There are a fair number of item and test analysis programs around, as you may well know.
If asked to name the two primary features which might distinguish Lertap from the others, I'd name these: it's an Excel application, and it'll score just about anything.
You've already seen examples of how Lertap leans on Excel for charts. Behind the scenes, Lertap uses Excel's in-built functions whenever it can, which is often. For example, whenever Lertap has need for a correlation coefficient, a macro tells Excel where the data are, and gets the Excel CORREL function to do the calcs. You want Lertap to calculate basic descriptive stats for a test score? No worries -- again Lertap tells Excel where the basic data are, and standard Excel functions, such as AVERAGE, STDEV, MEDIAN, KURT, and SKEW, are hauled in to do the work. The in-built functions are fast, optimised for Excel.
If you know how to use Excel, you can access the in-built functions yourself. Instead of a correlation matrix, for example, you might prefer a covariance matrix -- Excel's COVAR function may suit your needs. (Use Excel Help to find out more about its functions. There are many.)
Item weights? Item scores? Lertap has strengths here. Any item can use up to thirty (30) response codes. The codes themselves may be digits, or upper- or lower-case letters. A true/false item, for example, could use response codes of {t,f}, {T,F}, or perhaps {1,2}. Likert items usually employ response codes of {1,2,3,4,5}, but they can be anything you want.
Items in a test do not have to use the same number of options.
Items may be pre-scored. In this case, the entries in a Lertap Data worksheet become the actual number of points awarded for a person's answers to the items. (Usually the entries just indicate which option a person took.)
Any item response may have a non-zero score attached to it. This means, for example, that a cognitive item may have more than one correct answer. Likert items are very easily reverse-scored. Semantic-differential items may be scored on a -3 to +3 basis, or, for that matter, on any other basis.
Any item can be "credited" to all and sundry; no matter how a person has answered an item, s/he can be credited with a "correct" answer. This even applies to the people who didn't answer the question in question.
Lertap's "Halve & Hold" option, available from the Run menu, randomly divides a data set into halves, allowing for calibration and validation sampling.
Response similarity analysis, RSA, is supported in Lertap versions 5.5 and later. RSA is used to detect the possible presence of cheating.
DIF, differential item functioning using Mantel-Haenszel methods, was introduced in the September 2009 edition of Lertap for Excel 2007, version 5.7.5.
Coming back to the matter of Lertap being based on Excel: if Excel's functions are not enough to do the special job you may have on hand, chances are fair that a search of the Internet might uncover Excel macros which may meet your needs. For example, the free Matrix.xla Excel Add-In from the Foxes Group in Europe will support a principal components and a principal factors analysis using one of Lertap's standard correlation matrices as input. Lertap's DAT worksheet is made for programs wanting ASCII-type input with fixed fields, such as Bilog-MG.