5.2 Roots and Stems in Invented Word Families

Finding the stable center

A root is the central meaning-bearing part of a word. A stem is the form to which additional endings attach in a particular pattern. In simple cases the root and stem look identical. In harder DLAB-style items the stem may include a connector, a theme vowel, or a slightly altered shape before endings are added. The core habit is to locate the stable center before interpreting the moving pieces around it.

Imagine an invented system where dar means carry, daro means carrier, darok means carriers, and predar means recarry. The root dar is stable across all four. The ending -o appears to build an agent noun; -k marks plural after that noun; the prefix pre- means again. You do not need to know whether this resembles any real language. The point is to reason strictly from the examples in front of you.

Roots can be hidden by small, predictable changes. One item might show lum, luma, lumak; another might show feb, fev-a, fev-ak. If the data suggest b becomes v before a vowel suffix, the root idea is still recoverable. Do not assume every changed letter signals a new morpheme. Ask instead whether the change is predictable from its environment, which is far cheaper to track than a brand-new meaning.

Root-finding checklist

A root-first approach keeps reasoning efficient. Start from the longest word and you risk overfitting to one form; start from the repeated core and you can attach meanings to the extra pieces in order. This also sharpens elimination: an answer that quietly changes the root meaning without evidence is weaker than one that changes only a prefix or suffix.

Cross-root comparison is the guardrail

Be especially careful with English-looking forms. In an invented item, a root spelled pan need not mean a cooking pan, and an ending -ed need not mark past tense. Treat every form as internally defined by the item itself. If the prompt states pan = red and panol = redness, then pan is a color root in that system and -ol builds an abstract noun, full stop. Importing English meaning is one of the fastest ways to lose points.

The strongest answers preserve the whole word family. Suppose mar = write, maren = wrote, silar = sing, silen = sang. The ending -en marks past time, while mar- and sil- carry the actions. Any answer claiming -en means "write" collapses immediately, because -en also appears with sing. Testing a candidate affix against a second root is the single most reliable check you can run, and it costs only seconds.

As you work, write compact labels rather than full translations: ROOT, PL, PAST, DOER, AGAIN. These short tags reduce working-memory load and stop you from silently importing English grammar. On a roughly two-hour, 126-question exam where qualifying scores commonly start near 95 (and the Air Force expects about 110 for its hardest categories), shaving seconds per item by labeling instead of translating directly protects your score.

Worked family: stripping endings from the right

• Given: kel = walk, kela = walking, kelak = walking things, kelaket = many walking things.
• Strip rightmost first: -et recurs as "many/plural" → remove it, leaving kelak.
• -ak recurs as a noun-forming "thing" → remove it, leaving kela.
• -a recurs as the progressive "-ing" form → remove it, leaving the root kel = walk.

Working right to left in layers keeps each rule isolated. If a layer fails to recur on another root, you have caught the error early instead of building an answer on a false split. That layered discipline is exactly what separates a clean rule from a lucky guess.

Predictable sound changes versus real differences

One subtle skill is telling a predictable sound change apart from a genuinely different morpheme. Suppose an item shows tep = cut, teba = cutting, teban = the cutting (a noun). The root looks like it shifts between tep and teb. If every form with a following vowel uses b and every form ending in a consonant uses p, the change is predictable from the environment, and the root is still a single morpheme. You should not invent two roots to explain one rule.

Contrast that with tep = cut and dep = sew, where the first letters differ but the change is not predictable from any environment; here the difference signals two distinct roots. The deciding question is always the same: can the variation be predicted from its surroundings? If yes, fold it into one root and move on. If no, treat the forms as separate morphemes. Spending a few seconds on this test prevents a cascade of wrong answers, because a candidate who mis-splits a root early will mislabel every affix that follows it.

On a 126-question exam, that single discipline protects more points than almost any other morphology habit you can build.