 code in the C programming language: n "module" files with a ".c" suffix that
 contain the bodies of C functions and procedures, and one additional shared file
 with a ".h" suffix that contains definitions of global variables and types that
-are used by each of the modules. LDI, short for "logical data integration", is a
-collection of tools that make it possible to define this behavior by first
-substituting the ".h" file with a relational database schema abstracting all
-data relevant to the process or thread, including local heap memory data with
-the low level interface defined by the ".h" file, and then second, by appealing
-to a variant of the static embedded SQL protocol coupled with C in coding each
-of the n modules as a file with a ".sql" suffix.
+are used by each of the module files. LDI, short for "logical data integration",
+is a collection of tools that make it possible to define this behavior in
+another way. First, the ".h" file is replaced with a relational database schema
+that abstracts all data relevant to the process or thread, including local heap
+memory data with the low level interface defined by the ".h" file. And second,
+each of the ".c" module files is replaced with a ".sql" version in which all
+data manipulation in the bodies of C functions and procedures is re-coded by
+using a variant of the static embedded SQL protocol.
 
 Note that each of the tools comprising the LDI system define the behavior of
 such a process or thread, and are therefore also coded in this manner. Thus, the
-tools themselves also constitute important benchmarks for their own evaluation. 
\ No newline at end of file
+tools themselves also constitute important benchmarks for their own evaluation.
\ No newline at end of file