May 12, 2004

Hebrew Morphology

Over the years I have read descriptions of many languages, and seen many weird and wonderful things. But nothing I’ve seen comes close to Semitic morphology – it reminds me of nothing more than a multiplication table, with roots along one axis and patterns along the other. Each one has a meaning: combine the two and you get a word. It seems like an impossibly elegant system for creating words; I can’t imagine how it could have evolved. No wonder the speakers of Hebrew and Arabic (both Semitic languages) consider their languages divine. I will speak specifically about Hebrew, but the concepts apply to all Semitic languages (though the specific roots and patterns may be different).

A Hebrew root consists of three consonants. It is not a morpheme in the sense that by itself it has a meaning – that’s why when I give the meaning of a Hebrew root I say “basic meaning”. This is my generalized sense of the unifying concept of all words that have that root. It only takes on a meaning when combined with a pattern. For example, consider the root q-l-t; it appears in the following words:

qalat – to take in
niqlat – to become acclimatized
hiqlit – to record

haqlata – recording
maqlet – receiver
miqlat – refuge
qaletet – cassette
qelet – input
taqlit – record

Notice that they all have the letters q, l, and t, inserted into different patterns of other letters. Notice also that these words all have something to do with “taking in” or “being taken in”. Now consider the following:

ma`der – hoe
mafret – plectrum (pick)
magber – amplifier
maqlet – receiver
mahshev – computer
maqrer – refrigerator
masreq – comb
matleh – rack
mavreg – screwdriver
mazleg – fork
mazreq – syringe

Notice that they all have the following pattern maXXeX, where the Xs stand for root letters. Notice also that they are all tools – this pattern is a pattern for tools. Can you guess the roots of each word?

Patterns have meanings for nouns, verbs and adjectives. For example, the pattern XaXiX corresponds to English words that end in –able. The meanings of the seven verb paradigms are as follows:

pa`al – simple, either transitive or intransitive
nif`al – the passive of pa`al, sometimes active, but always intransitive
pi`el – always transitive
pu`al – the passive of pi`el
hif`il – causative, always transitive
huf`al – the passive of hif`il
hitpa`el – reflexive, sometimes repetitive, always intransitive

Each verb paradigm consists of a rather large collection of patterns, each with a specific function. For example, the following is the paradigm for lamad – to learn, from the pa`al paradigm:

Past Tense singular plural
1st person lamadti lamadnu
2nd person masculine lamadta l'madtem
2nd person feminine lamad't l'madten
3rd person masculine lamad lamdu
3rd person feminine lamda lamdu

Future Tense

singular plural
1st person elmad nilmad
2nd person masculine tilmad tilm'du
2nd person feminine tilm'di tilm'du
3rd person masculine yilmad yilm'du
3rd person feminine tilmad yilm'du

Present Participle singular plural
male lomed lomdim
female lomedet lomdot

Command singular plural
male l'mad limdu
female limdi limdu

infinitive lilmod
verbal noun l'mida

This system is productive – new words are created all the time using it; in fact you may notice that many of my examples are modern words. When I started learning Hebrew, I had a hard time believing that people could use it on an ad hoc basis, the way an English speaker can add –ize or –tion to a word as needed, which would be understood by all parties. But in time, I internalized the system, and am now productive in it. It’s amazing what the human mind can do.

UPDATE: The word for “software” in Hebrew is “tokhna”. On that basis a new pattern has been created for "–ware". It has produced:

homra – hardware (root: h-m-r – matter)
qoshha – firmware (root: q-sh-h – firm)
lomda – educational software (root: l-m-d – learning)
gonva – pirated software (root: g-n-v – stealing)

If I were to invent the word olna (root: '-l-n – tree) I bet it would be understood the way I intended!

UPDATE: More here.

Posted by David Boxenhorn at May 12, 2004 12:07 AM
Comments & Trackbacks

Wow, someone else who's interested in theorizing about language systems! It's a good thing you blogged about Den Beste or I might never have found you.

I would guess that the way the Semitic system arose originally was by analogy, after some initial pattern had developed in a few words. For example, in English we have the pair fall/fell, with the second one being a causative of the first. "Fell" in this sense was originally "fall" with a suffix, which affected the vowel of the stem. This vowel change remains the only evidence of the original suffix, after the suffix disappeared. One could imagine that this could have become a productive pattern in some alternative history of English, with pairs like call/ckell (cause to call), stall/stell (cause to stall), etc. Eventually this could be generalized quite broadly, and new patterns may be more likely to emerge once the basic idea of a consonantal root had become firmly planted in the language (such as the patterns which emerged in Hebrew in the late Biblical and Rabbinical period). I think I remember reading somewhere that there is some evidence that Afro-Asiatic languages (among them Semitic) originally had biconsonantal roots, which Semitic then developed into triconsonantal ones.

Now, how did the Semitic languages first develop the gutteral and emphatic consonants? :-)

Posted by: Adam at May 13, 2004 04:32 AM Permalink

The triconsonantal morphology of Hebrew and Arabic is descended from an earlier biconsonantal system. This is still seen in most of the other Semitic languages like Chadic. What happened was that certain prefixes became highly grammaticalized and ended up creating a 3-radical system. Most Hebrew and Arabic basic, primitive nouns are still biconsonantal. Eventually the old 2-radical system was replaced with a 3-radical one that became productive through analogy. This kind of templatic morphology is found in other languages families, although Semitic has the most productive system. But several Native American languages and Philipino languages are templatic.

Posted by: william t drewery at August 28, 2004 07:11 AM Permalink

As for the emphatics, it's not clear what their ultimate origins are but most linguist believe they are leftovers of an ejective series like those of the Caucasian language. In time, the ejectives became pharyngealized. I'm not sure where @ayn comes from, though.

Posted by: william t drewery at August 28, 2004 07:19 AM Permalink