Morpheme analysis: segmentation and classification

T. Poudel morphology I: Morpheme analysis: segmentation and classification 1 MORPHEME ANALYSIS: 1 Segmenting Morpheme analysis refers to segmenting parts of words into meaningful units. Such segmented units are called MORPHEMES. A native speaker knows the different MORPHEMES of a word or string of words intuitively. Here are a few simple English sentences. Segment the possible morphemes in these examples. The dog sits on a mat. The cats sit in a mat. The dogs sit on a mat. The cat sat on a mat. The dog sat on a mat. You can segment the MORPHEMES in these sentences something as given below: The cat sits on a mat. the cat sit s on a mat The dog sits on a mat. the dog sit s on a mat The cats sit on a mat. the cat s sit on a mat The dogs sit on a mat. the dog s sit on a mat The cat sat on a mattress. the cat sit. PST on a mattress In order to realize what we know intuitively or feel, try to think of as morpheme analyzer of a language, which you know little or nothing at all. How we make hypothesis and correct them in course of learning a language. Let us illustrate this with the example from Kekchi. First solve the task, then the proposed solution chart. Note: Hyphens are used in the morphological analysis usually for the identification of boundaries between meaningful parts of words. Clitics (explained later what that is) are usually marked with the equal sign. 1 Here is highly simplified data from Kekchi. It is simple: it compares and looks for similarities and differences between the compared items. tinbeq 'I will go' One needs to separate form and meaning Here are at least three forms, meanings, including: tinbeq 'I will go' tatbeq 'thou shalt go' ninbeq 'I am going' If we compare one with the other, . tinbeq 'I will go' tatbeq 'thou shalt go' equal: t…beq different: .in… /. at. .. Important difference: The subject is active in the process of walking The subject is the speaker, in the case of in- or an addressee if at-. tinbeq ‘I will go.’ ninbeq ‘I am going.’ equal:. inbeq different: t. .... / n. .... Important difference: The act of walking is the future t- or currently in progress—relative to the talk-point n-. tatbeq ‘Thou shalt go.’ ninbeq ‘I am going.’ equal: different: beq .. tat ... / nin ... 2 Conceptual differences: (i) The subject is the addressee tatthe speaker nin-. (ii) The operation time is present tatfuture nin-. ... So two important differences, so not very good analysis has yet begun (preliminary) result: -beq nt-in-at- 'go' 'is currently in progress' 'future' (the talk-point) 'subject = a speaker' 'subject = an addressee' So word formation (expressed in the word categories and the order of their expression): Time - participants: Person - Operation Or in more technical terminology: Tense - (and number of) the agent / subject - predicate (Maybe tense also aspect: what two different but related and not always easily distinguishable categories. Tense is a deictic category but aspect makes distinctions such as PERFECTIVE vs. IMPERFECTIVE or PROGRESSIVE. To review more data than those on which the morpheme analysis is based (possibly in fieldwork): ? natbeq 'You are going.' And also a more complex example of a familiar language, German, to practice the simple method: er schaut ‘he looks’ du schaust ‘you look’ Or schau- 'look' -t SBJ: 3SG -st SBJ: 2SG -t SBJ -s 2SG null 3SG 3 er schaut ich schaue -e er schaut er schaute -t -t -e ‘he looks’ ‘I look’ SBJ: 1SG ‘he looks’ ‘he looked’ SBJ.3SG PST SBJ: 3SG if preterite? If one begins his analysis with these two forms, and nothing else is there to compare – schaut 3SG PRES and schaute 3SG.PST— tenses, one will initially assume the minimal form of difference between null vs. -e at the end of the word goes with the minimal significance difference of 'present' vs. 'past'. More comparisons prove this hypothesis wrong. Ich schaue ‘I look’ Ich schaute ‘I looked’ -t -e past SBJ: 1SG du schaust ‘you look’ du schautest ‘you looked’ -t past -(e)st SBJ: 2SG er schaut er schaute -t1 -e -t2 ‘he looks’ ‘he looked’ SBJ: 3SG.PRES SBJ: 3SG.PST PST Revised analysis for German verb Stem (- Tense: PST) - Subject: PERSON.NUMBER (. TENSE) schau-(-t-) -e - (e)st -t / -e 1SG 2SG 3SG (PRES / PAST) 4 -en - (e) t -en 1PL 2PL 3PL Compare the structure of another type of verb, where the formal temporal distinction is a little different komm- -e -st -t -en -t -en 1SG PRES 2SG 3SG PRES 1PL 2PL 3PL kam- -Ø -st -Ø -en -t -en See Nida (1949) for further morpheme analysis. 1SG PST 2SG 3SG PST 1PL 2PL 3PL The morpheme Word: “minimal free form” (Bloomfield 1933:61) A word may contain smaller meaningful units e.g. anti-dis-establish-ment-arian-ism-s In some languages a single word can have a whole sentence. Morphology deals with the identification of these smaller meaningful units that form word in a particular language. To do this usually linguists use two procedures—substitution and recurring partials. Substitution Consider: We jumped We walked. We worked. Equal: we and –ed Different: jump, walk, work The substitution brings a change of meaning. The substituted form are in contrast in the frame we….ed. Because of this contrast jump, walk and work are different morphemes. 5 Recurring partials Consider the recurrent partials in the above examples we…ed. A native speaker of English knows, the word we means something like first person plural and the second element –ed as past tense. A careful analysis by a non-native of English can infer similar conclusion. The number of morphemes of a word in one language need not be the same when translated into another. Consider the following (Sierra Populuca, SP a Native American language) ikama ‘his cornfield’ way ‘hair’ ikoya ‘his rabbit’ kapay ‘sister-in-law’ Hypothesis 1.ik…a ‘his’ am ‘cornfield’ oy ‘rabbit’ Guess SP expression for ‘his hair’ and ‘his sister-in-law’. Now collect more data in the language. Consider the following data. kama ‘cornfield’ ikama ‘his cornfield’ koya ‘rabbit’ ikoya ‘his rabbit’ way ‘hair’ iway ‘his hair’ kapay ‘sister-in-law’ ikapay ‘his sister-in-law’ Revised Hypothesis The morpheme –i means ‘his’. The sequence k…a was just a coincidence. Elicit for my and your angkama ‘my cornfield’ ingkama ‘your cornfield’ angkoya ‘my rabbit’ ingkoya ‘your rabbit’ angway ‘my hair’ ingway ‘your hair’ angkapay ‘my sister-in-law’ ingkapay ‘your sister-in-law’ Till now we have discovered the following SP morphemes: ang ‘my’, ing ‘your’ i ‘his’ kama ‘cornfield’ koya ‘rabbit’ way ‘hair’ kapay’ sister-inlaw’ Can you describe the procedure how we discovered these morphemes? Steps in identifying morphemes. Step 1: Data collection (note that your data must contain partially like and different) Step 2: Substitution and contrast ( compare the differences in the forms in the target language and corresponding differences in meaning in the contact language) e.g. 6 kama ‘ cornfield’ koya ‘ rabbit’ way ‘ hair’ kapay ‘ sister-in-law’ These forms contrast in an identical environment (ang….) or (ing….) and differ in meaning leading to the conclusion that they are different morphemes. Step 3: Recurring partials (Check the forms that are alike. If a form has the same meaning wherever it occurs, it is considered to be one morpheme.) In the above example ang and ing translate each time my and your constantly. Possible assumptions about the potential diversity of morphological restrictions: (i) Each form carries a unit of meaning; (ii) The parts of a word contribute to whole word—compositional meaning; (iii) Difference in types of meanings (lexical, rather concrete vs. grammatical, more abstract), which is also reflected in the difference in the types of forms (more / less weight, important first / last); (iv) The same meaning is always determined by exactly the same form, and vice versa; (v) The meanings of parts do not contribute to the whole of the expression e.g. Kick the bucket. These assumptions are not implausible, but also not without problems— Ideals of the morphology but not always achieved in reality. Assumptions (iv) and (v) The ideal representation FORM MEANING Deviations from the ideal F M1 / M2 Ambiguity homonymy: Engl. bank - bank, Germ.-t PST— -t SBJ: PRES 3SG; Engl. no - know, -s SBJ: 3SG - s-PL Polysemy Engl. Foot, mouse, etc. 7 F1 / F2 synonymy e.g., Engl. lorry - truck; e.g. Germ.-s [Oma-s) -en (as in Frau-en); Engl. bucket, pail M F M1 - M2 F1 - F2 Cumulation e.g.Engl. punch, slap… Germ. –e SBJ:3SG.PST Lat. –mini SBJ:2PL.IND.PST.PASS Extension e.g. Lat. mo-mord-i PFV-laud-1SG.IND.PFV.ACT M Germ. ge-schau-t PARTIZIPII-look-PARTIZIPII Assumption (ii) Is the German word Hand-schuh compositional? (= The meaning of the whole is regular from the meanings of the parts) Bahn-hof? Haenge-matte? Arm-brust? be-greifen? ver-stehn? There are word-structures, which in real life, unfortunately, not compositional. Also phrase structures are not compositional. E.g. English. kick the bucket, a verbal phrase with the meaning 'scrape off, to die': How are the meanings of the familiar parts – preposition or verb, definite article, noun - automatically contribute to the meanings of the whole? Constructions, on the word as the phrase level, are often idiomatic expressions (non-compositional), and, historically speaking, idioms become lexicalized. Assumption (iii) Consider the following examples: Where is the initial English /  / and //? theory, theme, thesis, thin, thigh, thing, thistle, through, three, thug, thwart, etc. the, this, these, that, those, they, them, their, theirs, (thou, thee, thy, thine) there, thither, therefore, thus, then, thence, than, though, the2 (as in the sooner the better) 8 All grammatical words, often demonstrative, definite, etc —/  / ? What would be available for the remainder of the /  / be? • a form with a meaning which contrasts with the meanings of -is in this,-at in that, etc. • a form that WITHOUT th- occurs in a mat /  mæt / and the same meaning as in combination with th(Problem: the idea / i.adia/. - an idea / n ada/) For the (lack of) to the weightiness of this comparison: Latin dedissemus’ We have given ' - The 3rd loud, / d /, all alone to bear the meaning 'give' (Latin is an Indo-European language, Italic subfamily; not in strict sense "dead" language, but continued living in the present Romance languages.) Despite such cases, where the concrete lexical meaning is born by the minimal form (a single sound segment!) and heavy phonetic unit represents the abstract grammatical meaning. It seems that the trend in general but more so to be in accordance with assumption (iii), • that morphemes (particularly lexical) roughly correspond to syllables; • lexical morphemes that are weightier than grammatical morphemes. SYLLABLE= MORPH / WORD cf Eng. have, do, be, see, hear, feel, smell, taste, eat, drink, chew, run, walk, swim, fly, kick, kill, miss, hit, strike, punch, cut, bend, hold, blow, guess, love, hate, like, own, get, give, send, cast, throw, pay, sell, buy, speak, talk, cry, shout, yell, read, write, mean ... see also: linguistic (s), enamel, Tambourine, phenomenon, imbroglio, Hippopotamus, rhinoceros, symposium, etc. For which of the above assumptions, the following English data are problematic? hit strike kick punch slap smack 'bring hand or something held in hand forcefully in contact with surface of someone or something' 'hit sharply or forcefully' 'hit with foot' 'hit with closed hand (fist)' 'hit with flat part of open hand' 'hit with flat part of open hand, 9 spank butt nudge making a loud noise' 'hit with open hand, esp. on buttocks' 'hit/strike/push with head or horns' 'hit/push with elbow' => The meanings are complex, and although the same systematic complex ('to hit' + 'with a body part), but there is no systematic formal complexity of recognizable parts: a case for the lexical analysis, not for the morphological. In a language other than English, all the above may not be lexicalized. All these concepts may be expressed periphrastically with the addition to hit. Homework: Consider the following English examples for practicing morpheme analysis: Does mat form a separate morpheme in mattress? If problems arise, consult a dictionary. vgl. add address waiter waitress tiger tigress master/mister mistress murderer murder[er]ess seam seamstress ‘a line of stitches joining two pieces of cloth, ‘a woman whose leather, etc.’ job is sewing’ fort fortress ‘a large fort; place strengthened for defense’ butt buttress ‘to strike or push against s.o./s.th. with the ‘a solid structure head or horns; a push’; built against a wall ‘a person or thing that people make fun of’; as a support’ ‘a large, thick, or bottom end’ Historically: Eng. mattress <Old Fr. Materas <Arab. matrah 'Something on which you lie down; something for lying down' Further problems of morpheme analysis • What is the problem of raspberries, cranberries, scapegoats? Segment them as rasp-berries. Is rasp-meaningful? 10 What is the distribution of rasp? Does it occur as prefix, root, or suffix or what else? Same questions for cran- and scape-. Unique morpheme or empty morpheme • What is the problem with English receive? o Segment it as re-ceive? Is -ceive meaningful in English? o It has the same distribution o Also found in words as: de-ceive, per-ceive, con-ceive. o Really not easy to say what this –ceive in these four verbs mean. One possible meaning could be 'access' or' take '. o They have some similarities: all of them are verbs, take same form when nominalized (re-CEP-tion, de-CEP-tion, per-CEP-tion, conCEP-tion). o These words have same meanings and grammar. To ponder upon Does your intuitive knowledge help you to segment the parts of a word? Above we had two problem cases with: th-e and mat (t)-ress. What about English words lord, lady, female - complex or simple? 2.2 Classifying morphemes the building blocks of meaning • By Form free lexical items = Word, root, stems Lexemes e.g. Kekchi beq-'go' e.g. German schau ' look; perceive’ or bounded, grammatical = Affix o.a. Exponent Grammemes -n -t -in -at 'is currently in progress' 'future' 'subject=speaker 'subject = addressee' -t -e -(e)st -t / e -en -(e)t -en past SBJ=1SG SBJ= 2SG SBJ=3SG [PRES / PAST SBJ=1PL SBJ=2PL SBJ= 3PL 11 It is possible that the words may consist of all bound forms. Consider the following examples from Manipuri, a Tibeto-Burman language. poss 1 2 3 stems inama- -pa -ca -ma ‘father’ ‘child’ ‘mother’ In Manipuri, one cannot pronounce –pa ‘father’, -ca ‘child’, and –ma ‘mother’ in isolation i.e. without possessive morphemes. However, free morphemes do exist in the language. In the following example, possessives are optional: (poss) stems 1 2 3 inama- mit buk yum ‘eye’ ‘stomach’ ‘house’ "Free" may not always be taken literally. Sometimes the stem/root is a syntactic unit as in Imperative of 2SG (look!). • By meaning • persons, things, actions, properties, etc. (general criterion: Time stability?); concrete vs. abstract, specific vs. general. Although concreteness / abstractness, specificity / general are not easily understood. Everyone agrees that the meanings of Kekchi-beq- or German schau- are relatively concrete and specific than the meanings such as ' is currently in progress' (present tense), 'future' (FUTUR), 'subject = a speaker', or 'subject = addressee'. • by position • closer to the root / distant from the root un-de-cipher-abil-ity, re-place-ment-s anti-dis-establish-ment-ar-ian-ism • • before, after, in and around the root (Prefix, suffix, infix, Circumfix) Note:-ment and-s are both suffixes because they both occur after the root. –ment is closer to the root and –s comes at the end of the word. –ment is not inside the stem, it is after the root. 12