Words and morphemes

Words may consist of parts that carry parts of the meaning. These parts are called morphemes. They can be verbal and nominal endings, which signal grammatical functions such as person, number, gender, state, tense, mood. They can also be pronominal suffixes, which are almost separate lexemes.

The BHSA dataset distinguishes the following morphological elements in a word:

• prefix - lexeme - suffix - enclitic

• vowel-pattern

The information in each of these elements is carried by one or more features in the BHSA dataset. These features are also represented by a single shortcut character, which shows up in some feature values, which is why we mention them here.

prefix

pfm	!	preformative
vbs	]	verbal stem (root formation)

lexeme

lex	word (as dictionary entry)
g_word	word (as occurrence in the text)

suffix

vbe	[	verbal ending
nme	/	nominal ending
uvf	~	univalent final

enclitic

prs

+

pronominal suffix

vowel pattern

Not present as a separate feature in the dataset.

All these features encode the information that is encountered in the text. So, although the values carry grammatical information, these features do not label the grammatical information.

There is, however, one level of abstraction: the morpheme occurrences as they occur in the text, also called the graphical forms, are grouped around abstract, paradigmatical forms. The paradigmatical forms come close to specifying grammatical information, but usually they do not do that on their own, but the information of several paradigmatical forms must be combined to arrive at a definite grammatical label, if this is possible at all.

If you need to now the grammatical label assigned to a word, e.g. gender = f, or state = a, you need to use other features:

gn prs_gn	gender	m f
nu prs_nu	number	sg pl du
ps prs_ps	person	p1 p2 p3
st	state	a c e
vs	verbal stem	qal piel nif hif
vt	verbal tense	perf impf wayq