I've always taken a haphazard approach to conlang building, and usually end up painting myself into a corner and watching the project stall. I will coin a few hundred words more or less at random, and then come up with a morphological idea when it's too late because I don't want to have to scrap those hundred or more words that don't fit the new morphology.

Plus, the task of coining new words always seems kind of hit and miss to me. My conlang sketches seem to lack a coherent feel to their lexicon.

So for the last few weeks I've been toying around with ideas for a systematic morphology for a brand new conalng (known so far as "noname"). The plan is to work out all the details of the morphology before anything else so that I can make changes and modifications and do experiments before ever coining the first word for the conlang.

The preliminary design I've worked out so far looks like this:

1. Each part of speech has a specific ending. (The language will be non-inflecting, except perhaps for verb tense, noun plurals, and maybe possesives.)
2. Words can be changed from one part of speech to another by changing the affixes. (sad -> sadly -> sadness -> sadify (ensadden?) -> sadificationism -> sadificationistically ;-) ...)
3. The four principle parts of speech (noun, verb, adj, adv) will each end in a specific vowel.
4. When changing the ending, both the vowel before and the vowel after the final consonant will be changed. This is to make the spoken sound of the word different by more than just a single vowel, and thus more distinctive. For example, instead of changing a word like "kamo" to "kamu" both vowels would change, more like "kamo" to "kimu", or maybe "komo" to "kiamu".
5. Roots will contain at least two consonants, the last of which will be the "detached consonant", e.g. "bas-n-" where "-n-" is the detached consonant.
6. For each possible detached consonant there will be a unique permutation of the vowels which defines the pattern for roots ending in that consonant. For example: (very preliminary)

K: -ika -eke -oki -ako -uku
  e.g.: aj-k- -> ajika [n], ajeke [adj], ajoki [adv], ajako [v], ajuku [?]
L: -ela -yale -ili -ulo -olu
  e.g.: upan-l- -> upanela [n], upanyale [adj], upanili [adv], upanulo [v], upanolu [?]
M: -uma -ome -wami -imo -emu
  e.g.: k-m- -> kuma [n], kome [adj], kwami [adv], kimo [v], kemu [?]
N: -ana -une -eni -ono -inu
  e.g.: bas-n- -> basana [n], basune [adj], baseni [adv], basono [v], basinu [?]
T: -iota -ite -uti -eto -atu
  e.g.: samb-t- -> sambiota [n], sambite [adj], sambuti [adv], sambeto [v], sambatu [?]

(I'm not sure what words ending in -u are. Any suggestions? Vocative maybe?)

There would be as many vowel permutation families as there are possible "detached consonants", including clusters such as roots like: rot-nd- (rotunda, rotando, ...) or (in the extreme) pan-mpst- (panempsta, panompste). I need to draw up a list of which clusters will be permitted. I think it will probably be a pretty short list NOT including clusters like -mpst-.

(It might be interesting to have the vowel permutation families represent different semantic classes (somewhat Bantu-ishly) like "sentient creatures", "things of the sky or heavens", "food animals", "natural phenomena", "things supernatural", "liquids", etc. But that's not really part of the morphology, so I'm getting ahead of myself.)

Of course not every root will have a word for every form. Some forms just aren't very useful (like an adverb form of "yellow", for example.)

Then I want to work out a complete set of prefixes and suffixes for doing derivations. For example (hypothetically):

ma(n)- "more" (as in big -> bigger)
maha(n)- "most" (as in big -> biggest or "raja" -> "maharaja")
-ria - "[n] place to find X" (as in pizza -> pizzaria)
-(i)ta - "secondary noun naming a property" (as in acid -> acidity)
-(i)sta - "person who performs and action" (as in fly -> pilot)
-dina - "diminutive of a noun" (flora -> floradina)
(These are all just hypothetical since I haven't really started working on them yet.)

So far I've enumerated just over 90 different ways that one word can be changed into another word or part of speech along the lines of destroy -> destruction -> destructive -> destroyer -> destructible -> destructibility -> destructively -> ... or acid -> acidic -> acidity -> acidify -> acidifier -> acidification -> acidness -> acidly -> ...
(The list so far is here: <> )