| { |
| "paper_id": "P86-1023", |
| "header": { |
| "generated_with": "S2ORC 1.0.0", |
| "date_generated": "2023-01-19T09:13:06.177622Z" |
| }, |
| "title": "Morph~lo~leal Decomposition and 5tress Assignment for Speech Synthesis Kenneth Church Bell Laboratories 600 Mountain Ave. Murray Hill, N.J. research !alice !kwc", |
| "authors": [], |
| "year": "", |
| "venue": null, |
| "identifiers": {}, |
| "abstract": "", |
| "pdf_parse": { |
| "paper_id": "P86-1023", |
| "_pdf_hash": "", |
| "abstract": [], |
| "body_text": [ |
| { |
| "text": "A speech synthesizer is a machine that inputs a stream of text and outputs a speech signal. This paper will discuss a small piece of how words are converted to phonemes. Typically words are converted to phonemes in one of two ways: either by looking the words up in a dictionary (with possibly some limited morphological analysis), or by sounding the words out from their spelling using basic principles.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Background", |
| "sec_num": "1." |
| }, |
| { |
| "text": "\u2022 Letter to Sound Both appt~oaches have their advantages and disadvantages; dictionary lookup fails for unknown words (e.g., proper nouns) and letter to sound rules fail for irregular words, which are all too common in English. Most speech synthesizers adopt a hybrid strategy, using the dictionary when possible and turning to letter to sound rules for the rest. I discussed letter to sound rules at the last meeting of the ACL [Church] ; this paper will report on some new dictionary lookup approaches, with an emphasis on morphology.", |
| "cite_spans": [ |
| { |
| "start": 429, |
| "end": 437, |
| "text": "[Church]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022 Dictionary Lookup", |
| "sec_num": null |
| }, |
| { |
| "text": "Morphological decomposition is used to reduce the size of the dictionary and to increase coverage. Instead of storing all possible words, the system can store just a lexicon of morphemes and save a factor of 10 [Jon Allen (personal communication)] in storage. Now when the system is given a word and asked to determine is pronunciation, the system decomposes the word into known morphemes, looks up the pronunciation of each of the pieces and combines the results.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "\u2022 Dictionary Lookup", |
| "sec_num": null |
| }, |
| { |
| "text": "The best known morphological decomposition system is the Decomp module in the MITalk sysnthesizer [Allen et. al.] . This system attempted to parse an input word such as formally into morphemes: form, -al and -ly. It was assumed that morphemes are concatenated together (like \"beads on a string\") according to the finite state grammar shown below:", |
| "cite_spans": [ |
| { |
| "start": 98, |
| "end": 113, |
| "text": "[Allen et. al.]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "MITalk Decomp", |
| "sec_num": "2." |
| }, |
| { |
| "text": "The types of morphemes were:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "MITalk Decomp", |
| "sec_num": "2." |
| }, |
| { |
| "text": "1.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "MITalk Decomp", |
| "sec_num": "2." |
| }, |
| { |
| "text": "3. Costs were placed on the arcs to alleviate overgeneration. Note that the grammar produces quite a number of spurious analyses.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "2.", |
| "sec_num": null |
| }, |
| { |
| "text": "For example, not only would formally be analyzed as form-al-ly but it would also be analyzed as form-ally and for-mal-ly. The cost mechanism blocks these spurious analyses by assigning compounding a higher cost than suffixation and therefore favoring the desired analysis. Although the cost mechanism handles a large number of cases, it would be better to aim toward a tighter grammar of morphology which did not overgenerate so badly.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "The MITalk Decomp program performed its task quite well; it could analyze 95% of running text [Allen (personal communication) ]. In order to achieve this level of performance, the authors of Decomp made a conscious decision not to deal with stress alternations (festive I festivity), vowel shift and tensing (divine / divinity), and other phonological rules associated with latinate morphology. Basically, there was only one rule for combining the pronunciations of morphological pieces: simple concatenation with a few simple rules to account for spelling alternations at the juncture:", |
| "cite_spans": [ |
| { |
| "start": 94, |
| "end": 125, |
| "text": "[Allen (personal communication)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 Silent e deletes before a vocalic suffix: observe + ance \"-'* observance", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 Consonant doubles before a vocalic suttix: red + est -\" reddest", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 y -\" i before a suffix: glory + ous ~ glorious", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "\u2022 y deletes before a suffix starting with i: harmony + ize harmonize", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "All affixes were assumed to be stress neutral. Words like festivity and divinity which require a richer understanding of the interaction of morphology and phonology were entered into the lexicon as exceptions.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "The decision not to handle more complicated morphological and phonological rules was based on the belief that it is hard to do an adequate job and that it wasn't necessary to do so because the rules are not very productive and hence it is possible (and practical) to list all of the derived forms in the lexicon. I'd like to believe that morphology and phonology have progressed enough over the past ten years that this argument does not have as much force as it did. Nevertheless, I have to admit that the payoff may be marginal, especially if measured in short term savings in the size of the lexicon and memory costs. The real value in the enterprise is more long term; I am betting that pushing the theoretical linguistic understanding with a demanding application such as speech synthesis will uncover some new insights.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Prefixes (pref): UNtie, PERmit, REduce", |
| "sec_num": null |
| }, |
| { |
| "text": "It has long been recognized that \"stress-shifting\" morphology (e.g., divin+ity) differs in quite a number of respects from \"stress neutral\" morphology (e.g., divine#ness). It is a wellestablished convention to mark the \"stress-shifting\" morpheme boundary with a \"+\" symbol and to mark the \"stress-neutral\" boundary with a \"#\" symbol. (Scare quotes are placed around \"stress-shifting\" and \"stress-neutral\" because these terms are probably not quite right.) This paper will also use the terms Level 1 and Level 2 to refer to the two types of morphological combination, respectively. This terminology is taken from the literature on Level Ordered Morphology and Phonology (e.g., [Mohanan] ) which argues that \"+\" boundary (level 1) morphology is ordered before \"#\" boundary (level 2) morphology and that this ordering dependency has important theoretical implications.", |
| "cite_spans": [ |
| { |
| "start": 676, |
| "end": 685, |
| "text": "[Mohanan]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "It is worthwhile to review some of the well-known differences between \"+\" boundaries and \"#\" boundaries. Informally \"+\" morphemes such as in +, ad +, ab +, +al, +ity are (generally) derived from Latin whereas \"#\" morphemes such as #ness, #1y come from Greek and German. This historical trend is only a rough correlation and has numerious counter-examples (e.g., the German suffix -ist behaves like \"'+\"). The program uses the following set of prefixes and suffixes:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 Level 1 \"+\" Prefixes: a, ab, ac, ad, af, ag, al, am, an, ap, at, as, at, bi, col, corn, con, cor, de, dif, dis, e, ec, ef, eg, el, em, en, er, es, ex, ira, in, ir, is, ob, oc, of, per, pre, pro, re, suf, sup, sur, sus, trans \u2022 Level 1 \"+\" Suffixes: ability, able, aceous, acious, acity, acy, age, al, ality, ament, an, ance, ancy, ant, ar, arity, ary, ate, ation, ational, ative, ator, atorial, atory, ature, bile, bility, ble, bly, e, ea, ean, ear, edge, ee, ence, ency, ent, ential, eous, ia, iac, ial, ian, iance, iant, iary, iate, iative, ibility, ible, ic, ical, ican, icate, ication, icative, icatory, ician, icity, icize, ide, ident, ience, iency, ient, ificate, ification, ificative, if y, ion, ional, ionary, ious, isation, ish, ist, istic, itarian, ite, ity, ium, ival, ive, ivity, ization, ize, le, ment, mental, mentary, on, or, ory, osity, ous, ular, ularity, ure, ute, utive, y \u2022 Level 2 \"#\" Prefixes: anti, co, de, for, mal, non, pre, sub, supra, tri, ultra, un \u2022 Level 2 \"#\" Suffixes: able, bee, berry, blast, bodies, body, copy, culture, fish, ful, fulling, head, herd, hood, ism, ist, ire, land, less, line, ly, man, ment, mental, mentarian, most, ness, phile, phyte, ship, shire, some, tree, type, ward, way, wise There is also a well-known precedence relation between + and #. morphemes attach more productively than + morphemes.", |
| "cite_spans": [ |
| { |
| "start": 14, |
| "end": 26, |
| "text": "Prefixes: a,", |
| "ref_id": null |
| }, |
| { |
| "start": 27, |
| "end": 30, |
| "text": "ab,", |
| "ref_id": null |
| }, |
| { |
| "start": 31, |
| "end": 34, |
| "text": "ac,", |
| "ref_id": null |
| }, |
| { |
| "start": 35, |
| "end": 38, |
| "text": "ad,", |
| "ref_id": null |
| }, |
| { |
| "start": 39, |
| "end": 42, |
| "text": "af,", |
| "ref_id": null |
| }, |
| { |
| "start": 43, |
| "end": 46, |
| "text": "ag,", |
| "ref_id": null |
| }, |
| { |
| "start": 47, |
| "end": 50, |
| "text": "al,", |
| "ref_id": null |
| }, |
| { |
| "start": 51, |
| "end": 54, |
| "text": "am,", |
| "ref_id": null |
| }, |
| { |
| "start": 55, |
| "end": 58, |
| "text": "an,", |
| "ref_id": null |
| }, |
| { |
| "start": 59, |
| "end": 62, |
| "text": "ap,", |
| "ref_id": null |
| }, |
| { |
| "start": 63, |
| "end": 66, |
| "text": "at,", |
| "ref_id": null |
| }, |
| { |
| "start": 67, |
| "end": 70, |
| "text": "as,", |
| "ref_id": null |
| }, |
| { |
| "start": 71, |
| "end": 74, |
| "text": "at,", |
| "ref_id": null |
| }, |
| { |
| "start": 75, |
| "end": 78, |
| "text": "bi,", |
| "ref_id": null |
| }, |
| { |
| "start": 79, |
| "end": 83, |
| "text": "col,", |
| "ref_id": null |
| }, |
| { |
| "start": 84, |
| "end": 89, |
| "text": "corn,", |
| "ref_id": null |
| }, |
| { |
| "start": 90, |
| "end": 94, |
| "text": "con,", |
| "ref_id": null |
| }, |
| { |
| "start": 95, |
| "end": 99, |
| "text": "cor,", |
| "ref_id": null |
| }, |
| { |
| "start": 100, |
| "end": 103, |
| "text": "de,", |
| "ref_id": null |
| }, |
| { |
| "start": 104, |
| "end": 108, |
| "text": "dif,", |
| "ref_id": null |
| }, |
| { |
| "start": 109, |
| "end": 113, |
| "text": "dis,", |
| "ref_id": null |
| }, |
| { |
| "start": 114, |
| "end": 116, |
| "text": "e,", |
| "ref_id": null |
| }, |
| { |
| "start": 117, |
| "end": 120, |
| "text": "ec,", |
| "ref_id": null |
| }, |
| { |
| "start": 121, |
| "end": 124, |
| "text": "ef,", |
| "ref_id": null |
| }, |
| { |
| "start": 125, |
| "end": 128, |
| "text": "eg,", |
| "ref_id": null |
| }, |
| { |
| "start": 129, |
| "end": 132, |
| "text": "el,", |
| "ref_id": null |
| }, |
| { |
| "start": 133, |
| "end": 136, |
| "text": "em,", |
| "ref_id": null |
| }, |
| { |
| "start": 137, |
| "end": 140, |
| "text": "en,", |
| "ref_id": null |
| }, |
| { |
| "start": 141, |
| "end": 144, |
| "text": "er,", |
| "ref_id": null |
| }, |
| { |
| "start": 145, |
| "end": 148, |
| "text": "es,", |
| "ref_id": null |
| }, |
| { |
| "start": 149, |
| "end": 152, |
| "text": "ex,", |
| "ref_id": null |
| }, |
| { |
| "start": 153, |
| "end": 157, |
| "text": "ira,", |
| "ref_id": null |
| }, |
| { |
| "start": 158, |
| "end": 161, |
| "text": "in,", |
| "ref_id": null |
| }, |
| { |
| "start": 162, |
| "end": 165, |
| "text": "ir,", |
| "ref_id": null |
| }, |
| { |
| "start": 166, |
| "end": 169, |
| "text": "is,", |
| "ref_id": null |
| }, |
| { |
| "start": 170, |
| "end": 173, |
| "text": "ob,", |
| "ref_id": null |
| }, |
| { |
| "start": 174, |
| "end": 177, |
| "text": "oc,", |
| "ref_id": null |
| }, |
| { |
| "start": 178, |
| "end": 181, |
| "text": "of,", |
| "ref_id": null |
| }, |
| { |
| "start": 182, |
| "end": 186, |
| "text": "per,", |
| "ref_id": null |
| }, |
| { |
| "start": 187, |
| "end": 191, |
| "text": "pre,", |
| "ref_id": null |
| }, |
| { |
| "start": 192, |
| "end": 196, |
| "text": "pro,", |
| "ref_id": null |
| }, |
| { |
| "start": 197, |
| "end": 200, |
| "text": "re,", |
| "ref_id": null |
| }, |
| { |
| "start": 201, |
| "end": 205, |
| "text": "suf,", |
| "ref_id": null |
| }, |
| { |
| "start": 206, |
| "end": 210, |
| "text": "sup,", |
| "ref_id": null |
| }, |
| { |
| "start": 211, |
| "end": 215, |
| "text": "sur,", |
| "ref_id": null |
| }, |
| { |
| "start": 216, |
| "end": 220, |
| "text": "sus,", |
| "ref_id": null |
| }, |
| { |
| "start": 221, |
| "end": 259, |
| "text": "trans \u2022 Level 1 \"+\" Suffixes: ability,", |
| "ref_id": null |
| }, |
| { |
| "start": 260, |
| "end": 265, |
| "text": "able,", |
| "ref_id": null |
| }, |
| { |
| "start": 266, |
| "end": 273, |
| "text": "aceous,", |
| "ref_id": null |
| }, |
| { |
| "start": 274, |
| "end": 281, |
| "text": "acious,", |
| "ref_id": null |
| }, |
| { |
| "start": 282, |
| "end": 288, |
| "text": "acity,", |
| "ref_id": null |
| }, |
| { |
| "start": 289, |
| "end": 293, |
| "text": "acy,", |
| "ref_id": null |
| }, |
| { |
| "start": 294, |
| "end": 298, |
| "text": "age,", |
| "ref_id": null |
| }, |
| { |
| "start": 299, |
| "end": 302, |
| "text": "al,", |
| "ref_id": null |
| }, |
| { |
| "start": 303, |
| "end": 309, |
| "text": "ality,", |
| "ref_id": null |
| }, |
| { |
| "start": 310, |
| "end": 316, |
| "text": "ament,", |
| "ref_id": null |
| }, |
| { |
| "start": 317, |
| "end": 320, |
| "text": "an,", |
| "ref_id": null |
| }, |
| { |
| "start": 321, |
| "end": 326, |
| "text": "ance,", |
| "ref_id": null |
| }, |
| { |
| "start": 327, |
| "end": 332, |
| "text": "ancy,", |
| "ref_id": null |
| }, |
| { |
| "start": 333, |
| "end": 337, |
| "text": "ant,", |
| "ref_id": null |
| }, |
| { |
| "start": 338, |
| "end": 341, |
| "text": "ar,", |
| "ref_id": null |
| }, |
| { |
| "start": 342, |
| "end": 348, |
| "text": "arity,", |
| "ref_id": null |
| }, |
| { |
| "start": 349, |
| "end": 353, |
| "text": "ary,", |
| "ref_id": null |
| }, |
| { |
| "start": 354, |
| "end": 358, |
| "text": "ate,", |
| "ref_id": null |
| }, |
| { |
| "start": 359, |
| "end": 365, |
| "text": "ation,", |
| "ref_id": null |
| }, |
| { |
| "start": 366, |
| "end": 374, |
| "text": "ational,", |
| "ref_id": null |
| }, |
| { |
| "start": 375, |
| "end": 381, |
| "text": "ative,", |
| "ref_id": null |
| }, |
| { |
| "start": 382, |
| "end": 387, |
| "text": "ator,", |
| "ref_id": null |
| }, |
| { |
| "start": 388, |
| "end": 396, |
| "text": "atorial,", |
| "ref_id": null |
| }, |
| { |
| "start": 397, |
| "end": 403, |
| "text": "atory,", |
| "ref_id": null |
| }, |
| { |
| "start": 404, |
| "end": 410, |
| "text": "ature,", |
| "ref_id": null |
| }, |
| { |
| "start": 411, |
| "end": 416, |
| "text": "bile,", |
| "ref_id": null |
| }, |
| { |
| "start": 417, |
| "end": 424, |
| "text": "bility,", |
| "ref_id": null |
| }, |
| { |
| "start": 425, |
| "end": 429, |
| "text": "ble,", |
| "ref_id": null |
| }, |
| { |
| "start": 430, |
| "end": 434, |
| "text": "bly,", |
| "ref_id": null |
| }, |
| { |
| "start": 435, |
| "end": 437, |
| "text": "e,", |
| "ref_id": null |
| }, |
| { |
| "start": 438, |
| "end": 441, |
| "text": "ea,", |
| "ref_id": null |
| }, |
| { |
| "start": 442, |
| "end": 446, |
| "text": "ean,", |
| "ref_id": null |
| }, |
| { |
| "start": 447, |
| "end": 451, |
| "text": "ear,", |
| "ref_id": null |
| }, |
| { |
| "start": 452, |
| "end": 457, |
| "text": "edge,", |
| "ref_id": null |
| }, |
| { |
| "start": 458, |
| "end": 461, |
| "text": "ee,", |
| "ref_id": null |
| }, |
| { |
| "start": 462, |
| "end": 467, |
| "text": "ence,", |
| "ref_id": null |
| }, |
| { |
| "start": 468, |
| "end": 473, |
| "text": "ency,", |
| "ref_id": null |
| }, |
| { |
| "start": 474, |
| "end": 478, |
| "text": "ent,", |
| "ref_id": null |
| }, |
| { |
| "start": 479, |
| "end": 486, |
| "text": "ential,", |
| "ref_id": null |
| }, |
| { |
| "start": 487, |
| "end": 492, |
| "text": "eous,", |
| "ref_id": null |
| }, |
| { |
| "start": 493, |
| "end": 496, |
| "text": "ia,", |
| "ref_id": null |
| }, |
| { |
| "start": 497, |
| "end": 501, |
| "text": "iac,", |
| "ref_id": null |
| }, |
| { |
| "start": 502, |
| "end": 506, |
| "text": "ial,", |
| "ref_id": null |
| }, |
| { |
| "start": 507, |
| "end": 511, |
| "text": "ian,", |
| "ref_id": null |
| }, |
| { |
| "start": 512, |
| "end": 518, |
| "text": "iance,", |
| "ref_id": null |
| }, |
| { |
| "start": 519, |
| "end": 524, |
| "text": "iant,", |
| "ref_id": null |
| }, |
| { |
| "start": 525, |
| "end": 530, |
| "text": "iary,", |
| "ref_id": null |
| }, |
| { |
| "start": 531, |
| "end": 536, |
| "text": "iate,", |
| "ref_id": null |
| }, |
| { |
| "start": 537, |
| "end": 544, |
| "text": "iative,", |
| "ref_id": null |
| }, |
| { |
| "start": 545, |
| "end": 553, |
| "text": "ibility,", |
| "ref_id": null |
| }, |
| { |
| "start": 554, |
| "end": 559, |
| "text": "ible,", |
| "ref_id": null |
| }, |
| { |
| "start": 560, |
| "end": 563, |
| "text": "ic,", |
| "ref_id": null |
| }, |
| { |
| "start": 564, |
| "end": 569, |
| "text": "ical,", |
| "ref_id": null |
| }, |
| { |
| "start": 570, |
| "end": 575, |
| "text": "ican,", |
| "ref_id": null |
| }, |
| { |
| "start": 576, |
| "end": 582, |
| "text": "icate,", |
| "ref_id": null |
| }, |
| { |
| "start": 583, |
| "end": 591, |
| "text": "ication,", |
| "ref_id": null |
| }, |
| { |
| "start": 592, |
| "end": 600, |
| "text": "icative,", |
| "ref_id": null |
| }, |
| { |
| "start": 601, |
| "end": 609, |
| "text": "icatory,", |
| "ref_id": null |
| }, |
| { |
| "start": 610, |
| "end": 616, |
| "text": "ician,", |
| "ref_id": null |
| }, |
| { |
| "start": 617, |
| "end": 623, |
| "text": "icity,", |
| "ref_id": null |
| }, |
| { |
| "start": 624, |
| "end": 630, |
| "text": "icize,", |
| "ref_id": null |
| }, |
| { |
| "start": 631, |
| "end": 635, |
| "text": "ide,", |
| "ref_id": null |
| }, |
| { |
| "start": 636, |
| "end": 642, |
| "text": "ident,", |
| "ref_id": null |
| }, |
| { |
| "start": 643, |
| "end": 649, |
| "text": "ience,", |
| "ref_id": null |
| }, |
| { |
| "start": 650, |
| "end": 656, |
| "text": "iency,", |
| "ref_id": null |
| }, |
| { |
| "start": 657, |
| "end": 662, |
| "text": "ient,", |
| "ref_id": null |
| }, |
| { |
| "start": 663, |
| "end": 671, |
| "text": "ificate,", |
| "ref_id": null |
| }, |
| { |
| "start": 672, |
| "end": 682, |
| "text": "ification,", |
| "ref_id": null |
| }, |
| { |
| "start": 683, |
| "end": 693, |
| "text": "ificative,", |
| "ref_id": null |
| }, |
| { |
| "start": 694, |
| "end": 699, |
| "text": "if y,", |
| "ref_id": null |
| }, |
| { |
| "start": 700, |
| "end": 704, |
| "text": "ion,", |
| "ref_id": null |
| }, |
| { |
| "start": 705, |
| "end": 711, |
| "text": "ional,", |
| "ref_id": null |
| }, |
| { |
| "start": 712, |
| "end": 719, |
| "text": "ionary,", |
| "ref_id": null |
| }, |
| { |
| "start": 720, |
| "end": 725, |
| "text": "ious,", |
| "ref_id": null |
| }, |
| { |
| "start": 726, |
| "end": 734, |
| "text": "isation,", |
| "ref_id": null |
| }, |
| { |
| "start": 735, |
| "end": 739, |
| "text": "ish,", |
| "ref_id": null |
| }, |
| { |
| "start": 740, |
| "end": 744, |
| "text": "ist,", |
| "ref_id": null |
| }, |
| { |
| "start": 745, |
| "end": 751, |
| "text": "istic,", |
| "ref_id": null |
| }, |
| { |
| "start": 752, |
| "end": 760, |
| "text": "itarian,", |
| "ref_id": null |
| }, |
| { |
| "start": 761, |
| "end": 765, |
| "text": "ite,", |
| "ref_id": null |
| }, |
| { |
| "start": 766, |
| "end": 770, |
| "text": "ity,", |
| "ref_id": null |
| }, |
| { |
| "start": 771, |
| "end": 775, |
| "text": "ium,", |
| "ref_id": null |
| }, |
| { |
| "start": 776, |
| "end": 781, |
| "text": "ival,", |
| "ref_id": null |
| }, |
| { |
| "start": 782, |
| "end": 786, |
| "text": "ive,", |
| "ref_id": null |
| }, |
| { |
| "start": 787, |
| "end": 793, |
| "text": "ivity,", |
| "ref_id": null |
| }, |
| { |
| "start": 794, |
| "end": 802, |
| "text": "ization,", |
| "ref_id": null |
| }, |
| { |
| "start": 803, |
| "end": 807, |
| "text": "ize,", |
| "ref_id": null |
| }, |
| { |
| "start": 808, |
| "end": 811, |
| "text": "le,", |
| "ref_id": null |
| }, |
| { |
| "start": 812, |
| "end": 817, |
| "text": "ment,", |
| "ref_id": null |
| }, |
| { |
| "start": 818, |
| "end": 825, |
| "text": "mental,", |
| "ref_id": null |
| }, |
| { |
| "start": 826, |
| "end": 834, |
| "text": "mentary,", |
| "ref_id": null |
| }, |
| { |
| "start": 835, |
| "end": 838, |
| "text": "on,", |
| "ref_id": null |
| }, |
| { |
| "start": 839, |
| "end": 842, |
| "text": "or,", |
| "ref_id": null |
| }, |
| { |
| "start": 843, |
| "end": 847, |
| "text": "ory,", |
| "ref_id": null |
| }, |
| { |
| "start": 848, |
| "end": 854, |
| "text": "osity,", |
| "ref_id": null |
| }, |
| { |
| "start": 855, |
| "end": 859, |
| "text": "ous,", |
| "ref_id": null |
| }, |
| { |
| "start": 860, |
| "end": 865, |
| "text": "ular,", |
| "ref_id": null |
| }, |
| { |
| "start": 866, |
| "end": 874, |
| "text": "ularity,", |
| "ref_id": null |
| }, |
| { |
| "start": 875, |
| "end": 879, |
| "text": "ure,", |
| "ref_id": null |
| }, |
| { |
| "start": 880, |
| "end": 884, |
| "text": "ute,", |
| "ref_id": null |
| }, |
| { |
| "start": 885, |
| "end": 891, |
| "text": "utive,", |
| "ref_id": null |
| }, |
| { |
| "start": 892, |
| "end": 893, |
| "text": "y", |
| "ref_id": null |
| }, |
| { |
| "start": 928, |
| "end": 931, |
| "text": "de,", |
| "ref_id": null |
| }, |
| { |
| "start": 932, |
| "end": 936, |
| "text": "for,", |
| "ref_id": null |
| }, |
| { |
| "start": 937, |
| "end": 941, |
| "text": "mal,", |
| "ref_id": null |
| }, |
| { |
| "start": 942, |
| "end": 946, |
| "text": "non,", |
| "ref_id": null |
| }, |
| { |
| "start": 947, |
| "end": 951, |
| "text": "pre,", |
| "ref_id": null |
| }, |
| { |
| "start": 952, |
| "end": 956, |
| "text": "sub,", |
| "ref_id": null |
| }, |
| { |
| "start": 957, |
| "end": 963, |
| "text": "supra,", |
| "ref_id": null |
| }, |
| { |
| "start": 964, |
| "end": 968, |
| "text": "tri,", |
| "ref_id": null |
| }, |
| { |
| "start": 969, |
| "end": 975, |
| "text": "ultra,", |
| "ref_id": null |
| }, |
| { |
| "start": 976, |
| "end": 1008, |
| "text": "un \u2022 Level 2 \"#\" Suffixes: able,", |
| "ref_id": null |
| }, |
| { |
| "start": 1009, |
| "end": 1013, |
| "text": "bee,", |
| "ref_id": null |
| }, |
| { |
| "start": 1014, |
| "end": 1020, |
| "text": "berry,", |
| "ref_id": null |
| }, |
| { |
| "start": 1021, |
| "end": 1027, |
| "text": "blast,", |
| "ref_id": null |
| }, |
| { |
| "start": 1028, |
| "end": 1035, |
| "text": "bodies,", |
| "ref_id": null |
| }, |
| { |
| "start": 1036, |
| "end": 1041, |
| "text": "body,", |
| "ref_id": null |
| }, |
| { |
| "start": 1042, |
| "end": 1047, |
| "text": "copy,", |
| "ref_id": null |
| }, |
| { |
| "start": 1048, |
| "end": 1056, |
| "text": "culture,", |
| "ref_id": null |
| }, |
| { |
| "start": 1057, |
| "end": 1062, |
| "text": "fish,", |
| "ref_id": null |
| }, |
| { |
| "start": 1063, |
| "end": 1067, |
| "text": "ful,", |
| "ref_id": null |
| }, |
| { |
| "start": 1068, |
| "end": 1076, |
| "text": "fulling,", |
| "ref_id": null |
| }, |
| { |
| "start": 1077, |
| "end": 1082, |
| "text": "head,", |
| "ref_id": null |
| }, |
| { |
| "start": 1083, |
| "end": 1088, |
| "text": "herd,", |
| "ref_id": null |
| }, |
| { |
| "start": 1089, |
| "end": 1094, |
| "text": "hood,", |
| "ref_id": null |
| }, |
| { |
| "start": 1095, |
| "end": 1099, |
| "text": "ism,", |
| "ref_id": null |
| }, |
| { |
| "start": 1100, |
| "end": 1104, |
| "text": "ist,", |
| "ref_id": null |
| }, |
| { |
| "start": 1105, |
| "end": 1109, |
| "text": "ire,", |
| "ref_id": null |
| }, |
| { |
| "start": 1110, |
| "end": 1115, |
| "text": "land,", |
| "ref_id": null |
| }, |
| { |
| "start": 1116, |
| "end": 1121, |
| "text": "less,", |
| "ref_id": null |
| }, |
| { |
| "start": 1122, |
| "end": 1127, |
| "text": "line,", |
| "ref_id": null |
| }, |
| { |
| "start": 1128, |
| "end": 1131, |
| "text": "ly,", |
| "ref_id": null |
| }, |
| { |
| "start": 1132, |
| "end": 1136, |
| "text": "man,", |
| "ref_id": null |
| }, |
| { |
| "start": 1137, |
| "end": 1142, |
| "text": "ment,", |
| "ref_id": null |
| }, |
| { |
| "start": 1143, |
| "end": 1150, |
| "text": "mental,", |
| "ref_id": null |
| }, |
| { |
| "start": 1151, |
| "end": 1161, |
| "text": "mentarian,", |
| "ref_id": null |
| }, |
| { |
| "start": 1162, |
| "end": 1167, |
| "text": "most,", |
| "ref_id": null |
| }, |
| { |
| "start": 1168, |
| "end": 1173, |
| "text": "ness,", |
| "ref_id": null |
| }, |
| { |
| "start": 1174, |
| "end": 1180, |
| "text": "phile,", |
| "ref_id": null |
| }, |
| { |
| "start": 1181, |
| "end": 1187, |
| "text": "phyte,", |
| "ref_id": null |
| }, |
| { |
| "start": 1188, |
| "end": 1193, |
| "text": "ship,", |
| "ref_id": null |
| }, |
| { |
| "start": 1194, |
| "end": 1200, |
| "text": "shire,", |
| "ref_id": null |
| }, |
| { |
| "start": 1201, |
| "end": 1206, |
| "text": "some,", |
| "ref_id": null |
| }, |
| { |
| "start": 1207, |
| "end": 1212, |
| "text": "tree,", |
| "ref_id": null |
| }, |
| { |
| "start": 1213, |
| "end": 1218, |
| "text": "type,", |
| "ref_id": null |
| }, |
| { |
| "start": 1219, |
| "end": 1224, |
| "text": "ward,", |
| "ref_id": null |
| }, |
| { |
| "start": 1225, |
| "end": 1229, |
| "text": "way,", |
| "ref_id": null |
| }, |
| { |
| "start": 1230, |
| "end": 1234, |
| "text": "wise", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\"It is clear that #ness attaches more productively to bases of the form Xous than does +ity: fabulousness is much \"better\" than fabulosity, and similarly for other pairs (dubiousness I dubiety, dubiosity). There are even cases where the +ity derivative is not merely worse, but impossible acrimonious I *acrinoniosity, euphonious I *euphonosity, famous I *famosity. There is also the simple list test, which is still a good indicator. Walker (1936) lists fewer +ity derivatives than #ness derivatives of words of the form Xous.\" [Aronoff, .", |
| "cite_spans": [ |
| { |
| "start": 529, |
| "end": 538, |
| "text": "[Aronoff,", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "Aronoff continues to point out that the semantics of # boundaries tend to be more predictable and compositional than + boundaries. The meaning of callousness, for example, is more predictable from the meanings of callous and ness than the meanings of variety, notoriety and curiosity are from the meanings of their parts.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "The following list summarizes some of the differences between + and #:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes are (often) historically correlated with Latin; # with German and Greek", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes feed certain phonological rules (stress assignment, vowel shift); # do not.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes take precedence over #", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes can attach to bound morphemes; # cannot", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes are less productive than #", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "\u2022 + morphemes have less predictable semantics than #", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "The remainder of the paper will be divided into two sections, the first will be concerned with level 1 morphology and the second with level 2 morphology and compounding. Level 1 morphology has been studied more heavily in the lingusitics literature; level 2 is perhaps more important for practical applications, at least in the short term.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Types of Morphological Combination", |
| "sec_num": "3." |
| }, |
| { |
| "text": "A number of the differences between + and # ought to be relevant in decomposing level 1 affixes and reducing the posibility of spurious derivations. Not only is the feature approach ad hoc, but it also missing an important asymmetry. Note that most words ending with -ency (e.g., presidency) are derived from words ending with -ent (e.g., president), and crucially not the other way around. The intuition that the relation \"derived from\" is asymmetric has some distributional support: notice that the percentage of words ending in -ency which are morphologically related to words ending in -ent is much larger than the percentage of words ending in -ent which are related to words ending in -ency. (The program estimates these percentages to be 73% (36/49) and 5% (36/710), respectively, using a procedure described below.) This asymmetry is problematic for a concatenation model like MITalk's Decomp, which would place presidency and president on equal footing, deriving both from preside.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Aronoff-style [Aronoff] truncation rules provide an attractive mechanism for accounting for the asymmetry. Recall that Aronoff proposed that nominee be derived from nominate by truncating the -ate suffix and attaching -ee in a single step. The program uses truncation rules to capture the symmetry in the 'derived from' relation by permitting -ent to be truncated before -ency, but not the other way around. Thus, presidency is derived from president --ent + -ency, and president is not derived from presidency because does not truncate -ency before -ent. Truncation rules are subject to a number of constraints.", |
| "cite_spans": [ |
| { |
| "start": 14, |
| "end": 23, |
| "text": "[Aronoff]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "In particular, truncation is only found at level 1; truncation cannot apply at level 2 because, as mentioned above, level 2 affixes attach to words, not bound (-truncated) morphemes.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "How does the program decide which suffixes can be truncated and when? Let me introduce the notation -ency > -ent to mean (roughly) that words ending with -ency are likely to be derived from words ending with -ent. The precise status of the '>' relation should be to be explored more fully. In some cases, the relation is a necessary condition; if presidency is derived from an English word then it must be derived from president. In other cases, the relationship expresses a possibility but not a necessity. For example, words ending in -ation may be related to words ending in -ate, but not necessarily. Marchand describes the relation as follows:", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "\"The English vocabulary has been greatly enriched by borrowings, chiefly from Latin and French. In course of time, many related words which had come in as separate loans developed a derivational relation to each other, giving rise to derivative alternations. Such derivative alternations fall into three main groups. If 1) is a derivation from an English word, the only possible word is 2), ie., if piracy is a derivative from an English word, only pirate is possible. The statement does not imply that for every 1) there must be a 2). 1) may be a loan, or it may be formed on a Latin basis without any regard to the existence of an English word at all (enormity, for instance, is so coined). Nor does the derivational principle involve the existence of a 1) for every 2) (many words in -able or -ine are not matched by words in -ability resp. -inity).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Group B is represented by the pairs 1) -ation / 2) -ate (as creation ~ create), 1) -(e)ry / 2) -er (as carpentry carpenter), 1) -cress / 2) -erer (as murderess murderer), 1) -ious / 2) -ion (as ambitious ~ ambition, 1) -atious / 2) -ation (as vexatious ~ vexation).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "If 1) is a derivative from another English word, the derivational pattern 1) from 2) is possible, but not necessary. A derivative in -ation such as reforestation is connected with reforest, a derivative such as swannery is connected with swan, archeress is connected with archer, robustious is extended from robust (but otherwise an adj in -tious derived from a sb points to the sb ending in -tion, i.e. we have really type A).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Group C is nothing but a variant of A and concerns adjs in -atious as flirtatious. Originally deriving from sbs in -ation, the type is now equally connected with the unextended radical, i.e. flirt (the older derivation ostentatious 1658 has not entered this latter derivational connection).\" [Marchand, For pragmatic purposes, the program assumes that there is only one '>' relation, not three as Marchand suggests, and that the relation can be estimated statistically as follows:", |
| "cite_spans": [ |
| { |
| "start": 292, |
| "end": 302, |
| "text": "[Marchand,", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Probability (suffix I > suffix 2)number of words ending with both suffiX l and suffix2 number of words ending with suffix l", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The program estimates, for example, that -ency > -ent with a probability of 73% (36/49) and that -ent > -ency with a probability of 5% (36/710). The 36 words ending in ency which have a variant ending in -ent are: incumbency, complacency, indecency, excrescency, residency, presidency ascendency, dependency, independency, superintendency, despondency, exigency contingency, emergency, detergency, insurgency, deficiency, efficiency sufficiency, proficiency, expediency, clemency, permanency, transparency vicegerency, belligerency, currency, competency, prepotency, consistency inconsistency, frequency, delinquency, constituency, solvency and fervency. The estimate should be almost 100%; the program believes that decency, cadency, tendency, ambitendency, pudency, agency, regency, urgency, counterinsurgency, valency, patency, potency , and fluency are not derived from -ent. Most of the errors can be attributed to a heuristic which excludes short stems (e.g., ag-) on the grounds that these stems are often spurious. These errors could be fixed by ammending the heuristic to check a 'winners list' of one, two and three letter stems. Some of the other errors are due to accidental gaps in the dictionary.", |
| "cite_spans": [ |
| { |
| "start": 226, |
| "end": 640, |
| "text": "complacency, indecency, excrescency, residency, presidency ascendency, dependency, independency, superintendency, despondency, exigency contingency, emergency, detergency, insurgency, deficiency, efficiency sufficiency, proficiency, expediency, clemency, permanency, transparency vicegerency, belligerency, currency, competency, prepotency, consistency inconsistency, frequency, delinquency, constituency, solvency", |
| "ref_id": null |
| }, |
| { |
| "start": 726, |
| "end": 838, |
| "text": "cadency, tendency, ambitendency, pudency, agency, regency, urgency, counterinsurgency, valency, patency, potency", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The results of this statistical estimation are shown in the figure below (where -0 denotes the null suffix): -able (43%),-ate (29%) -0 (24%),-ation (18%),-ate (17%),-e (14%),-al (6%), -y (3%),-ion (2%), -ity (2%), -ous (2%),-ent (1%), -ive (1%) -0 (19%), -e (7%),-ate (7%),-ation (4%), -y (4%), -ous (4%),-al (3%),-ary (3%),-ic (3%) -acious (38%) -ate (42%),-ation (18%),-al (13%),-e (8%) -0 (51%),-y (13%),-e (12%),-al (5%),-ate (4%), -ation (4%),-able (4%),-on (4%),-ion (3%),-le (3%), -ic (3%),-ar (2%),-or (2%),-ial (2%) -0 (17%),-e (7%), -ic (2%), -y (2%),-on (1%), -le (1%) -al (76%),-0 (19%),-ate (13%),-e (9%),-ation (7%), -ary (5%),-ous (5%),-able (4%),-ative (4%) -0 (38%),-ate (29%) -0 (6%),-e (2%),-al (2%),-ous (1%), -y (1%),-on (1%), -ate (1%), -ation (1%) -ant (30%),-0 (26%),-e (15%),-ate (10%),-able (9%), -ation (9%),-or (7%),-al (4%),-ous (4%),-ion (4%), -ative (3%),-ive (3%),-y (3%) -ant (40%),-0 (19%),-ation (12%) -ate (27%),-ation (21%),-0 (21%),-e (11%),-able -ication (9%), -y (5%),-al (5%),-ous (5%),-ion (4%), -ent -icative (3%),-ity (3%),-or (3%),-ive (2%),-an (1%),-ar -icatory (1%),-ic (1%),-ize (1%),-on (1%) -ician -ate (13%),-e (9%),-ation (7%), -0 (6%), -ous (2%),-y -icity (2%), -able (1%),-al (1%), -ite (1%) -ar (63%),-ate (26%),-ation (22%),-0 (13%) -icize -0 (25%), -al (13%),-ate (10%),-e (8%),-ation (8%), -ar (6%), -ous (4%),-y (4%),-able (3%),-ion (3%),-ic -ide (2%),-ity (2%),-ize (2%),-ant (2%),-or (2%) -0 (13%),-e (9%), -al (8%), -ic (4%),-y (3%), -on (1%),-le (1%),-ion (0%) -ience -ate (42%),-e (21%),-0 (18%),-al (9%),-y (3%),-ous -iency (3%),-ion (1%), -ic (1%),-on 1%-ient -ation (40%),-e (25%) -ification -ation (56%),-ate (42%), -e (19%), -0 (17%), -able (17%),-ant (12%),-al (9%),-y (5%),-ity (4%),-ous -ify (3%),-ance (3%) -ate (61%),-ation (48%),-ant (18%), -ative (18%), -able (18%),-e (15%),-al (9%),-0 (7%),-ar (6%),-ity (5%),-ous (4%),-ary (4%),-on (4%) -ation (37%),-ator (26%),-atory (26%) -ation (63%), -ate (46%),-e (21%), -ative (20%), -ator (16%),-able (15%),-0 (13%),-ant (11%),-al (7%),-ar (4%) -ate (26%),-0 (21%),-ation (18%)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "-ion -ional -ionary -ious -isation -ish -ist -ble (62%),-on (14%) -on (5%),-0 (3%),-le (1%) -ble (73%) -0 (4%)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "-istic -0 (28%),-e (13%),-or (11%),-y (6%),-ation (6%), -ment (5%),-ate (5%),-ant (3%), -al (3%),-ion (3%), -itarian -able (3%) -ite -ent (54%),-e (18%),-0 (15%),-ment (3%) -ent (73%),-ence (24%),-e (14%),-0 (12%) -0 (6%),-e (6%),-y (1%),-ate (1%),-al (1%),-ation -ity (1%) -ence (59%),-ent (59%),-0 (26%),-e (20%) -ium -e (5%),-y (4%),-0 (3%), -ic (3%), -ous (3%),-ate (3%),-on (2%) -ic (14%),-0 (7%), -y (7%),-e (4%),-ous (2%),-al -ival (1%),-ate (1%) -ire -ia (44%),-ic (19%) -0 (26%),-y (15%),-e (5%),-ate (3%),-al (2%),-ic -ivity (2%),-ize (2%) -0 (23%),-y (14%),-ic (7%),-al (6%),-e (4%),-ize -ization (3%),-ia (3%),-ity (3%),-ium (3%) -ize -iate (27%) -ial (25%),-0 (22%),-e (22%) -ial (13%),-e (9%),-0 (7%),-ate (6%),-ium (6%),-ia (5%),-ious (5%) -le -iate (70%) -ment -ible (73%),-ive (45%) -ion (25%),-ive (22%),-0 (20%),-e (12%),-or (10%), -mental -ent (7%),-able (5%),-ory (5%),-enee (4%),-al (4%), -y (4%) -e (18%),-y (14%),-0 (12%) -y (55%), -ic (11%),-0 (8%), -ize (8%),-e (6%),-ist (6%),-al (2%),-ate (2%) -ication (26%),-ic (17%),-icity (15%),-e (14%),-y (11%),-0 (7%),-ical (7%) -y (66%),-ic (14%),-e (9%) -ieation (50%),-icate (38%),-y (38%) -ication (50%), -y (43%), -icate (36%) -ic (61%),-ical (32%),-0 (16%), -e (13%),-y (13%) -ie (63%),-e (18%),-0 (16%),-y (12%),-ieal (10%), -ize (8%),-al (7%),-ieation (7%) -ie (71%) -ate (8%),-ic (8%),-0 (7%), -ite (6%),-e (4%), -on (3%), -ous (3%),-al (3%), -ize (3%),-age (2%),-ium (2%) -ient (40%) -ient (100%) -e (11%),-0 (10%) -ify (71%),-0 (22%),-e (18%),-ity (16%),-y (16%),-ic (11%) -0 (25%),-e (15%),-ic (15%),-y (15%),-ity (13%),-al (11%),-ate (9%),-ion (7%),-ite (6%),-ize (5%),-or (5%), -ar (4%), -ary (4%),-ical (4%) -e (31%),-0 (15%),-ic (1%),-y (1%),-al (1%) -ion (57%),-ire (21%),-0 (18%),-e (18%),-or (11%) -ion (87%),-e (30%),-0 (26%),-ive (26%) -y (15%),-ity (13%),-ion (10%),-0 (9%),-e (9%),-ial (6%), -ium (5%), -ie (4%), -ate (3%), -ive (3%), -ist (2%) -ization (93%),-ize (70%),-0 (53%),-ity (33%),-ist (27%), -ic (20%),-e (17%) -0 (27%), -e (11%),-y (7%),-le (2%),-ic (2%) -0 (40%),-ie (19%),-ize (18%),-y (18%),-e (14%),-al (6%),-ity (5%),-ation (3%),-ate (2%),-able (1%),-ion (1%) -ist (46%),-ize (29%),-0 (27%),-e (17%),-ic (15%), -ity (13%),-y (13%),-al (10%) -ity (57%), -ize (43%),-0 (36%),-e (36%) -0 (13%),-ic (11%),-e (6%),-ate (6%),-ous (6%),-y (2%),-ia (2%),-on (2%),-al (1%),-able (1%),-ity (1%),-ation (1%),-ion (1%),-or (1%) -0 (37%),-e (24%), -ous (6%), -ate (5%),-al (4%), -ation (3%), -y (2%), -ion (1%),-ic (1%) -ic (11%),-0 (8%),-ial (6%),-y (6%),-ia (6%),-e (6%), -ite (5%),-ate (4%),-ous (4%),-al (2%),-on (2%),-ion (2%), -ize (2%),-ist (2%) -ire (47%) -ion (59%),-e (26%),-0 (22%),-al (1%),-y (1%), -ation (1%) -ive (66%),-ion (61%),-0 (39%),-or (32%),-anee (14%),-e (14%),-ible (11%) -ize (75%),-0 (59%),-ity (31%),-ist (25%),-ic (22%) -0 (47%),-ie (17%),-ity (17%),-y (14%),-e (12%), -ous (6%),-ate (4%),-al (4%),-ite (2%),-ation (1%), -ia (1%) -0 (11%), -y (3%), -e (3%),-on (2%),-ic (1%) -0 (63%),-able (6%),-e (4%), -ation (4%), -or (3%), -ant (2%),-ate (2%),-ble (2%) -ment (77%),-0 (20%) -ment (56%) -0 (4%), -e (2%),-ic (2%), -y (1%) -ion (30%),-e (27%),-0 (22%),-ive (16%),-ation (3%), -able (3%), -y (2%), -al (2%),-ate (2%), -ent (1%),-le (1%) -ion (56%),-e (34%),-ive (21%),-or (20%),-0 (I 1%) -ous (65%),-0 (15%),-al (12%),-ate (11%),-e (11%) -0 (13%), -ic (7%), -ate (6%), -e (6%), -y (4%), -al (4%),-on (2%) -le (31%),-0 (4%),-e (4%),-ate (4%) -ular (67%),-le (28%) -0 (21%),-e (15%),-ion (11%),-or (8%),-ire (4%),-al (2%) -e (8%) -ute (67%) -0 (19%),-e (6%)", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The decomposition program uses the table above to decide which suffixes can be truncated and when. Consider the word presidency. The program notices that this word ends in -ency so it looks in the table and discovers that -ency alternates with -ent (73%), -ence (24%), -e (14%) and -0 (12%). The program tries to replace the -ency with each of these sequentially until it finds a word in the dictionary. In this case, it will succeed on the first try when it replaces -ency with -ent and finds that the result president is a word in the dictionary.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Level 1 prefixes are processed through an analogous procedure, so that effect, for example, is derived from defect by truncating the efprefix and adding the prefix de-. The truncation mechanism is not generally employed by most authors for prefixing, and it may be a mistake to do so, but I used it anyways, mostly because it was available and filled a practical need.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The forest was constructed by applying the decomposition procedure to every word in the dictionary and then indexing the results to show which forms were derived from which stems.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Thus 38 words were found to be related to the stem port and 36 words were found to be related to infect. These results seems extremely promising; most of the relations appear to agree very closely with intuition.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Now that we have a fairly accurate method of decomposing words at level 1, how can this be put to practical use'?. For assigning stress, it would be useful to know the weight of the syllables in the stem. This is particularly necessary before socalled weak retraction suffixes (e.g., -ent, -ant, -ence, -able, ance, al, ous, ary) . General principles of stress retraction (e.g., [Liberman and Prince] ), predict strong retractors (e.g., -ate, -ation) always back the stress up regardless of syllable weight (degrhde I d~gradation), whereas weak retractors do so only if the preceding syllable is light (refir / rkferent with a light syllable before -ent, as opposed to (cohkre /cohkrent with a heavy syllable before -ent).", |
| "cite_spans": [ |
| { |
| "start": 284, |
| "end": 329, |
| "text": "-ent, -ant, -ence, -able, ance, al, ous, ary)", |
| "ref_id": null |
| }, |
| { |
| "start": 379, |
| "end": 400, |
| "text": "[Liberman and Prince]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Given syllable weight, it is relatively well-understood how to assign stress. A large number of phonological studies (e.g., [Chomsky and Halle] , [Liberman and Prince] , [Hayes] ) outline a deterministic procedure for assigning stress from the weight representation and the number of extrametrical syllables (1 for nouns, 0 for verbs). A version of this procedure was implemented by Richard Sproat last summer, and was discussed at the last ACL meeting [Church] .", |
| "cite_spans": [ |
| { |
| "start": 124, |
| "end": 143, |
| "text": "[Chomsky and Halle]", |
| "ref_id": null |
| }, |
| { |
| "start": 146, |
| "end": 167, |
| "text": "[Liberman and Prince]", |
| "ref_id": null |
| }, |
| { |
| "start": 170, |
| "end": 177, |
| "text": "[Hayes]", |
| "ref_id": null |
| }, |
| { |
| "start": 453, |
| "end": 461, |
| "text": "[Church]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "It it generally believed that syllable weight is derivable from underlying vowel length and the number of consonants, but if one is trying to assign stress from the spelling, it can be difficult to know the vowel length and the number of consonants. The fact that inhence has a heavy penultimate syllable and that ~nference has a light penultimate syllable is extremely difficult to determine from the spelling. It would be considerably easier if syllable weight (or some correlate thereof such as vowel length) were marked in a lexicon of stems, so that the program could determine syllable weight by decomposing a word into its peices, look them up in a morpheme lexicon, and then re-combine the results appropriately.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Not only is it convenient for practical application to assume that stems are marked in the lexicon for syllable weight, but it may be necessary for linguistic reasons as well. Consider the stress alternation confide I confidence. This alternation is problematic because the i in confide seems to be underlyingly long whereas the i in confidence seems to be underlyingly short, and yet, the two stems ought to share the same underlying form since the two words are morphologically related to one another. The solution to the confidence puzzle, I believe, is to say that the stem -fide is marked in the lexicon as underlyingly light at least with respect to stress retraction (and to account for the tense vowel in confide in some other way [Church (forthcoming) ", |
| "cite_spans": [ |
| { |
| "start": 739, |
| "end": 760, |
| "text": "[Church (forthcoming)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "]).", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "The table below is presented as evidence that the confidence alternation is determined, at least in part, by some sort of lexical marking on stems. Note, for example, that -fer, -cel, -side, and -fide words display the confidence alternation, but -here, -pel, and -pose words do not. -fer, -cel, -side, -fide, -main, -vail, -note, -cede, -pete, -pair, -pare \u2022 Retraction Class II Stems (heavy): -here, -pel, -pose, -hale, -pale, -grade, -vade, -flame, -suade, -place, -plore, -void, -clude, -prove,-sume, -fuse, - . Some results of the new decomposition procedure were presented, and they seem to agree very closely with intuition. It was suggested that the decomposition procedure could be used in stress assignment, by decomposing words into morphemes, look up the syllable weight of the pieces in a morpheme lexicon, and then recombine the results appropriately. This last suggestion has not yet been fully implemented.", |
| "cite_spans": [ |
| { |
| "start": 284, |
| "end": 357, |
| "text": "-fer, -cel, -side, -fide, -main, -vail, -note, -cede, -pete, -pair, -pare", |
| "ref_id": null |
| }, |
| { |
| "start": 402, |
| "end": 513, |
| "text": "-pel, -pose, -hale, -pale, -grade, -vade, -flame, -suade, -place, -plore, -void, -clude, -prove,-sume, -fuse, -", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Morphological Decomposition of Level I Affixes", |
| "sec_num": "4." |
| }, |
| { |
| "text": "Most of the linguistic literature deals with level 1 where we find extremely interesting stress alternations and vowel shifts and so forth. Generally speaking, the phonology of level 2 and compounding is believed to be relatively fairly straightforward. Something like the simple concatenation model in decomp is not a bad first approximation. In fact, I believe the stress of level 2 and compounding is more interesting than has generally been thought. In particular, I am beginning to believe that level 2 affixes are not stress neutral at all, but rather they stress as if they were parts of compounds. Note that under-, antiand superfollow the general compound pattern where stress is assigned the to the left member in nouns and to the right in verbs and adjectives.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Level 2 and Compounding", |
| "sec_num": "5." |
| }, |
| { |
| "text": "has two stress patterns, one with main stress on the left and one with main stress on the right, just like well-known compound blackboard. With left stress, the compound takes noncompositional semantics and with right stress the compound has a more compositional meaning. These facts suggest that the compound theory can be maintained to acocunt for cases like pro-life, but only if the compound stress rules are refined take the semantic facts into account.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Level 2 and Compounding", |
| "sec_num": "5." |
| }, |
| { |
| "text": "Level 2 suffixes provide additional support for the compound theory. Consider suffixes like ment, hood, ship and ness which appear to support the the stress neutral theory because they never receive main stress. But, they can also be accounted for under the compound theory because they form nouns, and therefore the main stress would be expected to fall on the left.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Level 2 and Compounding", |
| "sec_num": "5." |
| }, |
| { |
| "text": "Moreover, consider the level 2 adjectival suffixes -istic and -mental. l These suffixes refute the stress neutral theory because they take the main stress, but they are no problem for the compound stress theory which predicts that adjectivial compounds should receive main stress on the right.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Level 2 and Compounding", |
| "sec_num": "5." |
| }, |
| { |
| "text": "In attempting to include prefixes as a subcase of compound stress, I did stumble over a very interesting problem in the theory of compound stress. Consider the contrast between sl~perconductor and shperconductlvity.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Super-Puzzle and Compound Stress", |
| "sec_num": "7." |
| }, |
| { |
| "text": "Although both compounds are nominal, the first takes primary stress on the left member and the second takes stress on the right member. Upon further investigation, it appears than many compounds ending with level 1 suffixes. (e.g., -ity, -ation) take primary stress on the right member. For example, here is a breakdown of compounds ending with the letters ion. Note the strong tendency for primary stress to end up on the right member. ~", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Super-Puzzle and Compound Stress", |
| "sec_num": "7." |
| }, |
| { |
| "text": "\u2022 Left-Dominant: intersession, outstation, midsection supervision, anteversion, intercession, supersession, intermission, echolocation, intercolumniation, contravallation, overpopulation, interlunation, intermigration, overcompensation, aftersensation, superfetation, superelevation, interaction, intersection, contradistinction, superinduction, superconduction, underproduction, contraposition, superposition, interposition, postposition, interlocution, counterrevolution \u2022 Neither:", |
| "cite_spans": [ |
| { |
| "start": 54, |
| "end": 472, |
| "text": "supervision, anteversion, intercession, supersession, intermission, echolocation, intercolumniation, contravallation, overpopulation, interlunation, intermigration, overcompensation, aftersensation, superfetation, superelevation, interaction, intersection, contradistinction, superinduction, superconduction, underproduction, contraposition, superposition, interposition, postposition, interlocution, counterrevolution", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Super-Puzzle and Compound Stress", |
| "sec_num": "7." |
| }, |
| { |
| "text": "\u2022 Right-Dominant: intercommunion,", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "The Super-Puzzle and Compound Stress", |
| "sec_num": "7." |
| }, |
| { |
| "text": "Thus, it appears that compounds ending with a level 1 suffix take right stress. If correct, however, the generalization is a puzzle for the level ordering hypothesis, which assumes that the stress rules of level 1 are opaque to the stress rules of level 2 and compounding. In other words, level ordering suggests a structure like super [conductivity] where level 1 takes precedence over level 2 and compounding, but stress assignment requires a different structure [superconductive]ity where the compound stress rule applies before the level 1 suffix is analyzed.", |
| "cite_spans": [ |
| { |
| "start": 336, |
| "end": 350, |
| "text": "[conductivity]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "tourbillion, interrogation, foreordination, redintegration forestation, electrodeposition 3", |
| "sec_num": null |
| }, |
| { |
| "text": "1. These suffixes cannot be level 1, because they don't force the secondary stress to fall two syllables before the main stress: *dbpartmbntal (cf.. dbgrad [ttion) .", |
| "cite_spans": [ |
| { |
| "start": 156, |
| "end": 163, |
| "text": "[ttion)", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "tourbillion, interrogation, foreordination, redintegration forestation, electrodeposition 3", |
| "sec_num": null |
| }, |
| { |
| "text": "In this sense, words like superconductivity are very much like the well-known bracketing paradox ungrammaticality, where level ordering suggests one structure un[grammaticality] (un# is a level 2 prefix which must scope outside of +ity with is a level 1 prefix) and syntactic/semantic interpretation (LF) requires another [ungrammatical] ity (un# attaches to adjectives and not to nouns). Note that stress assignment seems to side with the syntactic/semantic arguments in suggesting a left branching structure that violates level ordering.", |
| "cite_spans": [ |
| { |
| "start": 322, |
| "end": 337, |
| "text": "[ungrammatical]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "tourbillion, interrogation, foreordination, redintegration forestation, electrodeposition 3", |
| "sec_num": null |
| }, |
| { |
| "text": "A solution to these bracketting paradoxes becomes apparent when we consider nominal Greek compounds like psychobiology with three or more morphemes. Notice that these compounds systematically take main stress on the middle morpheme. aeroneurosis, aerothermodynamics, astrobiology, astrogeology, astrophotography, autobiography, autohypnosis, autoradiograph autoradiography, biogeography, biophysicist, biotechnology, chromolithograph, chromolithography, chronobiology, cryobiology, diageotropism, electroanalysis, electrocardiogram, electrocardiograph, electrodialysis, electrodynamometer, electroencephalogram, electroencephalograph, electroencephalography, electrophysiology, endoparasite, epidiascope, geochronology, geomorphology, heterochromatin, heterochromosome, histopathology, hypnoanalysis, magnetohydrodynamics, metaphysicist, metapsychology, microanalysis, microbarograph, microbiology, micrometeorology, micropaleontology, microparasite, microphotograph, microphotography, multivibrator, myocardiograph, neoorthodoxy, neuropathology, neurophysiology, orthohydrogen, otolaryngology, paleoethnobotany, parahydrogen, parapsychology, photochronograph, photoelectrotype, photogeology, photolithograph, photolithography, photomicrograph, photopolymer, phototelegraphy, phototypography, photozincograph, photozincography, pneumoencephalogram, pneumoencephalography, psychoanalyse, psychoanalysis, psychoanalyze, psychobiology, psychoneurosis, psychopathology, psychopharmacology, psychophysiology, radioautograph, radiobiology, radiomicrometer, radiotelegram, radiotelegraph, radiotelegraphy, radiotelemetry, radiotelephone, radiotelephony, semidiameter, semiparasite, spectroheliograph, spectrophotometer, stereoisomer, stereoisomerism, telephotography, telespectroscope, telestereoscope, teletypewriter, thermobarograph, thermobarometer, ultramicrometer, ultramicroscope, ultramicroscopy Assume that compounds take stress on the right member when it is branching (bi-morphemic). Thus, psycho [biology] takes main stress on the biology because it is branching.", |
| "cite_spans": [ |
| { |
| "start": 233, |
| "end": 1895, |
| "text": "aeroneurosis, aerothermodynamics, astrobiology, astrogeology, astrophotography, autobiography, autohypnosis, autoradiograph autoradiography, biogeography, biophysicist, biotechnology, chromolithograph, chromolithography, chronobiology, cryobiology, diageotropism, electroanalysis, electrocardiogram, electrocardiograph, electrodialysis, electrodynamometer, electroencephalogram, electroencephalograph, electroencephalography, electrophysiology, endoparasite, epidiascope, geochronology, geomorphology, heterochromatin, heterochromosome, histopathology, hypnoanalysis, magnetohydrodynamics, metaphysicist, metapsychology, microanalysis, microbarograph, microbiology, micrometeorology, micropaleontology, microparasite, microphotograph, microphotography, multivibrator, myocardiograph, neoorthodoxy, neuropathology, neurophysiology, orthohydrogen, otolaryngology, paleoethnobotany, parahydrogen, parapsychology, photochronograph, photoelectrotype, photogeology, photolithograph, photolithography, photomicrograph, photopolymer, phototelegraphy, phototypography, photozincograph, photozincography, pneumoencephalogram, pneumoencephalography, psychoanalyse, psychoanalysis, psychoanalyze, psychobiology, psychoneurosis, psychopathology, psychopharmacology, psychophysiology, radioautograph, radiobiology, radiomicrometer, radiotelegram, radiotelegraph, radiotelegraphy, radiotelemetry, radiotelephone, radiotelephony, semidiameter, semiparasite, spectroheliograph, spectrophotometer, stereoisomer, stereoisomerism, telephotography, telespectroscope, telestereoscope, teletypewriter, thermobarograph, thermobarometer, ultramicrometer, ultramicroscope, ultramicroscopy", |
| "ref_id": null |
| }, |
| { |
| "start": 2000, |
| "end": 2009, |
| "text": "[biology]", |
| "ref_id": null |
| } |
| ], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "tourbillion, interrogation, foreordination, redintegration forestation, electrodeposition 3", |
| "sec_num": null |
| }, |
| { |
| "text": "Let me suggest further that this same sort of explanation might carry over to explain the stress in the bracketting paradoxes such as superconductivity and ungrammaticality where I claim that the right piece is 'branching' in order to account for the fact that main stress ends up on the right half. 4 Note that I am 2. None of the left dominant words above end in the suffix +ion. Note, for example, the contrast between lnter'session and inter-ebss+ion. The left dominant case does not end in the su/fix +ion: the right dominant case does. 3. Almost all of these exceptions are due to errors in morphological decomposition algorithm. Tour # billion, inter # rogation, fore # station. and electrode # position are all incorrect analyses. It is highly unusual for the algorithm to make this many mistakes. using the lexical category prominance rule in order to let one bit of information [+branching] pass through the opacity imposed by level ordering.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "tourbillion, interrogation, foreordination, redintegration forestation, electrodeposition 3", |
| "sec_num": null |
| }, |
| { |
| "text": "Two new ideas in machine morphological decomposition were presented. The discussion of level 1 proposed the application of Aronoff-style truncation rules as an effective means to capture the asymmetry in the 'derived from' relation. Secondly, the discussion of level 2 proposed ideas from the literature on compound stress as an alternative to the stress neutral approach taken in MITalk's Decomp.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "Conclusion", |
| "sec_num": "8." |
| }, |
| { |
| "text": ". The problem is to define 'branching' so that it gets the right results. 1 don't want to say that superconductor is branching, because that would incorrectly predict main stress on conductor. I don't know how to define branching to achieve the desired results, though 1 believe that thi~ approach is extremely promising.", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "", |
| "sec_num": null |
| } |
| ], |
| "back_matter": [ |
| { |
| "text": "The resulting decomposition program has been used to construct a forest of related words as illustrated below:( 38 port ( aport ) (comport (cosportmtnt)) (deport (depoEtatlon) (doporCee) It might be possible to extend this position to its logical extreme and say that all level 2 affixes stress like compounds, and thus completely do away with the concept of stress neutral affixes.\u2022 Compound Theory: (All) Level 2 affixes are stressed just like compounds; they receive main stress on the left in nouns and main stress on the right in verbs and adjectives.\u2022 Stress Neutral Theory: (At least some) Level 2 affixes are stress neutral; they are simply concatenated onto the stem (a 1~. MITalk's Decomp).The compound theory has much to recommend it. Indeed most level 2 prefixes are like under-, antiand superand show the compound stress pattern (stress on the left when nominal and on the right when verbal/adjectival). These prefixes cannot be accounted for easily under the stress neutral theory. The main support for the stress neutral theory seems to come from prefixes like unwhich (almost) never take the main stress. However, uncan also be accounted for under the compound theory by noting that unforms adjectives and verbs, and therefore main stress would fall on the right.Admittedly, there are a number of nominal compounds like pro-life and anti-abortion which take right stress, presumably because the semantics of the left member takes on a semiadjectival status. Notice, for example, that the word antimatter", |
| "cite_spans": [], |
| "ref_spans": [], |
| "eq_spans": [], |
| "section": "annex", |
| "sec_num": null |
| } |
| ], |
| "bib_entries": { |
| "BIBREF0": { |
| "ref_id": "b0", |
| "title": "Word Formation in Generative Grammar", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Aronoff", |
| "suffix": "" |
| } |
| ], |
| "year": 1976, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Aronoff, M., Word Formation in Generative Grammar, MIT Press, Cambridge, MA., 1976.", |
| "links": null |
| }, |
| "BIBREF1": { |
| "ref_id": "b1", |
| "title": "Conversion of Unrestricted English Text to Speech", |
| "authors": [ |
| { |
| "first": "J", |
| "middle": [], |
| "last": "Allen", |
| "suffix": "" |
| }, |
| { |
| "first": "R", |
| "middle": [], |
| "last": "Carlson", |
| "suffix": "" |
| }, |
| { |
| "first": "B", |
| "middle": [], |
| "last": "Granstrom", |
| "suffix": "" |
| }, |
| { |
| "first": "S", |
| "middle": [], |
| "last": "Hunnicutt", |
| "suffix": "" |
| }, |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Klatt", |
| "suffix": "" |
| }, |
| { |
| "first": "D", |
| "middle": [], |
| "last": "Pisoni", |
| "suffix": "" |
| } |
| ], |
| "year": 1979, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Allen, J., Carlson, R., Granstrom, B., Hunnicutt, S., Klatt, D., Pisoni, D., Conversion of Unrestricted English Text to Speech, incomplete draft, undergroland press, 1979.", |
| "links": null |
| }, |
| "BIBREF2": { |
| "ref_id": "b2", |
| "title": "The Sound Pattern of English", |
| "authors": [ |
| { |
| "first": "N", |
| "middle": [], |
| "last": "Chomsky", |
| "suffix": "" |
| }, |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Halle", |
| "suffix": "" |
| } |
| ], |
| "year": 1968, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Chomsky, N., and Halle, M., The Sound Pattern of English, Harper and Row, 1968.", |
| "links": null |
| }, |
| "BIBREF3": { |
| "ref_id": "b3", |
| "title": "Stress Assignment in Letter to Sound Rules for Speech Synthesis", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Church", |
| "suffix": "" |
| } |
| ], |
| "year": 1985, |
| "venue": "Proceedings of the Association for Computational Linguistics", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Church, K., Stress Assignment in Letter to Sound Rules for Speech Synthesis, in Proceedings of the Association for Computational Linguistics, 1985.", |
| "links": null |
| }, |
| "BIBREF4": { |
| "ref_id": "b4", |
| "title": "The Confidence Puzzle and Underlying Quantity, forthcoming", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Church", |
| "suffix": "" |
| } |
| ], |
| "year": null, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Church, K., The Confidence Puzzle and Underlying Quantity, forthcoming.", |
| "links": null |
| }, |
| "BIBREF5": { |
| "ref_id": "b5", |
| "title": "A Metrical Theory of Stress Rules", |
| "authors": [ |
| { |
| "first": "B", |
| "middle": [], |
| "last": "Hayes", |
| "suffix": "" |
| } |
| ], |
| "year": 1980, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Hayes, B., A Metrical Theory of Stress Rules, Ph.D. Thesis, MIT, 1980.", |
| "links": null |
| }, |
| "BIBREF6": { |
| "ref_id": "b6", |
| "title": "On Stress and Linguistic Rhythm", |
| "authors": [ |
| { |
| "first": "M", |
| "middle": [], |
| "last": "Liberman", |
| "suffix": "" |
| }, |
| { |
| "first": "A", |
| "middle": [], |
| "last": "Prince", |
| "suffix": "" |
| } |
| ], |
| "year": 1977, |
| "venue": "Linguistic Inquiry", |
| "volume": "8", |
| "issue": "", |
| "pages": "249--336", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Liberman, M., and Prince, A., On Stress and Linguistic Rhythm, Linguistic Inquiry 8, pp. 249-336, 1977.", |
| "links": null |
| }, |
| "BIBREF7": { |
| "ref_id": "b7", |
| "title": "The Categories and Types of Present-Day English Word-Formation", |
| "authors": [ |
| { |
| "first": "H", |
| "middle": [], |
| "last": "Marchand", |
| "suffix": "" |
| } |
| ], |
| "year": 1969, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Marchand, H., The Categories and Types of Present-Day English Word-Formation, University of Alabama Press, 1969.", |
| "links": null |
| }, |
| "BIBREF8": { |
| "ref_id": "b8", |
| "title": "Lexical Phonology, MIT Doctoral Dissertation, available for the Indiana University Linguistics Club", |
| "authors": [ |
| { |
| "first": "K", |
| "middle": [], |
| "last": "Mohanan", |
| "suffix": "" |
| } |
| ], |
| "year": 1982, |
| "venue": "", |
| "volume": "", |
| "issue": "", |
| "pages": "", |
| "other_ids": {}, |
| "num": null, |
| "urls": [], |
| "raw_text": "Mohanan, K., Lexical Phonology, MIT Doctoral Dissertation, available for the Indiana University Linguistics Club, 1982.", |
| "links": null |
| } |
| }, |
| "ref_entries": { |
| "FIGREF1": { |
| "num": null, |
| "uris": null, |
| "type_str": "figure", |
| "text": "(derv): laxiTY, existENCE, softNESS, kingDOM b. Inflectional (infl): boatiNG, toastED, coatS, roanS\" Roots a. Free (root): stay, squeeze, large b. Absolute (absl): the, than, but c. Left-Bound (lbrt): rePEL, conCEIVE d. Right-Bound (rbrt): CRIMINal, TOLERance e. Strong (root): women, rang" |
| }, |
| "FIGREF2": { |
| "num": null, |
| "uris": null, |
| "type_str": "figure", |
| "text": "Group A is represented by the pairs 1) -acy / 2) -ate (as piracy ~ pirate), 1) -ancy, -ency / 2) -ant, ent (as militancy ~ militant, decency ~ decent), 1) -ization / 2) -ize (as civilization ~ civilize), 1) -ification I 2) -ify (as identification ~ identify), 1) -ability / 2) -able (as respectibility ~ respectible), 1) -ibility /2) -ible as (convertibility ~ convertible), 1) -ician / 2) -it(s) (as statistician ~ statistics), 1) -icity / 2) -ic (as catholicity catholic), 1) -inity / 2) -ine (salinity ~ saline)." |
| }, |
| "TABREF1": { |
| "type_str": "table", |
| "text": "Consider how the first difference mentioned above, historical correlation, could be used to improve a decomposition program. It is very easy, for example, for a decomposition program to decide erroneously that acclamation is derived from clam, meaning roughly the result of having been clammed up. If the program could somehow split", |
| "num": null, |
| "content": "<table/>", |
| "html": null |
| }, |
| "TABREF2": { |
| "type_str": "table", |
| "text": "These truncation rules were necessary for him so that he could maintain his Word Based Hypothesis. The Word Based Hypothesis claims that words are formed from other words (possibly via truncation) and not from bound morphemes. Thus, in Aronoff's theory, there is no bound morpheme nomin-; there are only words (e,g., nominate and nominee).", |
| "num": null, |
| "content": "<table><tr><td>theories are captured in Aronoff's system by his truncation rules,</td></tr><tr><td>The</td></tr><tr><td>generalizations that would be attributed to nomin-in other</td></tr></table>", |
| "html": null |
| } |
| } |
| } |
| } |
