1. Morphology: Words and their Parts CS 4705 Slides adapted from Jurafsky, Martin Hirschberg and Dorr.

2. English Morphology Morphology is the study of the ways that words are built up from smaller meaningful units called morphemes We can usefully divide morphemes into two classes – Stems: The core meaning bearing units – Affixes: Bits and pieces that adhere to stems to change their meanings and grammatical functions

3. Nouns and Verbs (English) Nouns are simple (not really) – Markers for plural and possessive Verbs are only slightly more complex – Markers appropriate to the tense of the verb

4. Regulars and Irregulars Ok so it gets a little complicated by the fact that some words misbehave (refuse to follow the rules) – Mouse/mice, goose/geese, ox/oxen – Go/went, fly/flew The terms regular and irregular will be used to refer to words that follow the rules and those that don’t.

5. Regular and Irregular Nouns and Verbs Regulars… – Walk, walks, walking, walked, walked – Table, tables Irregulars – Eat, eats, eating, ate, eaten – Catch, catches, catching, caught, caught – Cut, cuts, cutting, cut, cut – Goose, geese

6. Why care about morphology? Spelling correction: referece – Morphology in machine translation Spanish words quiero and quieres are both related to querer ‘want’ – Hyphenation algorithms: refer-ence – Part-of-speech analysis: google, googler – Text-to-speech: grapheme-to-phoneme conversion hothouse (/T/ or /D/) – Allows us to guess at meaning ‘Twas brillig and the slithy toves… Muggles moogled migwiches

7. Concatenative Morphology Morpheme+Morpheme+Morpheme+… Stems: often called lemma, base form, root, lexeme – hope+ing hoping hop hopping Affixes – Prefixes: Antidisestablishmentarianism – Suffixes: Antidisestablishmentarianism – Infixes: hingi (borrow) – humingi (borrower) in Tagalog – Circumfixes: sagen (say) – gesagt (said) in German

8. What useful information does morphology give us? Different things in different languages – Spanish: hablo, hablar é / English: I speak, I will speak – English: book, books/ Japanese: hon, hon Languages differ in how they encode morphological information – Isolating languages (e.g. Cantonese) have no affixes: each word usually has 1 morpheme – Agglutinative languages (e.g. Finnish, Turkish) are composed of prefixes and suffixes added to a stem (like beads on a string) – each feature realized by a single affix, e.g. Finnish

9. epäjärjestelmällistyttämättömyydellänsäkäänköhän ‘Wonder if he can also... with his capability of not causing things to be unsystematic’ – Inflectional languages (e.g. English) merge different features into a single affix (e.g. ‘ s ’ in likes indicates both person and tense); and the same feature can be realized by different affixes – Polysynthetic languages (e.g. Inuit languages) express much of their syntax in their morphology, incorporating a verb ’ s arguments into the verb, e.g. Western Greenlandic Aliikusersuillammassuaanerartassagaluarpaalli. aliiku-sersu-i-llammas-sua-a-nerar-ta-ssa-galuar-paal-li entertainment-provide-SEMITRANS-one.good.at-COP-say.that-REP- FUT-sure.but-3.PL.SUBJ/3SG.OBJ-but 'However, they will say that he is a great entertainer, but...'SEMITRANSCOP FUTOBJ – So ….different languages may require very different morphological analyzers

10. What we want Something to automatically do the following kinds of mappings: Catscat +N +PL Cat cat +N +SG Citiescity +N +PL Mergingmerge +V +Present-participle Caughtcatch +V +past-participle

11. Morphology Can Help Define Word Classes AKA morphological classes, parts-of-speech Closed vs. open (function vs. content) class words – Pronoun, preposition, conjunction, determiner,… – Noun, verb, adverb, adjective,… Identifying word classes is useful for almost any task in NLP, from translation to speech recognition to topic detection…very basic semantics

12. (English) Inflectional Morphology Word stem + grammatical morpheme  different forms of same word – Usually produces word of same classclass – Usually serves a syntactic or grammatical function (e.g. agreement) like  likes or liked bird  birds Nominal morphology – Plural forms s or es Irregular forms (goose/geese)

13. Mass vs. count nouns (fish/fish(es), email or emails?) – Possessives (cat’s, cats’) Verbal inflection – Main verbs (sleep, like, fear) relatively regular -s, ing, ed And productive: emailed, instant-messaged, faxed, homered But some are not: – eat/ate/eaten, catch/caught/caught – Primary (be, have, do) and modal verbs (can, will, must) often irregular and not productive Be: am/is/are/were/was/been/being – Irregular verbs few (~250) but frequently occurring

14. Derivational Morphology Word stem + syntactic/grammatical morpheme  new words – Usually produces word of different class – Incomplete process: derivational morphs cannot be applied to just any member of a class Verbs --> nouns – -ize verbs  -ation nouns – generalize, realize  generalization, realization – synthesize but not synthesization

15. Verbs, nouns  adjectives – embrace, pity  embraceable, pitiable – care, wit  careless, witless Adjective  adverb – happy  happily Process selective in unpredictable ways – Less productive: nerveless/*evidence-less, malleable/*sleep-able, rar-ity/*rareness – Meanings of derived terms harder to predict by rule – clueless, careless, nerveless, sleepless

16. Compounding Two base forms join to form a new word – Bedtime, Weinerschnitzel, Rotwein – Careful? Compound or derivation?

17. Morphotactics What are the ‘rules’ for constructing a word in a given language? – Pseudo-intellectual vs. *intellectual-pseudo – Rational-ize vs *ize-rational – Cretin-ous vs. *cretin-ly vs. *cretin-acious

18. Semantics: In English, un- cannot attach to adjectives that already have a negative connotation: – Unhappy vs. *unsad – Unhealthy vs. *unsick – Unclean vs. *undirty Phonology: In English, -er cannot attach to words of more than two syllables – great, greater – Happy, happier – Competent, *competenter – Elegant, *eleganter – Unruly, ?unrulier

19. Morphological Parsing These regularities enable us to create software to parse words into their component parts

20. Morphology and FSAs We’d like to use the machinery provided by FSAs to capture facts about morphology Ie. Accept strings that are in the language And reject strings that are not And do it in a way that doesn’t require us to in effect list all the words in the language

21. What do we need to build a morphological parser? Lexicon: list of stems and affixes (w/ corresponding p.o.s.) Morphotactics of the language: model of how and which morphemes can be affixed to a stem Orthographic rules: spelling modifications that may occur when affixation occurs – in  il in context of l (in- + legal) Most morphological phenomena can be described with regular expressions – so finite state techniques often used to represent morphological processes

22. Start Simple Regular singular nouns are ok Regular plural nouns have an -s on the end Irregulars are ok as is

23. Simple Rules

24. Now Add in the Words

25. Derivational morphology: adjective fragment q3 q5 q4 q0 q1q2 un- adj-root 1 -er, -ly, -est  adj-root 1 adj-root 2 -er, -est Adj-root 1 : clear, happi, real (clearly) Adj-root 2 : big, red (*bigly)

26. Parsing/Generation vs. Recognition We can now run strings through these machines to recognize strings in the language Accept words that are ok Reject words that are not But recognition is usually not quite what we need Often if we find some string in the language we might like to find the structure in it (parsing) Or we have some structure and we want to produce a surface form (production/generation) Example From “cats” to “cat +N +PL”

27. Finite State Transducers The simple story Add another tape Add extra symbols to the transitions On one tape we read “cats”, on the other we write “cat +N +PL”

28. Applications The kind of parsing we’re talking about is normally called morphological analysis It can either be An important stand-alone component of an application (spelling correction, information retrieval) Or simply a link in a chain of processing

29. FSTs Kimmo Koskenniemi’s two-level morphology Idea: word is a relationship between lexical level (its morphemes) and surface level (its orthography)

30. Transitions c:c means read a c on one tape and write a c on the other +N:ε means read a +N symbol on one tape and write nothing on the other +PL:s means read +PL and write an s c:ca:at:t +N:ε + PL:s

31. Typical Uses Typically, we’ll read from one tape using the first symbol on the machine transitions (just as in a simple FSA). And we’ll write to the second tape using the other symbols on the transitions. In general, FSTs can be used for – Translators (Hello:Ciao) – Parser/generators (Hello:How may I help you?) – As well as Kimmo-style morphological parsing

32. Ambiguity Recall that in non-deterministic recognition multiple paths through a machine may lead to an accept state. Didn’t matter which path was actually traversed In FSTs the path to an accept state does matter since differ paths represent different parses and different outputs will result

33. Ambiguity What’s the right parse (segmentation) for Unionizable Union-ize-able Un-ion-ize-able Each represents a valid path through the derivational morphology machine.

34. Ambiguity There are a number of ways to deal with this problem Simply take the first output found Find all the possible outputs (all paths) and return them all (without choosing) Bias the search so that only one or a few likely paths are explored

35. The Gory Details Of course, its not as easy as “cat +N +PL” “cats” As we saw earlier there are geese, mice and oxen But there are also a whole host of spelling/pronunciation changes that go along with inflectional changes Cats vs Dogs Fox and Foxes

36. Multi-Tape Machines To deal with this we can simply add more tapes and use the output of one tape machine as the input to the next So to handle irregular spelling changes we’ll add intermediate tapes with intermediate symbols

37. Generativity Nothing really privileged about the directions. We can write from one and read from the other or vice-versa. One way is generation, the other way is analysis

38. Multi-Level Tape Machines We use one machine to transduce between the lexical and the intermediate level, and another to handle the spelling changes to the surface tape

39. Lexical to Intermediate Level

40. Intermediate to Surface The add an “e” rule as in fox^s# foxes#

41. Foxes

42. Note A key feature of this machine is that it doesn’t do anything to inputs to which it doesn’t apply. Meaning that they are written out unchanged to the output tape.

43. Overall Scheme We now have one FST that has explicit information about the lexicon (actual words, their spelling, facts about word classes and regularity). Lexical level to intermediate forms We have a larger set of machines that capture orthographic/spelling rules. Intermediate forms to surface forms

44. Overall Scheme

45. Cascades This is a scheme that we’ll see again and again. Overall processing is divided up into distinct rewrite steps The output of one layer serves as the input to the next The intermediate tapes may or may not wind up being useful in their own right

46. Porter Stemmer Porter Stemmer (1980) Used for tasks in which you only care about the stem – IR, modeling given/new distinction, topic detection, document similarity Lexicon-free morphological analysis Cascades rewrite rules (e.g. misunderstanding --> misunderstand --> understand --> …) Easily implemented as an FST with rules e.g. – ATIONAL  ATE – ING  ε Not perfect …. – Doing  doe

47. Policy  police Does stemming help? – IR, little – Topic detection, more

48. Summing Up FSTs provide a useful tool for implementing a standard model of morphological analysis, Kimmo’s two-level morphology But for many tasks (e.g. IR) much simpler approaches are still widely used, e.g. the rule-based Porter Stemmer Next time: – Read Ch 4 HW1 assigned; see web page: http://www.cs.columbia.edu/~kathy/NLP http://www.cs.columbia.edu/~kathy/NLP