ࡱ> bdafgRM kbjbj== WWglX X X X P L L L ,!$P!P j"^f"f"f"f"E#Z##iiiiiii$k mpi #A#E###i&X X f"f"i&&&#|X Rf" f"i&#i&&#.2a " Lff"! 0`P L G$ c.LfDi0jNcOo %OoLf&P P X X X X A best-first alignment algorithm for automatic extraction of transfer mappings from bilingual corpora Arul Menezes and Stephen D. Richardson Microsoft Research One Microsoft Way Redmond, WA 98008, USA  HYPERLINK "mailto:arulm@microsoft.com" arulm@microsoft.com  HYPERLINK "mailto:steveri@microsoft.com" steveri@microsoft.com Abstract Translation systems that automatically extract transfer mappings (rules or examples) from bilingual corpora have been hampered by the difficulty of achieving accurate alignment and acquiring high quality mappings. We describe an algorithm that uses a best-first strategy and a small alignment grammar to significantly improve the quality of the transfer mappings extracted. For each mapping, frequencies are computed and sufficient context is retained to distinguish competing mappings during translation. Variants of the algorithm are run against a corpus containing 200K sentence pairs and evaluated based on the quality of resulting translations. Introduction A machine translation system requires a substantial amount of translation knowledge typically embodied in bilingual dictionaries, transfer rules, example bases, or a statistical model. Over the last decade, research has focused on the automatic acquisition of this knowledge from bilingual corpora. Statistical systems build translation models from this data without linguistic analysis (Brown, 1993). Another class of systems, including our own, parses sentences in parallel sentence-aligned corpora to extract transfer rules or examples (Kaji, 1992) (Meyers, 2000) (Watanabe, 2000). These systems typically obtain a predicate-argument or dependency structure for source and target sentences, which are then aligned, and from the resulting alignment, lexical and structural translation correspondences are extracted, which are then represented as a set of rules or an example-base for translation. However, before this method of knowledge acquisition can be fully automated, a number of issues remain to be addressed. The alignment and transfer-mapping acquisition procedure must acquire rules with very high precision. It must be robust against errors introduced by parsing and sentence-level alignment, errors intrinsic to the corpus, as well as errors resulting from the alignment procedure itself. The procedure must also produce transfer mappings that provide sufficient context to enable the translation system utilizing these mappings to choose the appropriate translation for a given context. In this paper, we describe the alignment and transfer-acquisition algorithm used in our machine translation system, which attempts to address the issues raised above. This system acquires transfer mappings by aligning pairs of logical form structures (LFs) similar to those described by Jensen (1993). These LFs are obtained by parsing sentence pairs from a sentence-aligned bilingual corpus. (The problem of aligning parallel corpora at the sentence level has been addressed by Meyers (1998b) Chen (1993) and others and is beyond the scope of this paper). We show that alignment using a best-first strategy in conjunction with a small alignment grammar improves the alignment and the quality of the acquired transfer mappings.  EMBED Visio.Drawing.6  Logical Form A Logical Form (LF) is an unordered graph representing the relations among the most meaningful elements of a sentence. Nodes are identified by the lemma of a content word and directed, labeled arcs indicate the underlying semantic relations. Logical Forms are intended to be as independent as possible of specific languages and their grammars. In particular, Logical Forms from different languages use the same relation types and provide similar analyses for similar constructions. The logical form abstracts away from such language-particular aspects of a sentence as constituent order, inflectional morphology, and certain function words. Figure 1a illustrates the LFs for the following Spanish sentence and its corresponding English translation, which we use in example below. En Informacin del hipervnculo, haga clic en la direccin del hipervnculo. Under Hyperlink Information, click the hyperlink address. Alignment We consider an alignment of two logical forms to be a set of mappings, such that each mapping is between a node or set of nodes (and the relations between them) in the source LF and a node or set of nodes (and the relations between them) in the target LF, where no node participates in more than one such mapping. In other words, we allow one-to-one, one-to-many, many-to-one and many-to-many mappings but the mappings do not overlap. Our alignment algorithm proceeds in two phases. The first phase establishes tentative lexical correspondences between nodes in the source and target LFs. The second phase aligns nodes based on these lexical correspondences as well as structural considerations. The algorithm starts from the nodes with the tightest lexical correspondence (best-first) and works outward from these anchor points. We first present the algorithm, and then illustrate how it applies to the sentence-pair in Figure-1. Finding tentative lexical correspondences We use a bilingual lexicon that merges data from several sources (CUP, 1995), (SoftArt, 1995), (Langenscheidt, 1997), and inverts target-to-source dictionaries to improve coverage. Our Spanish-English lexicon contains 88,500 translation pairs. We augment this with 19,762 translation correspondences acquired using statistical techniques described by Moore (2001). Like Watanabe (2000) and Meyers (2000), we use a lexicon to establish initial tentative word correspondences. However, we have found that even a relatively large bilingual dictionary has only moderately good coverage for our purposes. Hence, we pursue an aggressive matching strategy for establishing tentative word correspondences. Using the bilingual dictionary together with the derivational morphology component in our system (Pentheroudakis, 1993), we find direct translations, translations of morphological bases and derivations, and base and derived forms of translations. Fuzzy string matching is also used to identify possible correspondences. We have found that aggressive over-generation of correspondences at this phase is balanced by the more conservative second phase and results in improved overall alignment quality. We also look for matches between components of multi-word expressions and individual words. This allows us to align such expressions that may have been analyzed as a single lexicalized entity in one language but as separate words in the other. Aligning nodes Our alignment procedure uses the tentative lexical correspondences established above, as well as structural cues, to create affirmative node alignments. A set of alignment grammar rules licenses only linguistically meaningful alignments. The rules are ordered to create the most unambiguous alignments (best) first and use these to disambiguate subsequent alignments. The algorithm and the alignment grammar rules are intended to be applicable across multiple languages. The rules were developed while working primarily with a Spanish-English corpus, but have also been applied to other language pairs such as French, German, and Japanese to/from English. The algorithm is as follows: Initialize the set of unaligned source and target nodes to the set of all source and target nodes respectively. Attempt to apply the alignment rules in the specified order, to each unaligned node or set of nodes in source and target. If a rule fails to apply to any unaligned node or set of nodes, move to the next rule. If all rules fail to apply to all nodes, exit. No more alignment is possible. (Note: some nodes may remain unaligned). When a rule applies, mark the nodes or sets of nodes to which it applied as aligned to each other and remove them from the lists of unaligned source and target nodes respectively. Go to step 2 and apply rules again, starting from the first rule. The alignment grammar currently consists of 18 rules. Below we provide the specification for some of the most important rules. Bidirectionally unique translation: A set of contiguous source nodes S and a set of contiguous target nodes T such that every node in S has a lexical correspondence with every node in T and with no other target node, and every node in T has a lexical correspondence with every node in S and with no other source node. Align S and T to each other. Translation + Children: A source node S and a target node T that have a lexical correspondence, such that each child of S and T is already aligned to a child of the other. Align S and T to each other. Translation + Parent: A source node S and a target node T that have a lexical correspondence, such that a parent Ps of S has already been aligned to a parent Pt of T. Align S and T to each other. Verb+Object to Verb: A verb V1 (from either source or target), that has child O that is not a verb, but is already aligned to a verb V2, and either V2 has no unaligned parents, or V1 and V2 have children aligned to each other. Align V1+O to V2. Parent + relationship: A source node S and a target node T, with the same part-of-speech, and no unaligned siblings, where a parent Ps of S is already aligned to a parent Pt of T, and the relationship between Ps and S is the same as that between Pt and T. Align S and T to each other. Child + relationship: Analogous to previous rule but based on previously aligned children instead of parents. Note that rules 4-6 do not exploit lexical correspondence, relying solely on relationships between nodes being examined and previously aligned nodes. Alignment Example In this section, we illustrate the application of the alignment procedure to the example in Figure 1. In the first phase, using the bilingual lexicon, we identify the lexical correspondences depicted in Figure-1a as dotted lines. Note that each of the two instances of hipervnculo has two ambiguous correspondences, and that while the correspondence from Informacin to Hyperlink Information is unique, the reverse is not. Note also that neither the monolingual nor bilingual lexicons have been customized for this domain. For example, there is no entry in either lexicon for Hyperlink_Information. This unit has been assembled by general-purpose "Captoid" grammar rules. Similarly, lexical correspondences established for this unit are based on translations found for its individual components, there being no lexicon entry for the captoid as a whole. In the next phase, the alignment rules apply to create alignment mappings depicted in Figure-1b as dotted lines. Rule-1: Bidirectionally unique translation, applies in three places, creating alignment mappings between direccin and address, usted and you, and clic and click. These are the initial best alignments that provide the anchors from which we will work outwards to align the rest of the structure. Rule-3: Translation + Parent, applies next to align the instance of hipervnculo that is the child of direccin to hyperlink, which is the child of address. We leverage a previously created alignment (direccin to address) and the structure of the logical form to resolve the ambiguity present at the lexical level. Rule-1 now applies (where previously it did not) to create a many-to-one mapping between informacin and hipervnculo to Hyperlink_Information. The uniqueness condition in this rule is now met because the ambiguous alternative was cleared away by the prior application of Rule-3. Rule-4: Verb+Object to Verb applies to rollup hacer with its object clic, since the latter is already aligned to a verb. This produces the many-to-one alignment of hacer and clic to click Acquiring Transfer Mappings Figure-2 shows the transfer mappings derived from the alignment example in Figure-1.  EMBED Visio.Drawing.6  Transfer mappings with context Each mapping created during alignment forms the core of a family of mappings emitted by the transfer mapping acquisition procedure. The alignment mapping by itself represents a minimal transfer mapping with no context. In addition, we emit multiple variants, each one expanding the core mapping with varying types and amounts of local context. We use linguistic constructs such as noun and verb phrases to provide the boundaries for the context we include. For example, the transfer mapping for an adjective is expanded to include the noun it modifies; the mapping for a modal verb is expanded to include the main verb; the mapping for a main verb is expanded to include its object; mappings for collocations of nouns are emitted individually and as a whole. Mappings may include wild card or under-specified nodes, with a part of speech, but no lemma, as shown in Figure 2. Alignment Post-processing After we have acquired transfer mappings from our entire training corpus, we compute frequencies for all mappings. We use these to resolve conflicting mappings, i.e. mappings where the source sides of the mapping are identical, but the target sides differ. Currently we resolve the conflict by simply picking the most frequent mapping. Note that this does not imply that we are committed to a single translation for every word across the corpus, since we emitted each mapping with different types and amounts of context (see section  REF _Ref511634362 \r 4.1). Ideally at least one of these contexts serves to disambiguate the translation. The conflicts being resolved here are those mappings where the necessary context is not present. A drawback of this approach is that we are relying on a priori linguistic heuristics to ensure that we have the right context. Our future work plans to address this by iteratively searching for the context that serves to optimally disambiguate (across the entire training corpus) between conflicting mappings. Frequency Threshold During post-processing we also apply a frequency threshold, keeping only mappings seen at least N times (where N is currently 2). This frequency threshold greatly improves the speed of the runtime system, with negligible impact on translation quality (see section 5.6). Experiments and Results Evaluation methodology In the evaluation process, we found that various evaluation metrics of alignment in isolation bore very little relationship to the quality of the translations produced by a system that used the results of such alignment. Since it is the overall translation quality that we care about, we use the output quality (as judged by humans) of the MT system incorporating the transfer mappings produced by an alignment algorithm (keeping all other aspects of the system constant) as the metric for that algorithm. Translation system Our translation system (Richardson, 2001) begins by parsing an input sentence and obtaining a logical form. We then search the transfer mappings acquired during alignment, for mappings that match portions of the input LF. We prefer larger (more specific) mappings to smaller (more general) mappings. Among mappings of equal size, we prefer higher-frequency mappings. We allow overlapping mappings that do not conflict. The lemmas in any portion of the LF not covered by a transfer mapping are translated using the same bilingual dictionary employed during alignment, or by a handful of hard-coded transfer rules (see Section 5.7 for a discussion of the contribution made by each of these components). Target LF fragments from matched transfer mappings and default dictionary translations are stitched together to form an output LF. From this, a rule-based generation component produces an output sentence. The system provides output for every input sentence. Sentences for which spanning parses are not found are translated anyway, albeit with lower quality. Training corpus We use a sentence-aligned Spanish-English training corpus consisting of 208,730 sentence pairs mostly from technical manuals. The data was already aligned at the sentence-level since it was taken from sentence-level translation memories created by human translators using a commercial translation-memory product. This data was parsed and aligned at the sub-sentence level by our system, using the techniques described in this paper. Our parser produces a parse in every case, but in each language roughly 15% of the parses produced are fitted or non-spanning. Since we have a relatively large training corpus, we apply a conservative heuristic and only use in alignment those sentence-pairs that produced spanning parses in both languages. In this corpus 161,606 pairs (or 77.4% of the corpus) were used. This is a substantially larger training corpus than those used in previous work on learning transfer mappings from parsed data. Table-1 presents some data on the mappings extracted from this corpus using Best-First. Total Sentence pairs208,730Sentence pairs used161,606Number of transfer mappings1,202,828Transfer mappings per pair7.48Num. unique transfer mappings437,479Num. unique after elim. conflicts369,067Num. unique with frequency > 158,314Time taken to align entire corpus (on a 800MHz PC)74 minutesAlignment speed35.6 sent/sTable-1: Best-first alignment of training corpus Experiments In each experiment we used 5 human evaluators in a blind evaluation, to compare the translations produced by the test system with those produced by a comparison system. Evaluators were presented, for each sentence, with a reference human translation and with the two machine translations in random order, but not the original source language sentence. They were asked to pick the better overall translation, taking into account both content and fluency. They were allowed to choose Neither if they considered both translations equally good or equally bad. All the experiments were run with our Spanish-English system. The test sentences were randomly chosen from unseen data from the same domain. Experiment-1 used 200 sentences and each sentence was evaluated by all raters. Sentences were rated better for one system or the other if a majority of the raters agreed. Experiments 2-4 used 500 sentences each, but each sentence was rated by a single rater. In each experiment, the test system was the system described in section 5.2, loaded with transfer mappings acquired using the techniques described in this paper (hereafter Best-First). Comparison systems In the first experiment the comparison system is a highly rated commercial system, Babelfish (http://world.altavista.com). Each of the next three experiments varies some key aspect of Best-First in order to explore the properties of the algorithm. Bottom Up Experiment-2 compares Best-First to the previous algorithm we employed, which used a bottom-up approach, similar in spirit to that used by Meyers (1998a). This algorithm follows the procedure described in section 3.1 to establish tentative lexical correspondences. However, it does not use an alignment grammar, and relies on a bottom-up rather than a best-first strategy. It starts by aligning the leaf nodes and proceeds upwards, aligning nodes whose child nodes have already aligned. Nodes that do not align are skipped over, and later rolled-up with ancestor nodes that have successfully aligned. No Context Experiment-3 uses a comparison algorithm that differs from Best First in that it retains no context (see section 4.1) when emitting transfer mappings. No Threshold The comparison algorithm used in Experiment-4 differs from Best First in that the frequency threshold (see section 4.2.1) is not applied, i.e. all transfer mappings are retained. Comparison SystemNum. sentences Best-First rated betterNum. sentences comparison system rated betterNum. sentences neither rated betterNet percentage improvement Babelfish93 (46.5%)73 (36.5%)34 (17%)10.0%Bottom-Up224 (44.8%)111 (22.2%)165 (33%)22.6%No-Context187 (37.4%)69 (13.8%)244 (48.8%)23.6%No-Threshold112 (22.4%)122 (24.4%)266 (53.2%)-2.0%Table-2: Translation Quality Discussion The results of the four experiments are presented in Table-2. Experiment-1 establishes that the algorithm presented in this paper automatically acquires translation knowledge of sufficient quantity and quality as to enable translations that exceed the quality of a highly rated traditional MT system. Note however that Babelfish/Systran was not customized to this domain. Experiment-2 shows that Best-First produces transfer mappings resulting in significantly better translations than Bottom-Up. Using Best-First produced better translations for a net of 22.6% of the sentences. Experiment-3 shows that retaining sufficient context in transfer mappings is crucial to translation quality, producing better translations for a net of 23.6% of the sentences. Experiment-4 shows that the frequency threshold hurts translation quality slightly (a net loss of 2%), but as Table-3 shows it results in a much smaller (approx. 6 times) and faster (approx 45 times) runtime system. Num mappingsTranslation speed (500 sentences)Best-First58,314173s (0.34s/sent)No-Threshold359,5288059s (17s/sent)Table-3: Translation Speed (500 sentences) Transfer mapping coverage Using end-to-end translation quality as a metric for alignment leaves open the question of how much of the translation quality derives from alignment versus other sources of translation knowledge in our system, such as the bilingual dictionary, or the 2 hand-coded transfer rules in our system. To address this issue we measured the contribution of each using a 3264-sentence test set. Table-4 presents the results. The first column indicates the total number of words in each category. The next four columns indicate the percentage translated using each knowledge source, and the percentage not translated respectively. As the table shows, the vast majority of content words get translated using transfer-mappings obtained via alignment. Our alignment algorithm does not explicitly attempt to learn transfer mappings for pronouns, but pronouns are sometimes included in transfer mappings when they form part of the context that is included with each mapping (see section 4.1). The 31.89% of pronoun translations that the table indicates as coming from alignment fall into this category. Our algorithm does try to learn transfer mappings for prepositions and conjunctions, which are represented in the Logical Form as labels on arcs (see Figure-1). Mappings for prepositions and conjunctions always include the nodes on both ends of this arc. These mappings may translate a preposition in the source language to a preposition in the target language, or to an entirely different relationship, such as direct object, indirect object, modifier etc. As the table shows, the system is currently less successful at learning transfer mappings for prepositions and conjunctions than it is for content words. As a temporary measure we have 2 hand-coded transfer rules that apply to prepositions, which account for 8.4% of such transfers. We intend for these to eventually be replaced by mappings learned from the data. Number of instancesAlignmentDictionaryRulesNot translatedContent words21,24593.50%4.10%0%2.4%Pronouns2,15831.89%68.20%0%0%Prepositions/Conjunctions6,64032.00%59.70%8.4%0%Table-4: Coverage of transfer mappings, dictionary & rules Conclusions and Future Work We proposed an algorithm for automatically acquiring high-quality transfer mappings from sentence-aligned bilingual corpora using an alignment grammar and a best-first strategy. We reported the results of applying the algorithm to a substantially larger training corpus than that used in previously reported work on learning transfer mappings from parsed data. We showed that this approach produces transfer mappings that result in translation quality comparable to a commercial MT system for this domain. We also showed that a best-first, alignment-grammar based approach produced better results than a bottom-up approach, and that retaining context in the acquired transfer mappings is essential to translation quality. We currently rely on a priori linguistic heuristics to try to provide the right context for each transfer mapping. In future work, we plan to use machine-learning techniques to determine the extent of the context that optimally disambiguates between conflicting mappings. References Peter Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and Robert L. Mercer, 1993. The mathematics of statistical machine translation Computational Linguistics, 19:263-312 Cambridge University Press (1995), McCarthy, M. ed., Cambridge Word Selector Stanley F. Chen, 1993. Aligning sentences in bilingual corpora using lexical information Proceedings of ACL 1993 Karen Jensen, 1993. PEGASUS: Deriving argument structures after syntax. In Natural Language Processing: The PLNLP Approach. Kluwer Academic Publishers, Boston, MA. Hiroyuki Kaji, Yuuko Kida, and Yasutsugu Morimoto, 1992. Learning Translation Templates from Bilingual Text Proceedings of COLING 1992 Langenscheidt Publishers 1997, The Langenscheidt Pocket Spanish Dictionary Adam Meyers, Roman Yangarber, Ralph Grishman, Catherine Macleod, and Antonio Moreno-Sandoval, 1998a. Deriving transfer rules from dominance-preserving alignments, Proceedings of COLING 1998 Adam Meyers, Michiko Kosaka and Ralph Grishman, 1998b. A multilingual procedure for dictionary-based sentence alignment Proceedings of AMTA 98 Adam Meyers, Michiko Kosaka and Ralph Grishman, 2000. Chart-based transfer rule application in machine translation Proceedings of COLING 2000 Robert C. Moore 2001, Towards a Simple and Accurate Statistical Approach to Learning Translation Relationships among Words Proceedings of the Workshop on Data-Driven Machine Translation, ACL 2001 Joseph Pentheroudakis and Lucretia Vanderwende 1993, Automatically identifying morphological relations in machine-readable dictionaries Ninth Annual conference of the University of Waterloo Center for the new OED and Text Research Stephen D. Richardson, William Dolan, Monica Corston-Oliver, and Arul Menezes 2001, Overcoming the customization bottleneck using example-based MT, Workshop on Data-Driven Machine Translation, ACL 2001 SoftArt Inc (1995) Soft-Art translation dictionary. Version 7 Hideo Watanabe, Sado Kurohashi, and Eiji Aramaki, 2000. Finding Structural Correspondences from Bilingual Parsed Corpus for Corpus-based Translation Proceedings of COLING 2000     789NOPQR[ /isBW###2%H%%&m&n&&&&&ոբՏՏH*mH nH ujUmH nH u*j:> CJOJQJUVaJmH nH ujUmH nH u5\mH nH u6]mH nH u mH nH umH nH tH ujU0JjU jUmH sH 8fgPQR[u V/ $a$%h`h ]^  (#]kkk/is&B !!_"V#W##2%%&'( & F1$ & F% & Fh`h hL^h`L&&&E'F'T'U't'u'{'|'''''''9(:(`(a(((((((=)>)@)*+L+X+\+r++,@,?-@----.%.).8.<.A.E.I.N.S... /-/B/K/O/X/p/w/////q0|0000081K1^1d1t1x11111112\25\6]6]mH nH uH*mH nH u mH nH uV(?)@)))@---.00112]2z2236!69;:O:^;v;; 2^2` LC$^`L L^`L% & F\2]2^2v2w2x2y2z223!66878M8N8Q8R89O:^;;==%AAAXGdGIKKmLLM[OOOZQ[QQQQQQQ*R+RIRJRURVV]^^%_a_b_~_ackclcc"dXd6]mH nH sH tH CJOJQJaJmH nH tH unH tH jhUmH nH u*jq|> CJOJQJUVaJmH nH u mH nH ujUmH nH umH nH tH u?;==%AAAEEEEEFF Fttti$$Ifl0Q <q04 la $$$$Ifa$ $$$$Ifa$1$% F%F/F0FKFPFQFoFwFxFFFFFF}}}}i$$Ifl0Q <q04 la $$$$Ifa$ $$$$Ifa$FFFG GG%G&GXGdGI#KKK}t}xvtrrv%$x $$$$Ifa$ $$$$Ifa$i$$Ifl0Q <q04 la KmLLLMPO[OOOPPPPQ>QZQ $$$$Ifa$%ZQ[QeQpQ{QQQQYPDDDDY $$$$Ifa$ $$$If$$IfTlrGqE\    064 laQQQQQQQQD$$IfTlrGqE\    064 la $$$$Ifa$ $$$IfQQQQQQR RMD $$$If$$IfTlrGqE\    064 la $$$$Ifa$ RR$R*R+RHRIRJRM@@> $ (#1$$$IfTlrGqE\    064 la $$$$Ifa$JRURRSTKU#V$V%V2VTVUV`Vn~$$IflF <v/06    4 la $$$Ifh`h `VgVyVzVVVVVVVTYY([xxtrpjjh`hx~$$IflF <v/06    4 la $$$If ([\]^^_^`^t^~^^^^ $$$$Ifa$ $$$If%h`h ^^^^^^F=111 $$$$Ifa$ $$$If$$IfTlֈiT5 d!////064 la^^^^^^:1 $$$If$$IfTlֈiT5 d!////064 la $$$$Ifa$^^^^^^:$$IfTlֈiT5 d!////064 la $$$$Ifa$^____!_$_ $$$$Ifa$ $$$If$_%_`_a_b_F9,(x $ (#1$ $ (#1$$$IfTlֈiT5 d!////064 lab_~_1``yaQbaclc#dqddef^fgg?hiijjkkkkkkk1$1$&(h`hXdpddd1e`eef2f]fgggg$h?hhhiiijjjjkkkkkkkk mHnHtH nH tH 66] kkkkkk&/0&P P. A!"#$b%2 00&P P. A!"#J$b% P h* 00&P P. A!"#J$b%2 00&P P. A!"#J$b% P h* 00&P P. A!"#J$b%2 00&P P. A!"#J$b% P h* 00&P P. A!"#J$b%2 00&P P. A!"#J$b% P hDyK arulm@microsoft.comyK 6mailto:arulm@microsoft.comDyK steveri@microsoft.comyK :mailto:steveri@microsoft.comfDdd)eOP  S .A? "2Ufj'qZ{猛%1f`!)fj'qZ{猛%hAЪf5ex] O?ϙo'#%)T^*+Ϙ0+)yEBɣ7.%W+zQ*tL 9w~Y{~N>w,l)G\pr*]|~f.@d c劮dupꨫmq9?\y'g>ͰLPUv_t59i k\tu9N S_ե~*_pٽ{7{ҙ3sէufߴKp-sԝ0ݫw3}#rz3{t8ԧ6gJꪢq>VWuu *Tn#`/=߽J@\WvwWw&w{{z׽kΘk?ճw}9.P'vd)uf?O.H;.|렽=u}Lu_SʞBw!VgCoqv}S:)+] oK;W\]Puu'Eȿ9NwMc-E~Gߒ#wzSxy3RۇwE)1Fb"wW M\0)py.;#:܍]ʽ{To\7\9l{ ++sTU>>ؕ_ M2 =ۮknWCA`eW=Me3#v%JAWbJbWS5]]]@ l 6W8\ΎKd3Ր^bmO>؞,fk>W~kW d[9ZY]w*[/xc9x{ J|A}AQQl1(-{"p-%B7-zEE[;DXLwz[xjghgĶJnx?(8G;C2½= K/$m4Wʺz2?1^IՇݬn]}(ln(K~CCƎC;Eڇ=? ԏ6\i^s 8ooǹ9Vx߫DKپQbh_w7~و Pu[P7\琈m/q+`Nr?|5I𣻃(pp wUbd8HoV'N,y_"= *K]71,_{v[?Cas*ۣ~~/X#%TW u}kݦ^~c^d4ë=IoqIʊSP;K^5K5?[9,w@I|0п*Ս=aph˩@Ͻb/E [~5*ukrnֺ_姨[~1ukQnu xWec*n)VIWV@׭Gp]DuK[{ܮ.̑˯nsxn戁=߃isԺ*|Z7/Tqw~2L/+;bޢW>\+UyeR._sH%9?_NR6/oXp-_TeqzԶRY]Vʤ0[Jb`AlBmNwLk^m}7B%gzpj.C)0]|V̉wQ?pm\YyXH^1:G@ux&Wǻnu=f%?7R9Z تzi Ru79vAho }e $\}{ٽ_?йc1whuW&*t "cUWK]Y?{bP?%/_)HQz:4n &AauN9Jk:z> d^W׆8.p0|UϠ6(vFy早ynøpMp͕|ڿ)l>\Q | _.:smzĻՍ#~.,Xڡv(p 3!nb V?3HQ?Z-FZkW E Jڤ&EPIW.W؏L˧Em(7[{*>e01X%UՔ6k(CƟ-m@ɿ2Q0z} Ma !=-\ Wd9Joo二Pɍ>kZXɭQIɗ21d|w%_JC=OJ}<5"Oc)NP<_G?XO3~(V?`<+ +k7mo!++2C}xs#rv JyO<˄.ہ=?4BXM#$v? Ϗ Z(xq_K>AE~ggHώ#|HQCڿXk[I[~.3ݥs#rƃ<EIҶ  !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXZ[\]^_`cQijklmnopqrstuvwxyz{|}~Root Entry F`e Data Y}WordDocumentObjectPoolpK``_1051605725 FpK`pK`Ole PRINThpCompObjq !"#$%&'(*+,-./01235 FMicrosoft Visio DrawingVISIO 6.0 ShapesVisio.Drawing.69qOh+'0@HXdp|arulm      !"#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~I% &" WMFC = 2OI% EMF=w lV`)2VISIODrawing% % Rp Arial dhEwpwx\d4`d ddԵ4`ddv% % (   ?& %  W$+7b11\% ( % & %  W$71% ( % & %  W$Iz60O% ( % & %  W$*z0% ( % & %  W$&{g,a% ( % RpB B  ArialEw"wwĆ ĆĆEwpwxxje<kdEww=xĆĆ4#` jxdv% % x %  % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Tl2|@ϝ@nLXhacer%%"% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % x %   TlQ@ϝ@CLXusted%"%% % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Tq@ϝ@ Ldinformacin%%9%"%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % x %   T2@ϝ@$ Ldhipervnculo%%%!%"%% % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   T92@ϝ@$ Ldhipervnculo%%%!%"%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   TdAQ@ϝ@CLTclic"" % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Tq @ϝ@ L`direccin%%""%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   Tlf@ϝ@LXclick """ % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   T& @ϝ@LxHyperlink_Information0!%%%"%%%9%%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   T`lf@ϝ@LTyou!%% % % Rp ArialdEw"wwdĆd ĆĆ&" WMFC =Ewpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Tx @ϝ@L\addressx%%%%"" % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   TM@ϝ@ L`hyperlink%!%%%" % & %  W$:F@@% ( %  '% % V0#nnn##% % (  Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   TX-dw@ϝ@iLPde%% % & %  W$% ( %  '% % V0Y#Y#Y#Y% % (  Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   TXa-w@ϝ@aiLPde%% % & %  W$<z6gBa0% ( % Rp ArialdEw"wwkdĆkd ĆĆEwpwxxqe<`dEww=xkdĆkdĆd4#` jxdv% % x %  % %  '% % V0\\\\% % (  ( Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % x %   TX@ϝ@LPen%% % %  '% % V0""""% % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   TXQ@ϝ@QLPen%% % %  '% % V00{{{00{% % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Tl-`w@ϝ@iLXundere%%%% % %  '% % V0*; *  ;*;*% % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   TdL9@ϝ@LuLTDsub0"%% % %  '% % V0V;7V77;V;V% % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Td9 @ϝ@uLTDobj0%% % & %  W$0C6=% ( %  '% % V0@@@@% % (  Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   T`I@ϝ@ILTMod7%% % %  '% % V0% % (  Rp&" WMFC = ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Td~@ϝ@LTDsub0"%% % & %  W$+1% ( % %  '% % V00{{{00{% % (  Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   Td-w@ϝ@iLTDobj0%% % &%  66!6$'6+.61467:6=@6CF6F6H$K+6N3Q;6TCWK6YS\Z6^baj6crez6hj6l6ln6pr6tv6xz6z6{}6~66 66"*62:6A6BJ6RZ6bj6rz6}666666666 66"*62:6BJ6RZ6bj6~p6~r|z6{y6wv6ts6r6qo6mk6ig6ec6b6a^6\Y 6WT6R O(6N,6M0J76H?EF6BN?V6=^:e68k67m4t61|.6+(6%"6666 66% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % &%  #6+36;C6K!S6#[%c6'i6'k)s6+{-6/062466869:6<=6?@6BC6D6EF6G H6IJ#6K+L36M;NC6OKPS6Q[Qc6RkSs6T{T6U6UU6VV6VW6WW6XX6XX6XX6XX6X X6XX#6W+W36V;VC6VKUS6U[Tc6Ti6TkSs6S{R6QQ6PO6ON6N6ML6KJ6IH6GF6ED6B A6?>#6=+;36:;8C67K5S64[2c60k/s6-{+6*(6'%6#!66666 6  6"6*26:% ( % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  % &%  66666}z6vs6ok6he6b#_*6\1Y86V?SF6QJ6PMMT6J\Gd6DlAt6>|<69765260-6-6+)6'%6#!6666 6$6,46<D6LT6&" WMFC =X6\d6l t6 | 6  6  6  6  6 6  6  666 6$6,4666<D6LT6\d6l t6"|$6&(6*,6-6.163567:6<?6AD6GI6L O6Q6RU"6X)[16^8a@6dGgO6hR6jVm]6pdtk6wrzy6~6666666% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % &%  H6PX6`h6px66666666666 (6086@H6PX6`h6px6}666666666666 (6086@H6PX6]6`h6px666666% ( % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  % &%  g6ow66666666666 6$6,46;6<D6LT6\d6lt6|6666666666 6$6,46<D6J6LT6\d6lt6|66666666666  6$+6296@ G6$N'U6*\.c61j5q69x<6@D6GK6OS6W[6`d6hl6m6qu6z~66 6 66#6*067=6DK6QX6^e6kq6w~6% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % & %  6g67o7w68869:6:6::6;;6;<6<<6;;6;:6::698687'66/5764?3G62O0W6/_-g6,o*w6(&6%6$"6 6666 6 66666$+62+6296@G6NT6X6[a6hn6u{6|6666&" WMFC =6~y6s% ( % Rp  ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  % &%  H6~P{X6y`vg6snqv6n}k6j6hf6ca6^\6YW6V6TR6PN6KI6GE6C A6?=#6;+9367;9367;5C64K2S61[/c6.k,s6+{*6)(6'&6%6%$6##6"!6  6  66 6#6 + 36 ; C6!K"S6#[$c6%k&s6'{)6*,6-,6-/613657686:<6?A6DG6JM6PS6VY6]#a*6e1i86k;6m?rE6vL{S6Z`6gm6tz6666666666666#6#*6186?F6M F6M T6[b6ip6w!~$6'*6-06366668;6>@6CF6H6IK6NP6SU6XZ6\&^6.`6b6>dFg6Kh6NiVj6^lfn6npvq6~su6u6vx6yz6|}666666 66&6.66>F6NF6NV6^f6nv6~6666666666&6.66>F6NV6^f6nv6~6666666666&6.66>F6NV6^f6nv6~66666666% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % % ( RpTimes New RomanEw"wwĆ ĆĆEwpwxx5< Eww=xĆĆ4#` jxdv% % x %   TGR@ϝ@G"LFigure 1a: Lexical correspondences712!,2,;,3,,,2!!,'2222,2,,' % % ( RpTimes New RomanEw"wwH|ĆH| ĆĆEwpwxx5< Eww=xH|ĆH|ĆH|4#` jxdv% % ( x %   T1 R8@ϝ@1 LFigure 1b: Alignment Mappings712!,22H12N,2Y,2221' % & %  W$  b  \% ( % & %  W$    % ( % & %  W$ z  % ( % & %  W$ z  % ( % & %  W$ {ig ca% ( %&" WMFC }= RpB B  ArialEw"wwĆ ĆĆEwpwxxje<kdEww=xĆĆ4#` jxdv% % ( x %  % % ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Tl 2g |@ϝ@ nLXhacer%%"% % % ( Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   Tl 9 Q@ϝ@ CLXusted%"%% % % ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Ttq @ϝ@t Ldinformacin%%9%"%% % % ( Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % x %   Tn 2@ϝ@n$ Ldhipervnculo%%%!%"%% % % ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   T 2@ϝ@ $ Ldhipervnculo%%%!%"%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   Tdy  Q@ϝ@y CLTclic"" % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   T q@ϝ@  L`direccin%%""%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   Tl H @ϝ@ LXclick""" % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   T~ @ϝ@~LxHyperlink_Information0!%%%"%%%9%%% % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   T` lG @ϝ@ LTyou!%% % % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Tx  @ϝ@ L\address%%%%"" % % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   T M@ϝ@  L`hyperlink%!%%%" % & %  W$ ( " &" WMFC ]=" % ( %  '% % V0#O O O ##% % (  Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   TX-F w@ϝ@iLPde%% % & %  W$bnhh% ( %  '% % V0;#;#;#;% % (  Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   TXC-w@ϝ@CiLPde%% % & %  W$ z g$ a % ( % Rp ArialdEw"wwkdĆkd ĆĆEwpwxxqe<H|Eww=xkdĆkdĆd4#` jxdv% % x %  % %  ' % % V0= =   = = % % (  ( Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % x %   TXm @ϝ@m LPen%% % %  ' % % V0      % % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   TX3 | @ϝ@3 LPen%% % %  ' % % V0w 0d {w {d {d 0w 0w {% % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Tl -B w@ϝ@ iLXunder%%%% % %  ' % % V0 ;    ; ; % % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %   Td. 9 @ϝ@. uLTDsub0"%% % %  ' % % V0 ;    ; ; % % (  ( Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Td 9 @ϝ@ uLTDobj0%% % & %  W$e0qCk6k=% ( %  '% % V0""""% % (  Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %   T`*@ϝ@*LTMod7%% % %  '% % V0       % % (  Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % ( x %   Td ` @ϝ@ LTDsub0"%% % & %  W$ q k% ( % %  '% % V0 0 { { { 0 0 {% % (  Rp ArialEw"wwĆ ĆĆEwpwx&" WMFC ==x< Eww=xĆĆ4#` jxdv% % ( x %   Td - w@ϝ@ iLTDobj0%% % &%  6666 6 6 6 66!)6,61 96"A$I6&Q(Y6*a,i6-l6.q/y6126456786969;6<=6>?6AB6C6CD6EE 6FG6G!H)6I1I96JAJI6JQKY6KaKi6LqLy6LL6LM6MM6MM6ML6LL6KK6KJ6J6JI 6HH6G!F)6E1E96DACI6BQAY6@a?i6>q=y6<;698676643621606/-6,*6('6%# 6" 6!)6196AI6QY6[6 a i6 qy66666666% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % & %  #6+36;C6KS6[c6i6k s6 {666666!"6$%6&6'(6) *6+,#6-+.36/;0C61K2S62[3c64k4s65{6666676776886996::6::6::69969 9699#68+8368;8C67K7S67[6c66i66k5s64{463261160/6/6.-6,+6*)6('6&%6$ "6!#6+36;C6KS6[c6ks6{ 6  66666666 6#6+36;% ( % Rp  ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  % &%  66~|6zx6wu6sq6q6pn 6mk6j$h,6g4e<6dDbL6bM6aT`\6_d^l6]t\|6[Y6XW6W6VU6UT6SR6QP6ON6N6NM6LL 6KK6J$I,6I4H<6H@6HDHL6GTG\6GdFl6FtF|6EE6E6EE6EE6EE6EE6EE6E6EE6FF6FG 6GG6G$H,6H26H4I<6IDJL6KTK\6LdLl6MtM|6N6NO6PQ6RS6TT6UV6W6WX6YZ6[]6^_ 6`a6b$a6b$c,6e4f<6hDiL6kTl\6ndol6qts|6t&" WMFC =v6xz6{}66% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % &%  H6PX6`g6ow6x666666666666 66'6/76?G6OW6_g6ow6666666666666'6/76?G6OW6_g6ow666666% ( % Rp ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  % &%   g6 o w6 ~ 6  6  6  6  6  6  6  6  6 6  6  $6 , 46 < D6 L T6 \ d6 l t6 | 6  6  6 6  6  6  6  6  6  6  6 6  $6 , 46 < D6 L T6 \ d6 l t6 | 6 6  6  6  6  6  6  6  6 6  6  6  6 " )6 ,6 0 76 > E6 L S6 Z a6 d6 h o6 v }6  6! % 6' 6) - 61 6 6: > 6B G 6J 6K P 6T Y 6] b 6f k 6o t 6y } 6 & ,6 3 96 ? F6 H6 L R6 X ^6 d j6 p w6 }% ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % &%   g6 o w6  6  6 6  6  6  6 6  6  6  6 6  6  6  '6 / 76 ? G6 O W6 _ g6 o w6 ~6  6  6  6 6  6  6  6  6  6 6  6  6  #6 * 16 8 ?6 F M6 T [6 b i6 l6 p v6 } 6 { 6v q 6l g 6b % ( % Rp ArialdEw"ww`dĆ`d ĆĆEwpwxxye<H|Eww=x`dĆ`dĆ4#` jxdv% % ( x %  % &%  _ Z6c ah g6l np u6u {y 6~  6  6  6  6  6  6  6  6  6  6  6 " )6 0 76 > E6 F6 L S6 Z a6 h o6 &WMFC =v }6 ~6  6  6  6  6" % 6( + 6. 1 63 64 6 69 ; 6> @ 6C E !6G )I 16K 9M A6O IQ Q6S YU a6V iX q6Z y[ 6\ ^ 6_ ` 6a b 6c c 6d d 6e e 6e e 6e e 6e d 6d c !6b )a 16a 56` 9_ A6^ I\ Q6[ YZ a6X iV q6T yR 6P 6O M 6J G 6E B 6? < 69 5 62 / 6+ ' 6#  6 6  6  6  6 " (6 . 46 : @6 F L6 R X6 ^ X6 ^ c6 i n6 t y6 }6 ~ 6  6  6 6 { 6t m 6g ` 6[ % ( % Rp ArialdEw"wwdĆd ĆĆEwpwxx< Eww=xdĆdĆd4#` jxdv% % x %  % & %  #6+36; C6 KS6[c6ks6{6666666666 6"6)168@6GO6V]6el6s{6666666666}6zw6sp 6li6e!a(6^/Z66V=SD6OKKR6GYC`6?g;n68r66u2|6-)6% 666 66666666 66"6%+6066;A6F6GL6~QxV6r[l`6fe`j6YoRs6LxE}6>760)6"6 666666666666{6sk6c[6Z6SK6C;63+6(6#6 6666666666{6sk6c[6T6SK6C;63+6#6 6666666666{6sk6g6c\6TM6E>66/6(6' 66 66666~6{x6w6ur6ok6he6xbq^6j[cX6\TUQ6NNGJ6@G9C62@+<6$9% ( % Rp  ArialEw"wwĆ ĆĆEwpwxx< Eww=xĆĆ4#` jxdv% % ( x %  %   6-- Arial-"System- -%11\-- -%1-- -%0O-- -%0-- -%,a--B B  Arial-"System-.- -- Arial--.- 2 nhacers%%"%-- Arial--.- 2 Custeds%"%%-- Arial--.- 2  informacin%%9%"%%-- Arial--.- 2 $ hipervnculo%%%!%"%%-- Arial- -.-  2 $ hipervnculo%%%!%"%%-- Arial--.-  2 Cclic""-- Arial--.- 2  direccinu%%""%%-- Arial--.- 2 clicks"""-- Arial--.- '2 Hyperlink_Information0!%%%"%%%9%%%-- Arial-- .-  2 you!%%-- Arial- -.-  2 address%%%%""-- Arial--.- 2  hyperlinku%!%%%"- -%@@----$nn##-- Arial--.-  2 ide%%- -%----$Y#Y#Y-- Arial--.-  2 iade%%- -%Ba0-- Arial--.- --- -$\\\--  Arial--.-  2 en%%--- -$"""--   Arial- -.-   2 Qen%%--- -${{00{--  Arial--.- 2 iunders%%%%--- -$*  ;*;*--  Arial--.-  2 uLDsub0"%%--- -$V77;V;V--  Arial--.-  2 uDobj0%%- -%6=----$@@@-- Arial--.-  2 IMod7%%----$-- Arial-- .-  2 Dsub0"%%- - %1- --- -${{00{--  Arial- -.-   2 iDobj0%%--!$'+.147:=@CFF$H+K3N;QCTKWSYZ\b^jarczehjllnprtvxzz{}~ "*2:ABJRZbjrz} "*2:BJRZbjp~r~z|{ywvtsrqomkigecba^\ YWT R(O,N0M7J?HFENBV?^=e:k8m7t4|1.+(%" -- Arial--.- -- #+3;CKS![#c%i'k's){+-/024689:<=?@BCDEF GHI#J+K3L;MCNKOSP[QcQkRsS{TTUUUVVVWWWXXXXXXXX XXX#X+W3W;VCVKVSU[UcTiTkTsS{SRQQPOONNMLKJIHGFED BA?#>+=3;;:C8K7S5[4c2k0s/{-+*('%#!   "*2:- - Arial- -.-  --}zvsokhe#b*_1\8Y?VFSJQMPTM\JdGlDtA|><97520--+)'%#! $,4<DLTX\dlt |                $,46<DLT\dlt |"$&(*,-.1357:<?ADGI LOQR"U)X1[8^@aGdOgRhVj]mdpktrwyz~-- Arial--.- -- HPX`hpx (08@HPX`hpx} (08@HPX]`hpx- - Arial- -.-  --gow $,4;<DLT\dlt| $,4<DJLT\dlt|  $+29@G N$U'\*c.j1q5x9<@DGKOSW[`dhlmquz~  #*07=DKQX^ekqw~-- Arial--.- -- g6o7w7889::::;;;<<<;;;:::988'7/675?4G3O2W0_/g-o,w*(&%$"   $+2+29@GNTX[ahnu{|~ys- - Arial- -.-  --HP~X{`ygvnsvq}nkjhfca^\YWVTRPNKIGE CA?#=+;39;739;7C5K4S2[1c/k.s,{+*)('&%%$##"!     #+ 3 ; C K!S"[#c$k%s&{')*,-,-/13578:<?ADGJMPSVY#]*a1e8i;k?mErLvS{Z`gmtz##*18?F MF MT[bip!w$~'*-03668;>@CFHIKNPSUXZ\^&`.b6d>gFhKiNjVl^nfpnqvs~uuvxyz|} &.6>FNFNV^fnv~&.6>FNV^fnv~&.6>FNV^fnv~-- Arial--.- --Times New Roman--.- :2 G"Figure 1a: Lexical correspondences712!,2,;,3,,,2!!,'2222,2,,'-- Times New Roman- -.-  32 1 Figure 1b: Alignment Mappings712!,22H12N,2Y,2221'- -%  \-- -%  -- -% -- -% -- -% ca--B B  Arial-- .- -- Arial-- .- 2 n hacers%%"%-- Arial-- .- 2 C usteds%"%%-- Arial--.- 2 t informacin%%9%"%%-- Arial--.- 2 $n hipervnculo%%%!%"%%--  Arial- -.-  2 $ hipervnculo%%%!%"%%-- Arial--.-  2 Cy clic""-- Arial--.- 2  direccinu%%""%%-- Arial--.- 2  clicks"""-- Arial--.- '2 ~Hyperlink_Information0!%%%"%%%9%%%-- Arial-- .-  2  you!%%-- Arial- -.-  2  address%%%%""-- Arial--.- 2  hyperlinku%!%%%"- -%" " ----$O O ##-- Arial--.-  2 ide%%- -%hh----$;#;#;-- Arial--.-  2 iCde%%- -%$ a -- Arial--.- --- -$=   = = --  Arial--.-  2 m en%%--- -$     --   Arial- -.-   2 3 en%%--- -$w {d {d 0w 0w {--  Arial--.- 2 i unders%%%%--- -$   ; ; --  Arial--.-  2 u. Dsub0"%%--- -$   ; ; --  Arial--.-  2 u Dobj0%%- -%k6k=----$"""-- Arial--.-  2 *Mod7%%----$     -- Arial-- .-  2  Dsub0"%%- - % k- --- -$ { { 0 0 {--  Arial- -.-   2 i Dobj0%%--    !),19 A"I$Q&Y(a*i,l-q.y/12457899;<=>?ABCCDE EFG!G)H1I9IAJIJQJYKaKiKqLyLLLLMMMMMMLLLKKKJJJ IHH!G)F1E9EADICQBYAa@i?q>y=<;987643210/-,*('% #" !)19AIQY[a i q y-- Arial--.- -- #+3;CKS[ciks { !"$%&'( )*+#,+-3.;/C0K1S2[2c3k4s4{56667778899::::::99 999#9+838;8C8K7S7[7c6i6k6s5{4432110//.-,+*)('&% $"!#+3;CKS[cks{    #+3;- - Arial- -.-  --~|zxwusqqp nmk$j,h4g<eDdLbMbTa\`d_l^t]|\[YXWWVUUTSRQPONNNML LKK$J,I4I<H@HDHLHTG\GdGlFtF|FEEEEEEEEEEEEEEEEFFF GGG$G,H2H4H<IDILJTK\KdLlLtM|MNNOPQRSTTUVWWXYZ[]^ _`a$ba$b,c4e<fDhLiTk\ldnlotq|stvxz{}-- Arial--.- -- HPX`gowx '/7?GOW_gow'/7?GOW_gow- - Arial- -.-  --g o w ~                       $ , 4 < D L T \ d l t |                        $ , 4 < D L T \ d l t |                         " ) , 0 7 > E L S Z a d h o v }   ! % ' ) - 1 6 : > B G J K P T Y ] b f k  o t y  } & , 3 9 ? F H L R X ^ d j p w } -- Arial--.- -- g o w                         ' / 7 ? G O W _ g o w ~                         # * 1 8 ? F M T [ b i l p v }   { v q l g b - - Arial- -.-  --Z_ ac gh nl up {u y ~                       " ) 0 7 > E F L S Z a h o v } ~         " % ( + . 1 3 4 6 9 ;  > @ C !E )G 1I 9K AM IO QQ YS aU iV qX yZ [ \ ^ _ ` a b c c d d e e e e e e  e d d !c )b 1a 5a 9` A_ I^ Q\ Y[ aZ iX qV yT R P O M J G E B ? < 9 5 2 / + ' #         " ( . 4 : @ F L R X ^ X ^ c i n t y } ~        { t m g ` [ -- Arial--.- -- #+3;C K S[cks{ ")18@GOV]els{}zws pli!e(a/^6Z=VDSKORKYG`Cg?n;r8u6|2-)%   "%+06;AFGLQ~Vx[r`lefj`oYsRxL}E>70)" {skc[ZSKC;3+(# {skc[TSKC;3+# {skgc\TME>6/('  ~{xwurokhebx^q[jXcT\QUNNJGG@C9@2<+9$- - Arial- -.-  -ObjInfo VisioDocumentSVisioInformation" SummaryInformation( Visio (TM) Drawing H %Rl !fffMMM333$ $ Ud8@ T Arial@NWingdzs@N@tMonotype Sort*NTimes New RomaUn@+ NtSymbol5T?? Y@-1TJDT1EWP-hTT<U*U U =/Ub b酸0zGz?@8H2!kWbU%U  +PL/^&9^$? { Ak^&,",'%/v&Q&  1y   )? 2    J12?k9aUBBHEHEHEUHEHEHEH@?>?:`2BBHEHEHEHEHEHEHEH@$%O9F7AOY@; A*sVsVAgLTkY 1 W_W__ !`#ku4lb6Pu`kW 4l 4l %Y?:?-\ *#!+|tKf*2|2|2|2wG)QUoTMeEttA%_8BOTOfOxOO??O?7 ܻuW+? sU42 T*PbtφϘϪ(:L^p߂߂O߳ 1C@Ugb{.);M_xa"4FX 0Bzas//'/9/K/]/o/////////?#?zaB?T?f?x????????OBOUbb7OQV.e_OqOOOOOOOO_wk#_5_G_Y_k_}______L__UlAoSoeowoooooooxa(:L^p cwo!3EWi{Ï٦ .@Rdvk}*B'Ck#bz@.`c.`/%b /=QN3L/^/p////////??7?Pbtί(:L^p݇Z;ɼٿِ>.@*[m߅3O'Nk P2Ǎq?߃?\.?o AqT$$$'9K]ڊ䐾伩jPPj|xԕIWؙ&V03E)0<kDB!ߵ V /ASew+=Oa/09K]o?!?%`Apple LaserWriter 16/600 PSS odXX;LetO_cPRIV0|'ro\?KhCO;afup//$/6/H/Z/l/~////////? ?2?D:C-winspoolApple LaserWriter 16/600 PSLPT1:UFDf h-TUU[U@@??I?`d buoqYkQhu23u` Connector ` e1Crw UH ^   -}|5 p`I`V?}`xa833ސ]3σ3u 33?, ?Gxpx^& CThis connector automatically routes between the shapi ts.HD @# =h8T YY9 BT#F oU@? P6 u `u bA@]u  .(#DB uu`h?\hr|uVa@-?bl;'bE-ho'$y( 2rq?@I ?$%? @"U*5L -br  ^vv"(2uI."q28v"uh9Bd&</MSz #145 |`Vis_SE.chm!#20)@`7Copyright 1999 @io Corporation. All ;Bs reserved.4U#1#7)A,@A+4j'vQ8l>$UdvE \b4 *1B$b24R(@][v[D ZQi@ a59 93O'2"q?.g;2GHl'w-* O`E )FT 9#  7 B ? >Gdo@4k8' (Ul4,ǟٸN@|p .@ P jA-$38Ul4,ǟٸN@|p .@ P N A-3X7;'Ul4,ǟٸN@|p .@ Q *A-37"A(Ul4,ǟٸN@|p .@ LR jA-D3 AU~@dB ER0B 4P ^MT[GcQ FH<( H<( H<( H<( U~EB  R0B> |P _ /T6DUmA@ ?ۿI?Y{*$O@SeeeZ?% $H. \?o  _# ?..:? ?j n`Uyek_k|f`cConnect two objs together. or can be ei  a curve a straight line. shap.f).bW"-ٿL&d2?P2k~i? FH H D # =hT#4>T #3 AAU@;?@} P!3|@u `u bu`u J- q6u` ?mu`b"Y<,O#  AO8A U[iJ , '>މU@dx-E?@d2L?@??@m6f?.@t'+"o"~//'/%/*:$! tNz)sOQ2 W< 9&70dQ46 &""2N 6 !LO-]{eE?AJ4?2rqWI9@.E7ES??rbBhC*'@`A&@bgCj OBuo QA^Su^]]}#`uAy@ b P;u`282 &"{C#]QF1F37 h`Vis_Sba.chm!#2244732`7Copyright 1999 Pio Corporation. All Rs reserved.)l>(>U i5 /7(C$9(8^C$<%E??< hEY fA|8Q92 Yn^a@u9 aZX^;c`Change Arrowhead..PVwsthe a~u sizpnptyle forpis con?nectorPbB?b)O?q":%7]U^}[cJfc SwEt {,U`@`_Set As StraRLined9P"Nv l_? to bepWP)BszF, # KB  ca @)+k=a5kNPUFDfP h> /T6DUmA@ ?ۿI?Y{O1@3EEbOeZ HG 9V?o)>  `r/`wpw> ȳpke{rpgTHYY9 mAUA? @ ?Q6 u` ?Mu ~YYA|XA44 4> !bJfvqQ >!"q06DNX lL>U5 LO@%#A?AE^?AL#l%u%?P6 @*  L#&H   " ' ]{%#W1\!\!5 `7Copyright 1999 Visio Corporation. All 2s reserved.`0_Sba.chm!#22427Bd(,*9 l>(>Ud!E (!(ZU$ B$bW* /T6DUmA@ ?ۿI?Y{\ $Oc@SeeeZ? $H S &4G ?rp{ri^ARectangle with variabcorner rounding.f s on r.mb?贁N? ?4 HD # =h(>TP9 AUAZ@ ??Q6 V  , A+L(2u` 6u>d7 ; vB@J{(!-$5 `7Copyright 1999 Visio Corporation. All R"s reservedo.`] _Sba.chm!#22433Kl> >Ud]!! %.?!<r r<%G"bb%I6B3 . -3!3! #y 2!:?AԲ?$6#$3  C'a9n2 @r"NH6 "I5 3ҡ2G2GBG22B! K(;?0<::E3H81|j!<7GB 9D#G"AEBA3 AEK!:096CHHBFh322p!:E"B]dB!J\068D cfBDFo#^o.:B\lq,^E@/3/7065"AChUTb 1}Y0s{V#rCQVS 1%{yq|\ޏ0dA!a| = o6b0\; Y01FC B*ABEV!!.!11Q!YsR5gtrw5gtrwgtdO?O1u?OO6N{a5@ST_uhacer/KeqG_uq5s52/U@VU`M$@@Y"[?@?$d,ez/!q/-]f/./+@?R?d?v???O __._?U$gQvOuGOx.eO<ccOFc5_G_usted/%m_____//,/>%X"e@@K8,h>o**qao-o.`јo+9 ??0?ooo /, PtJn-?ٿEu(3EύM[infor?macin/ 2DVhzp ˖}ݟ") #/G/k ̯ޯ`rϖ\n 꿩?^pO ΎM\ipervnculo/6%7I[mrZ!'"`֦p .fן(6ߦ-(:2LQUgyCUg-/y//#8%@? ψ?/'?/?? !____[侑.j"`,hISPj!smVo/o^/p/////O(O:OS??HWuo?od?h?o???"o?Xqo|ύpclicOOOOO 0BTfc|b__ __/$+ 3/E/W/oo'o9oKoo4E-ionQc^(AL^}direcA1CUgOO __/Z?@̄e3y?@]A ̷fF_j_ȟڟ"|ί$7IZЂߦt6_]C k '9K]oρˆ_۶m6ij?@JR@RVC{);M_q߃ߕߧ߹);TBoe>3/R/kv//}jKHyz@link_Informatio ˆ(ߊ(b 3?@Oō<?Ul/.!'/!f/)ٟwWi{1G?ا 3B&!AI a?s:T. #B1^R /,/UV!UA#YSObOǿ8JOGOOO2WbQ@Y U_g_y________ ogycooo<oor&_ n&/pobjkg# 5G!# ! WiG!LS!rq@@fŒ2Y@@x6Ov2Ȋ3CohĀR6S!<1 &&q940x2T~0"\МEP$y.d?@j"]s23CI4@k?*D2E?7*C,P#9P k" tNк!zs:!GFS! G%O贗Nk  ('?,+y tN,hxzsH1dXx l~TϿ"-'\п_qd}Ϗϡϳv/ 1CUgyߋŎ x\ߣ5GOwDOp$A!OEWi{, 炠@r@GzS9UvA cYeYɨq@@ 3&7b=7?EWi{6 qD.OGpAUgy2O ~ ( y^S՝ӶuR/d/v//`/////pAعf?!?36UD8f ggJonf'wmg1??? OOb_BOTOfOxOOO^G+ OO_ _ =_O_s_cVX@@F׋ @@F8`uwU___oo+o0=oOoaososعo =j&|30oq+5 F6pJq'¯gy X-I*^"ĸ ߃{,t?ɃMɴƏ؏I[63e @@}{?@?]?PĀ.QW iH֋3c͕Lĸr֠!3!10>`s3mq҄.RD2ٿ):aZ￵s|ŕñ3%Lñ7B|hacer'i$6HZl~ !я&۾~<5@@qYo?@?Ө*)i)as߅ڼtpU((/:/ s//i'//_usted.*{ύϟϱ 88@@(@(@@8Cٺt1iF1ّc8 ߤ߶[m??%?7?8F_/_V/X/o///Ko_@/o_ooK= informacin{2/u????????OO)I%6E޵GF޵YOkO}O KqsOOOG O__,_>_P_b_t_ooo_[{\ooПpo`!oWp{J>ipervnculoFrOasyo'SO/P/LW0OOpH_/d_O /_O__e-}-{direcBv3/////Vhz4=w@@e3?@y?]@ u)us???gbawv8?\>OPObOtOOO_ooO6"vNo_r@_+u_9P`_C`.o"{*kib4couoooooooo48iit?@W6r?Ug>O /wɼ׏)Ȱ_Fk*_rݿe-Hylink_Informatio\c5,>Pbt480 0p 3R n??wOa:O_0I=Q>MA_S_e_?(A5Q#5Pe@ > @@7od7Dd7Pfw:B!Ja:b``u bgRx`@" XR RQfT_okA(iVbRUPS @CT @|AB8bgW bWo*eVVg"mW`o*`'!3/E/W#!`E?XO$?6>I9*bfD?u0uNI$7 B;O$OyY!yROdOvOooL GHd5ad5Qρ__[qYā!M5/M2+Lhx@6Lfd07a%yq U"{;>y?\r?]};5]u45G7ɟퟥ((¯.RyYf+=OasL `X lEi!/E/i/u|svunderE>?QcuaPXb3I-?Q1Y)F!3////X6#_/Y_:x__,/_P/,__@/ot_,o/u {/Dsub2???&?8?J?\?n???SUvp ????OOavBOTOfOxOOOOOOOOTofoxooP_____=O_͏񏩟Loԏ98oo`objb!ao'ao(a"u8/w yqaaaU@0 0@@ 3/@@x&-IN(:L^p(:wF/j/l~//0//T/ ?7//T?VhhB.@BgM}lH1CA C@?@'DfEPA-K^@`u >bB}@",HBBA=T@.{1yV۹ɶBSA2bWWRW_eV-g]W7o`;|%GYk@aO,O/#+*~?=D?t-uN k31C'?K]oCOUOoL__q_DyqRgy //y????u/////?;_q???0objru|suquqzt|fDU@pgP`&(@@Z敩@hHqRoD3͏߉i fwuyc|dus'k2UŋbA̘+;Figure 1a: Lexical correspondencesq1CUgyׯM-?gcx߀į,؁qEp?@@쪴2r?@[I@?DEdFrx,&3HPb02g tNwp1zs灄%!Sp1 pa5|eO/Nkc mq(]U%?7?8^g?oo?*J^I??O'O9OKOJiOOOOOOO_SU-U?@޳6at_UMU@ U_ng:%u6>п__ॽd i,ofN tYNzs5i!Bo푯! Yoko}ooTooroIпίLdj|c 0BTfx=2/SdÏ㽮쏐Ϣ"4Ϯdv1П]HK{G2DVhz-n@@hm.@@?XhQX?@cZ [b[i5(P;n9\n?P;nAɳ $j5i+ܿ '9TTf;)ϲx^rs?т&8Jߋߝ߀,?߶߀" =[TfΊymY/k/4yO ODVhzL`Q_$6~: - [b@@Ę{`n0cć/ `Q  @@!A[q0p3A)'\Q:9?PUe5GY`Qs8w`Qv/The///`QQ/0{OO/oܿAo/ҫ{ //??T?f?I?m????????4е"eO/OAJ0n_O"4OJBnOOï __1_C_Ma_)_[______j籺?@̄en/~vQ )lo~eb}@@`7q3o5`?@#`kpoo?o"}~)siWi)J֔T1)DVP ψ"Th/ɏۏ.5GYk}>k_ AJ(^J N֯ UTUVUU !"$U%&)*U+23456*Ul4,ǟٸN@|p .@ G qHC-A BrAU2@@ Hr7PRH<( U2E$A r7 RUlL k,~E q@?%rRD ;U$ +"4F`Xj|C:\Program Files\Microsoft Visio\Solutions\ Extras\Backgrounds.vss-, #5GYk}C:\Program Files\Microsoft Visio\Solutions\Block Dia\Basic Shapes.vss2>+"4FXj| C:\Program Files\Microsoft Visio\Solutions\ Extras\Borders and Tit.vssv(BC*d >sC!! +/0- -H*= MYk PE/i#*o'2U"?%%. 1#&GuideTheDocPage-1Gesture FormatVisio 90ConnectorVisio 00Visio 01Visio 02Visio 03Visio 10Visio 11Visio 12Visio 13Visio 20Visio 21Visio 22Visio 23Visio 50Visio 51Visio 52Visio 53Visio 70Visio 80BasicBasic ShadowDynamic conn?ectorSchemeNameLine-curve connectorScaleFactorLine-curve connector.31Line-curve connector.34Line-curve connector.35Line-curve connector.32Line-curve connector.36Dynamic connector.56Dynamic connector.57Dynamic connector.64EllipseRounded rect?angleLine-curve connector.72Line-curve connector.77Line-curve connector.75Line-curve connector.70Line-curve connector.71Line-curve connector.73Line-curve connector.74Line-curve connector.76 #93~ uEv  - R-G.-  U;-,qIc UDqVc \qcc Utqpc q}c Uqc qc Uqc qc _ c %Uc 4%c L%Uc d%c |%c % wd %%&R%3%= NU$i<xU\R"UU7*7* x8)/>m* Qm A1 Um '< t*\ *|J m  m  m*̍ m* y8)  ?!0'/0C3:/0G3:$/0K3:,/0O3:4/0S3: 7DVhz .@Rdv 763* 3!/3/E/W/i/{//////// ??/?A?S?e?w??????;Ro0tE;n0ryO 5!AOZ@VB!ALOOO5.O/vPYbAThisD0cume2 __fA1 _S_fA_qD_PROJ0C@__.TA __+%!.fA(!- kAU3u/Bu/RU!"#$TQ&'(U)*+,3U./013645678/oo(:L^pX7E..q.ykT.p@*zaA"U_N`WfA%v!qx d8MQE;7/߿ A 2DV7aSL?S5t?%3Jk@wPD3D2dm6,SUyem`L@bҮA [e@vRP1K]%Z3=@BT^nzWUNT?S@s0)eP\d0Ѫtby`OEu0oP&Y3@k@  ʋ;߿68[J+T.pM`OS`1cb@1q0`U-RA%f@ //,/>/P/b/t////////??(?:?L?^?p??????!_F?_$Ij1`k`-R  d`VisioVBA"@Win16~5D32,AMac߳/A6# ACL_2001_Fig1!stdole` 26t< @_EvaluoEjӲ-<ݼ42P< @b5b_ CS(q8R0* pHd[!`bE@jB@@ = +x * J< ruBe>@rf%\*\G{020430-;@C06}#2.0x#C:\WIT\System32\c2.b#OLE ation`t.pF2G@QcPSnY2`H1 ,C" +BIoID="{A2B73420-2F4B-4CA2-AB9E-3CF23EE?2B154}@E=I/&HMb )e="bJ@HelpContexta0@Ver'@nComp`ble32="39322f@&qCMG="FDFF0E2032F636ZyaPB="CFCD3C12DC?36AF37~s@GCp1A352o9753q68@ [Host Extender Info] f1={3832D640-CF90-11CF-8E43-00A0C911005A};VBE;i  ao&roo gyӏ -?Qcuϟ);M_q˯ݯ%7I[mǿٿ!3EWi{ύϟϱ /ASew߉ߛ߭߿+=Oas'9K]oA: 4i B{ * g"4pFX(Y@(Ϗ@pLz J |O :]RLOQ  |Uݐ&UW 9!B lGZ > OXo>OU +$ b,' rN 5s)ޯtJ1< (u=`2 zwrV<v{|B?ܟ m}GD 5 C=d ~.U OY B z$q ՜.+,D՜.+,V@HP\h t   PagesMastersPage-1Dynamic connectorLine-curve connectorEllipseRounded rectangle8_VPID_ALTERNATENAMES_VPID_DocumentSummaryInformation8 _1048342922FP`P`Ole CompObjqPREVIEWS_PID_LINKBASE A  FMicrosoft Visio DrawingVISIO 6.0 ShapesVisio.Drawing.69q՜.+,D՜.+,ObjInfoVisioDocument VisioInformation"SummaryInformation(FVisio (TM) Drawing  $Q zRl !fffMMM333$ $ Ud8@ TD Arial@NWingdzs@N@tMonotype Sort*NTimes New RoWman@+ NtSymbol5T?? Y@-1TJDT1EWP-hTT<U*U U =/Ub b酸0zGz?@8H2!kWbU%U  +PL/^&9^$? { Ak^&,",'%/v&Q&  1y   )? 2    J12?k9aUBBHEHEHEUHEHEHEH@?>?:`2BBHEHEHEHEHEHEHEH@$%O9F7AOY@; A*sVsVAgLTkY 1 W_W__ !`#ku4lb6Pu`kW 4l 4l %Y?:?-\ *#!+|tKf*2|2|2|2wG)QUoTMeEttA%_8BOTOfOxOO??O?7 ܻuW+? sU42 T*PbtφϘϪ(:L^p߂߂O߳ 1C@Ugb{.);M_xa"4FX 0Bzas//'/9/K/]/o/////////?#?zaB?T?f?x????????OBOUbb7OQV.e_OqOOOOOOOO_wk#_5_G_Y_k_}______L__UlAoSoeowoooooooxa(:L^p cwo!3EWi{Ï٦ .@Rdvk}*B'Ck#bz@.`c.`/%b /=QN3L/^/p////////??7?Pbtί(:L^p݇Z;ɼٿِ>.@*[m߅3O'Nk P2Ǎq?߃?\.?o AqT$$$'9K]ڊ䐾伩jPPj|xԕIWؙ&V03E)0<kDB!ߵ V /ASew+=Oa/09K]o?!?%UFDf h-TUU[U@@??I?`d buoqYkQhu23u` Connector ` e1Crw UH ^   -}|5 p`I`V?}`xa833ސ]3σ3u 33?, ?Gxpx^& CThis connector automatically routes between the shapi ts.HD @# =h8T YY9 BT#F oU@? P6 u `u bA@]u  .(#DB uu`h?\hr|uVa@-?bl;'bE-ho'$y( 2rq?@I ?$%? @"U*5L -br  ^vv"(2uI."q28v"uh9Bd&</MSz #145 |`Vis_SE.chm!#20)@`7Copyright 1999 @io Corporation. All ;Bs reserved.4U#1#7)A,@A+4j'vQ8l>$UdvE \b4 *1B$b24R(@][v[D ZQi@ a59 93O'2"q?.g;2GHl'w-* O`E )FԸ "#t K7 B  'Gdn$o@dk E' (*#'Ul4,(\(@a,Œ?@  *A-37Ul4,(\(@a,Œ?@  : A-3D7;'Ul4,(\(@a,Œ?@  *A-37"A(Ul4,(\(@a,Œ?@ \ *A-T37"A#Ul4,(\(@a,Œ?@  5*A- 3;7"AU@ CR0B 4 JM| Gcmq l =rH<( H<( H<( H<( H<( UE  R0B>  _L _ m_  UFDfP h> /T6DUmA@ ?ۿI?Y{*$O@SeeeZ?% $H. \?o  _# ?..:? ?j n`Uyek_k|f`cConnect two objs together. or can be ei  a curve a straight line. shap.f).bW"-ٿL&d2?P2k~i? FH H D # =hT#4>T #3 AAU@;?@} P!3|@u `u bu`u J- q6u` ?mu`b"Y<,O#  AO8A U[iJ , '>މU@dx-E?@d2L?@??@m6f?.@t'+"o"~//'/%/*:$! tNz)sOQ2 W< 9&70dQ46 &""2N 6 !LO-]{eE?AJ4?2rqWI9@.E7ES??rbBhC*'@`A&@bgCj OBuo QA^Su^]]}#`uAy@ b P;u`282 &"{C#]QF1F37 h`Vis_Sba.chm!#2244732`7Copyright 1999 Pio Corporation. All Rs reserved.)l>(>U i5 /7(C$9(8^C$<%E??< hEY fA|8Q92 Yn^a@u9 aZX^;c`Change Arrowhead..PVwsthe a~u sizpnptyle forpis con?nectorPbB?b)O?q":%7]U^}[cJfc SwEt {,U`@`_Set As StraRLined9P"Nv l_? to bepWP)BszF [ݴ# B  cad fo@)+}go=a5}oPUFDfP h> /T6DUmA@ ?ۿI?Y{O1@3EEbOeZ HG 9V?o)>  `r/`wpw> ȳpke{rpgTHYY9 mAUA? @ ?Q6 u` ?Mu ~YYA|XA44 4> !bJfvqQ >!"q06DNX lL>U5 LO@%#A?AE^?AL#l%u%?P6 @*  L#&H   " ' ]{%#W1\!\!5 `7Copyright 1999 Visio Corporation. All 2s reserved.`0_Sba.chm!#22427Bd(,*9 l>(>Ud!E (!(ZU$ B$bW* /T6DUmA@ ?ۿI?Y{\ $Oc@SeeeZ? $H S &4G ?rp{ri^ARectangle with variabcorner rounding.f s on r.mb?贁N? ?4 HD # =h(>TP9 AUAZ@ ??Q6 V  , A+L(2u` 6u>d7 ; vB@J{(!-$5 `7Copyright 1999 Visio Corporation. All R"s reservedo.`] _Sba.chm!#22433Kl> >Ud]!! %.?!<r r<%G"bb%I6B3 . -3!3! #y 2!:?AԲ?$6#$3  C'a9n2 @r"NH6 "I5 3ҡ2G2GBG22B! K(;?0<::E3H81|j!<7GB 9D#G"AEBA3 AEK!:096CHHBFh322p!:E"B]dB!J\068D cfBDFo#^o.:B\lq,^E@/3/7065"AChUTb 1}Y0s{V#rCQVS 1%{yq|\ޏ0dA!a| = o6b0\; Y01FC B /T6DUmA@ ?ۿI?Y{O1@3EEbOeZ! Hu )??/CQbP<]sUKE8Arrow with a 45-degree ahead.HDB # ?hZ4>THYY9 dA[UA? ?P-DT! @A-u `u bu  @"Y!T FWu`5 tue.A2+8&>s@$U'P"_/q%'i-/ ĉ7$#5 ÀL!#  6 Bd*%#/ Fh?Q(>u{%6{`("u?"F&`Qps-8R? WI@ !Q6Nu@{$#AD5 `7Copyright 1999 Visio Corporation. All Bs reserved.h`@_Sba.chm!#22435pB!$C tNLs!OR U\Ul> <>UdP!QQ Y!b(Z$+B!4b?GQ X b#=b(3$$T!v8 `RrB!db`SaEcjp= ףpwoOR"ccb5$*0g23oeE/mOQ|zbeb)U2`mIwzl_{_o:zofaSaQvy_uA w'kpdTAha9 G75 ~NpFm@QB?bZt!(! %?2vB@jIsJ::Y +@ hhJb??贁N8i6??Hl_-s !Oya GE_2szFT 2+w#_ KB < O`,oc|kUo@k͍oPKT  d0;a(  ;S{ +]&B?9߄NRD HS" "/"T $w *w 0T UPmm@$I$Ir @FDTey P ah$T DUI(\(@Ia,Œ?k@??I?. K]o]ؽ|/!/3/E/W/d)d//////// ??/?R<9U2a=?s=h8u` 222u$9[32{5FN,1NHsUWAUFt\wOgO7KTDR2O8&C` Black & WhitenyO8[3QoU60SColor Schemes...1`'Set thek cRsR fPthis doc_umentP `1R` ZQh0eJ PYBBdE3ONM%b#aPReST@VU[\]YfcdQAijklRm%QopUqrstUuvwxUyz{|U}mH_1eT=Ѧl4_Aw9Cy;=xV4@GzT @@_Mp@J:W@:WP2lA-`Bu_@`u dPbu 6 ;(@"i1D4 @PP:4?|ID?OK%I㹶ʿS@ @@)\(40?Q}IQ X#0V3kE}BUK*TD?o]W%p/c@:?M_qH?/j@S@@ZwRBՠV3TjoxJ(:L^p! nF~[I+m+t,nFGmGSONnFWM}G_/I?)/;/M/_///T>o#ki8}f0//gѣi ??.?__*T 0 qqmj:sʿ1o=oautXo+㟵os+oH-q(Noun)iׄҁwM9爼'2q11CUgyu)\}w`@| r@,>IRj|ď֏Xj|Bџ>ߢb63=!83Pro̯ޯ&2DVhz¿Ժ(j@gE#v@@r@H?͸CC F'62l~ϢϴZl~D?cv߬-Iu-/J.Figure-2 : Transfer mappings acquired}* U*:2 4!?? Tf?$9U@*;L]?@HQ?@8@dECE{O23ϱJO1IDwFk-z[&, Iq0F@U/g/y//QUGy__XE__d_! o_DoK/o_Pzoo/Qd ccinV21S?e?w????????@p= ףUEE5OGOYOFɏqVLOOOG OOO__+_=_O_oooo_kvT7o_zKo9Kf@o0ITf+hipervn/culo+r%0>t7qC/t#,茁D{Epp"@Ѽ4GnAim8/vؘھǿjQ苦Cn8@ &/d//qAzx @?yF@y΃ <݇qgyѮъE1r 3@oֆ08Frp7~끥x8Fz󁄑1,qAA=u/;9(уu灛ر(G(ў>L A{ѿ0ѿpsܢ|ƀAu ou ohqCA? A@FP BNogjvfveozpj Q2P:˯;sŁӀ7pܢ tNLoO1zsjܢ Xq! ON_  /VWqdC2rq?@Iw ?m$u%ps,lǂ߂-8f|o҅Z~#02!(э]q?Zrz`0Af?ϲTll?AЁ sₚ(߁U1iΠ/x7J0LO^O aeT3 weV4%7I[my(y4,7U@d0@bVe?Te .@APfel?ЏO* DOzOUWooן#xU6H o`o}o5hů rKHylink_InformationBR ďSewoP}ڿU U1=oU7Oylϯ}?Ēʼn$6HZ~ߐߢ#ߧo2V׿zݿ@OO__&_8_JY7̢??_U@Gz@ ?@R8S"U___Yi ,j-֟Ln6Ց jQBcYQ]ooooro'9KU3dWj|v' 'XJHaddress ߀1CUgyߋߝРZۧ?ħ׃i@i&8J\ gyCUgy?Q/?*8=?O?/ ?/?>/b-lXo'*g:)˝! dg Rk t`zs&!R bc.! EFbrRrEb &v 0BaqU2rq?@Ip?tu8OJAΨ H#0W$F`(ՔP+`0 ɑ՟iHT67O6d!%#v".@/U@/4O 6?sP7".ModBR$OOM[=!,&E?Y!A Ce?@T @@_Mp@JbV+@P2!;``u QbX5@%"!X*p@T/*5SRR.ӶYc\R @@0 @-+ҷ&ҷ5GӶs?}GChEOasd%rN`**Oߟ=Mf36S/ASew ߿Ѭl硲cb/  .4 !3EWi{d~Fy5ƀ"dcIĽds?1CUߤycvA!l8,PѬ}:H}lPZwRBՠV# TjoxJ0,>PvAh&}<\m+tF#߼&M7'IN?`>&7a=7?DVhz5@ LT.O WTfVu*Y&O8OTߏ§4 @@@QEQT B?T, S5vBA//Żh%fi8iLai!iaA@E}0GoQ}} YAQGMOUaKě}IޟBMĕV>7EY~UrMq~)Mqa@IC_hacerRVd\_,_>_uXYQBqlTt\CB[_U@HzGͦ@Y݂ ̅__oǹh>oPobjooooooQcuMί;}@PtF٢K(Pron)iVDQlT=Iڡx4t['2q2QWO_F_+=OX@@88UUğXh 1CUgy ӯ֫b3| WͿxz IclicZpOϠU\Ut,iWt\t`WE)?\(<5 *-%$AW9& In  pP/aQt[oa{f BfBÏ8CQD/CՄ QdDVhyy:!aQ?〃F@/cYAA<GhQ)AAnXr5xA=t5&3>p= ףp4t;;>%3X>!Kr3O 5t?6\X_j_|__\F#߼Zs 6߼߀"`A)iVߙbjB@yǒ3do>P0?tfnVoi(olZ-}Qflo@o enB]OOOOOO_#_5_^O|j_\ Ѓ____d__ oo-o?oQocouooo -or.~=,ۯGkׯ41Ds ubk^ewя^p &8J\n&N֥ɟ۟#5GYɿۿF'2q ༭qiq ߯VhzZQU@1E?QEQT?@8¸?{׃ʤO %7I[mU{[,-?u߽x,|s2hacerGữͯ'9K]Hz[Gl@$~̐G"Y6 Yk}/#/5/MO<3eOO[HO_~OH_aOpl_~_!3ProD@Vhzj///// ?~ZFy5p@@88?@a+?9tAu[?m??zav?????O"O4OFO___ON:vf_OCuӏ7_Qc x0_[x_ 2clicbk{ooooo%7IVE#߼ZUl75ҷ·\nϹ ߹4zNS4gMS /Aß՟֦.$dԪ8%@ω #.@^ 2en"4FXj|^|jŮ׮Ip!M_q˿ݿ_q߃ߕ[ϥ$CUyT?Dsubm' 9K]+xT:9& 1CUgy1CU??uOO?PO?tO?OO_qobjnoR}]Y? /j}PAxV4@?@ȥ_Mp@J7@b7P2`1-ruP`u b@bu`uUd+q"dQ\hLhq[CRaLb Wacab\bd/`/{PQyf8OS@ cT@@)\( dT/@QOR"MwgHrgWiu@fwa}wip8m#O0r?CFgQpS>2_/gQw?U;*OD_=o&chjCXQoc_u____._\j@S`Zw`BՠVh-hTjoxJ@oRodovooooooo8tI+)pOm+tP2pI"2p>}ާkt߯&򭕧0 aASewƏ@ĥjC,Y wp*U@,4Ƣày?#%,5GYyeZ_//o@H/????/U$Q$?u/.>?clickltOx1ҏ?TW?@¿ҿGYkƟ؟ 2z߻ گ,?Q#uЊ߮>ȿ9 direcciniPBoToؔ#C'2Ǖqbmu /ASewH-R"P| 5GYk}ߏߡ#5G  /m0(M/////?=q8addresskv!3E Wi{+@= 1"V9K]o'?9?K?/_C_/oq/2h/Oo?//oo ?o_oϽ(Pron)??????kwfx#O5OYIQ$jG`QOYS_OU@RQ?@(?@8À?@q2ÃʅFO@_|q';Z-[^yq@Q !hNScPwq$鰎 tN&qzshQ ֱ&q<Ҁ+2rr?@[I?!Vx̃ſ&%S/0t0"2)BCYA{?`0C0- -H*= MYk PE/i$:/. E&GuideTheDocPage-1Gesture FormatVisio 90ConnectorVisio 00Visio 01Visio 02Visio 03Visio 10Visio 11Visio 12Visio 13Visio 20Visio 21Visio 22Visio 23Visio 50Visio 51Visio 52Visio 53Visio 70Visio 80BasicBasic ShadowDynamic conn?ectorSchemeNameLine-curve connectorScaleFactor45 degree single.745 degree single.845 degree singleEllipseRounded rect?angleDynamic connector.3845 degree single.78Dynamic connector.8145 degree single.9045 degree single.114Dynamic connector.12145 degree single.123 :3Tj 7p EdA tL- RWGUo |Uc ՗c c kW c ,c UDc \c Utc c դc qd Թ;&;l 3;m@;4mM;LmZ;dmg;|Rmt;m~;m;UĪm;m;Um;m ;Ay3 ;U 7DVhz .@Rdv 763* 3!/3/E/W/i/{//////// ??/?A?S?e?w??????;Ro0tE;n0ryO 5!AO1V@VB!ALOOO5.O^EThisD0cume2x __ fA1 _S_fA_qD_PROJ0C@__`.TA __+%!.fA!- R kA3U/BW/RW!"J#$Q&U'()*U+,3./U013345683/oo+=Oas*..q.kT.pS@zaA"U_N`WfA$%v!qx d8ME;7/ A 2DxV7aSL?St?2Lk@!wP3D7dm6,Syem`Lr@bA [%e@vRP\1]%Z`3=BT^nzWNT?SJ@s0eP\Qd0tby`UOEu0oPt&Y3@_@  ʋ;׿>FpM`OS`1cJa`f`0`FU-RA%f@ //,/]/o/////////?#?5?G?Y?k?}??????6p$OkPj1`k`-R  d`VisioVBA"@Win16~5D32,AMac/A6# ACL_2001_Fig1!stdole` 26t< @_EvaluEjӲ-<ռ4P< @N,b5b_5f MlY?rq8R0* pHod!`bE@jB@@ = +x * J< ruBe>@rf%\*\G{020430-;@C0~6}#2.0xп#C:\WIT\System32\c2.?b#OLE ation`t.pF2G@QcPSnY2`H1 ,C"? +KP%ID="{A2B73420-2F4B-4CA2-AB9E-3CF23EE2B154}@E=I/&HMb )e="bJ@HelpContextra0@Ver'@nComp`ble32="39322f@aCMG="9E9C6D1195?313735yaPB`C3ECFAFF34CF4>s@GC="DAD82955C656Wq39@ [Host Extender Info] f1={3832D640-CF90-11CF-8E43-00A0C911005A};VBE;I)i A&8J\nȏڏ"4FXj|ğ֟ 0BTfxү,>Pbtο(:L^pςϔϦϸ$6HZl~ߐߢߴ 2DVhz .@Rdv\: $<ua BU^U U UUUl4,(\(@a,Œ?@ ԣ KC-3 7AUl4, < " AJ-43,7A_*<N@$ Հ MRVga .6uH<( H<( _*<NEl RVgDb q &AM3'AM3(PPT)PPT1_K*/t aV$u L;t;՜;;;v <dZܧw ,T|}r~V ln<tģ즣.d.j...{.T{.|{. AԪ*+*x 8 "&*'()YU  !1UTU*!"#7%5{ * g"4pFX(G;v@(pMy S X :]R}4Y  d&;!B 4Z \Ƶ >"@g _91$  ɓ'lN [ m)<obb =$_`2,e wsV o YW+?L _dGD_r 5 C=\t .U FY CR<_7 By>      !"#$%&'()*+,-./0123456789:;<=?@BCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxOh+'0@HXdp|arulmG4Eg EMF,El@VISIODrawingLD ??d(@(@DocumentSummaryInformation8(1TableAOoSummaryInformation( DocumentSummaryInformation8)l(@HP\h t   PagesMastersPage-1Dynamic connectorLine-curve connectorEllipseRounded rectangle45 degree single8_VPID_ALTERNATENAMESm_VPID_PREVIEWS_PID_LINKBASE A {K۾HFg"biקҮo+#:i]7Ko?!#O?L_G`w5/m^+Ɵxח1ywƃxh1ڋd,㫃=v8=ۧفݞnA` ;\x-g|yxr} ͌}Kg<x?pp:Qv?.0?7?C? S= &3g0ua<-{aQƟE&Pv P〫_ߛ@]kk0~`<ԛ]? ׈\:XPW+02xkN23E++oA>; l e./5|>w8>ܷI]0.2iK ۿǍBF#|-> 7D "%KD#uyT=S##C~O }=/awmO>=~xwwi̿C});"77> #C}wc{S>@}~0 0q?i`"9u31%G/{B]2~EmZ6$OvPdi1V/S{Oܟ vxzOS]3rS2g] .=;g] G,0Y<[gYy,jΫg y%_<˿<;ϲ³,zv~ϳEgY1w.˞wYyy%Νٹ=ϝ<8bΝ<9 FG#/|ߒWE~ǖ󇈾wEϝx =;y۳s'zv3ϝ@]ؗN_fΕ|ٹJ~p6| kO#[OxzN;؋xWnƃo X*90_pg*K) }=&oripa~qCWXs'fνo᫦1#ws|{d)JL_31j%OcJx4+y%?sMJ~JAtw1~J^䁸6䍌ҟ3~"% JsC7}%1_Odͥ`)ߥ䊌|%Wg<̗'܀Og Jn|=o#xG_+3*'90j 㡟=G2{)bu}Yk_^@"ޮ/xKxdoC OJ|;A a>JsgNy>m}$|[_G>m}Ws:|2OQrSr||ΨmS*ub*r>'0?㫃=0(38})Lv?xҎa~u1>/q[}7=V yH%x'?ev  <O,tT9xȋe<{U/j^SS!x(ӵd<0uaJ_YXҕ/ ;%yyz-^Z&z~'0UxXTY<)1VUIE+p+7v^v8 x; )Aͫ:EhEHΏiii i|0.|<{g\n1l>cSW@c|w=."sX`_pc}d2֞PZ {-D^ ʱB.$(0l^̱>D;ڇƎՍhxFs0_9!ϟk#`IȜ60Gl"sZ߆m|-yAd _x^98|m̗k#`9<##2i9&2d6^qck#*9ڈJ6d6~/_k#@_|m]:Zx_|-O?_ B=0iOQ{k%`_+Fcbٜ'V<< owIQ|=iwu=6+= }k3>txh305a~g|ց^׫RÜ.=&+8 9{ #s>F<g=b,c_ *_uy&//#_ƿc>!a;og?:Z7`}LjOSF')W򗌯|-~_۵׊3LrEZ'<;|{W&2W=!1k]>x+2~ߘ1]+ƞ]^k\̉Sr]??tz'3료z=qPWz=(V7z=Wqq}'Ӻz}]>͌szwz9"|FoDs"E~doE#y #Bogy\n0}!OIҮ?߿[gi_Xwgȓ*?1_ƸoF؃({{?1noojik8P6 u{`QQ׳JQ8< +ySsR<F^O$$~ʜMTH}Tpy/)pU}Ia>{ {9CWrj7Hm=Y+p>y@${'ϯc1/Uȗ|nM;U2kϠϩjtOQdNu&Ձ\>)i_Pgg.t]7Nڻnubu|Fd83yZs/%`,a+9TXl,0:2:o ^KΌ˩=VTd;En=M{w/NM1HE(KG9~vw|#iqsjV|=k]xV7\d[IJ}*Z>G7,u`Yr&ءw{S<)={65Tua\ryE9qz& [v> (&vP[qN6[*`2őހ{ɹyɞswuv$}a%h%g:S Kk֧3'ao$֡ɬ3h9s7XrswqW.ej<*]w塸_L,> v4<{ǃA̾ sy˫`>LSReJzT6vurj8q}}>ιVw'^-xl`~/'7ί~_Ldh_?TW}`KE<a*™Õ}>>.|7n71,:%P3g1۪OWҏB_L}67YzN.7`ga}{ ks ;0K1zؕ 8>]sr׺E+w=W '](]w6G?N~#?4}Gt}LbcpU~LkW%{q{qw~WoB> W-&W} E ۙ"8]3ڤ= oqw\yb?u~{G0{zjS"%han&S&ㅒg|䍌OW)yJݟ__J.; '[쬜Õ|Pr9Úc_A'0k2h%fWRrC(LWQ?.׽=ܑՕܕ5zL>I'+y8k+y4(y")VƟYl&%^R쥄wO񳍔?){4g$f>O&J~g&yğ竚3 |B.ovB7X̄`?o Ta9`??7+䆌? /d|+~Ke|>OS"^g|-%ֲ?x`㫁}3:|-s؃anI8VFi?d>{,ˀ=04=DBAr&?#c=Оf<xaw*^Cʲݑe52~{}WYzVsƀbV%-JMY/^Jcwu%ϕYkW%moBΕηe]?3>-܇ ߎGv|=ַI3f3|~6bBw[lń}FG~FO~&1 LcXX33 Y>3NHgdA{ȪȂ?# ? C }ȗE2? U?Ăz$X?, #? d~V}"~swDgX 3@Ύ#X ,x_ d~&? 셟2? 썟ϼg^ϼ{g^A}g^ ,_ ,++3>nv|Cc'#Sv`j0u6lڌpgXA~/v|`P`Ǘ#XC|9)=L*hfv{6?ۻ. rq` SgJA! d<ğ;UYO2>s%g33%x2a=c P",}&RY(zĮ^ulf~Os+Vɫ34}Ɩ9_L[?D_1~`@%b<ȿ3ykq2 dc|4;KSɧ13[3U]܉+;ٹ޿+y0;)Z_qɌD30仙Ax1= ہ0x_ ^ ۂ0 A&"?^x. ^x.q?{^ {ac`?9s~c{a<=G`C{~-a/qi@E=W y6,-fcGm1:-Cl,at= ki8-E{V:y'm_@h^,TYŘ޳Ĵ1}v9?)_䙌2*c:bBLQ$(ntŜSy<1G1u?(y< c:ob|SLM鼉mp 8k[gt֘Σyc:b|LQLyoI}&fΛgr'OI}&eΛg~\gD9pD~4"ϻ$¿"Xƃ~0o< _"S}Exlc) ~*¶ǚ3$ωۜ1;o%ԉ\9}$q_{g9b:oB<.}%_0CϹ3c:b+7?+Cff 6Cn(:>0cO^%9Faf*yΰUS]L>ץ1 ;VJ7s 21F:B<[*y-{e_R7yB%zތ({wS򯌿"C/0DJo%g2~p=?\M u'(&'e1}zTrRr V9L[ɭ"?K;3iO?+D ` XOsxd+q{Kp>:5234%`OQPOTu<Ȏ54ܪ7nٽXɫrlߴo/c<_fZəl2Xk9˜mS-C,3f<-[Wr;_Kɽݟ~ s6U`PYJך+y?Sɓr%Oc|g12d|%?c s2J/kL]W:Ɵ_[ɛ{k2zoW7TN7Vf+9Ȧ'_k֛Vzm֛Mz-gfQ^Koff>Q^Koffb^Ko3ffO^;Ozi,Rr጗Jh+rS% 9qh T<π~o/b<-.v_kb<-0nׄX%f{|J(yci1n ~ϝLnN3|?8!~DŽ3R{rvCI Ru`_}XhNZ^ڱ^k^-B{6k^;fk ٿ{m6^m`ka_E`,fo ٛ{mǁ6^~`Zwq^ x^?J>?dǁޫ;/"Rv{,s_9|fD?W{X`硞 Zm Zm3hnƩw2>=?{ l<+y'G 9}%>O+sny ۅ)Mn r:M<0- ?a꿌uf`O|}`U3Kǎ|:ansƷ ް"Ʒ /f %wa]x17sM|z?w6{uH=\TƐaKkǡנ0=U^KnR㏔qVm|\%iK?/qiXG A9L~Lwf|Jɖzh3Gkݟǹ=n}6kR[\/dRVuX҅WH]'M(ql+um9W3\x#20de1O!0B؁#<~?0ͼ双_W #??ώ#?6|#"?(D~wԟ>aw^|["A_e;_ۇ_ۏ_ۗ_>|3|m{{gtoAɟp{t}2fXj؃<0~O }xsD&y_?3W}(>2aRn/_aa/{ <=݇0<݇0񝒟`J~%?j(xγݝx;000006qa삷T^a+W!L{>{S8/Uh?<^rnvu|`̸ctX]~]SO^5eI}&x|\S^wW :3uUm{ty}s݋<,nd%e}sM$ԓgg GkvQGU=-]Mjqyb؜>H9C|`OQ-9I-3\nܼ>}yn-eڊI=Gh$] ?á5JKpӵGz|ƪƪ?P$`oU}$W]6u8I8IaL'g^q 6\;K^5do|m^spӣ `kATA4oQuL^4ӒT^hHZ(lA WczN4ZIm$ٳO T }{i5x3y4M^[[۸/ѽW߿`*m/-,yr׶&ش/}mM֮ͅ)=\ L~uaLK_rZs>璉,jRHxٰe-tˆ{b܊=5nv![-Imsk#[_9^#BtIJ5CT;^+zmk]X3=bkuDRbݼΈ]uE*;bx=F&JĦz}4 Ke;SK:H'TtGkэ+"茘+;!.;"v#e{*ɋ;^ElZ|dk"og#FXΒq<Y*b سb/{EFlLedbHV/NX؁unbO-R p%u#kX3?v>RWE,קx럈$dĦAlO_>b!b}iOG b}jd;f7Nld}AmbhX8\* 5D+ҩhM:TW ?D[ҩt*.$ jcD;Ҥvw:Qt%SHXR:5BAbA~`VKHXR:5fm34/ ːe*Iu%=KcILhev~et\S#I)\D>KzIHz |ƔO\Ed%@5mBVC%:،9;8tW yIA#i@uU@>]p 4G: &Zddp"i2A &ճ * 4˃bz[䝈3ī+ !ޫ2;T.b? _2U"V?~وFbRS?% KJ~iKJ~&i׏vgvk l^1 SZe6(횧<w_$D)ҩ  tUEj SAzoJ z ]Ao7I[!MJSC:UH8ҩ8t*NEMdH6iW%SHT4"H TP|ܡCdkJ˶)&%/B6G %Ֆ!vXSbme/.A^e?zJCl$7JDl$O8Yr3卤{9ɡ8{JކNĖʻȞbFN?ui}%݈uEAlOZڧVj;b7>\w3V!+bS{ٿZ=fS.Tϟ$ľm߄#&"&,e#YB%,~jT1XR3`] ` v3QHڕvҮ|x'Iiҩ\9/ \%Ğ{I.GU:QLu>EA~~46ہw#6E:=/<>(QѣXc" EX{aQG:"7=Q\ڣjGqIS{aˣkG֫EXxaӣ2beZ/{k-(ZQG֋EXK) <[̓~ev_) m b9T s=iRh.'VNֿIb T%7Ib=TM:NSt*>"I4)>Eգ e`D 5boo{_@l //}'v"/bk,Ǿ ;\ 3"K!VCf#VGE<'/ZOy*MVCSV$Į(OAjiM"#6CDll<e;Ğ#''UOYF95}#G".91?Xz0`J3-BbpOٛyU@V,Byb0X%Z2:o>@lX b#- ۜ^mN}ImNImNImΩMRwnlF*݃%f 1ba)[j׼[2"csVo1;v }>=>=1>j>ʓ5OA_~]uدɈO$M5I~uҤGTCAk##EZ%uA!UE !^=d'"4bp'bL v1NP?gpAYA b B[ŋ#½Mt铎0}K> >e5ot.P~sbwjZM^C]W{wh"ƽ(oCYC`|ʷMFhW3F(Z390uC]jhWf&oJsha FNvhPLjXa Y1ef ʦW, ʵ.Q>mm0ʻm349_R6Cml>klfk%%MҤ]o#44VRztI-O,/m-ʺSI岁點JڕHKM I$w}V@#4:P!id!W;I9 4PMnC*4. oqip=b]q (w/2SPkt@y < C4 "vObK{$x JĞ "$"J׃/[|C(wmK92 W#6&|!Dž!vCQ!o G=1 741LMan "6)<qM OB>b7g vKa[vFlZxb<ĦCDlf8͈ކm݈>{;—#\aՈ FlV>3v B aYFlvxN2X}akcl+ gi`g<{`%AzYAw`o#AzL'A3w%5ђj3ÓRyFx$j!ɮHM ;K+$dbSálw${O=p),VF=nĦ 69|\RzVR-{IR[!6TWߖTW?B3Iuz,|]܁ RaϚvK)>_AVÚFYtsTEn!WTkJ{^۽kJ_uUA׽j;ucN99ߺT;jW3yj^m_Z^i[JgׯgJw0 7נ :_|IR:&KPjE\C^+OEv|Y{єr>29s⣝Yn/۞b`od6.-G!}{ɴK^r =ieu?ٔۓՌƞ4#Iaơ'=i>YŽ'͚d&\8\4ŋkO",-qCl;b+݈z y wCľp38Y^)ҽlGwbzG!vW^<ن^%r;׫؅quGW\f|INFlW^n!6;{=읊^C,=2 ^cĖy!w:bk3{kBz$={g!׌%g9?%goIzƒҳi?%ؓ<.v oEŒҮIXR:5DjYS,))`e{|Zyly/}g_oݔ\A2]6&iXa4)Z&™9iRPEg&ESҤ ;(6 тlh Ud c>iR &EҤؿE-ҩ8t*l5I:TO:H*iRV&MJ$c%$ڒACy$iR%M2IY4)K%h\t*3I2t*I2 JA:TF8K>^/Y,Zi;V툭_!VĶ-Yæξ2Ԫ΢}e~͢Mfh_'Aʐ/I}z#~ZIKо2ҌSK3ɧf_:AʔIо2ȇ Wʹ-d+}u1 _{lPWIڪ Wʹ"e4ӖOyQ#A3m5f՞YT:JP{NP{uMMcj ,SHKL4IX HƯ7:0b㓖 jWFS,Z6lvl<_eH² V` VCՕ| `$v&u\S#IC$#$y$=OuIlIm$ϻ@"MJG1Œ咼JIxA:l'N%yyAyA4d& d@1|<Ȃ€<‼<`WsxbO䗃h_h7YY4KIJ+~X|p#ycY}eƦQbt,Gw$bk[^jĶ vݔ7ٸX_&'v^oy;Xsbni {]uAw9b}Έ ص^'y!6 x#v)u@Qo],)zI^;Үwi׻]@ڒv6iڌJ?`I;&bIioEŒҮi?%]ĚRs,)=ߒ%][vkq٤],Ta9`l2zm0FSbد y]晅 ,.{+j US'H }sAo-# zoo_Aw8+ qdhKvŅ]qiW#]alW=['ǒҳ+:vť]Auk;iWv&I iWv%n`5ewҳIzWeoҳ%=>gٗ,Hϲ?Y$=A]9+vդ]y iW HzҮMڕc:{ìLuxJux:i4ZbDbWi&4Z0V>F+QH")=?FkRFl b#Ɵ;<\dFgbو4aX*!@q;bm,f"@@FN#VLCb,e{[IXRzSIXR5b-],)Sz°l'1;>6O<7l~nԁҩ$m$$iIZ;ҔF|$7fHz[-]SIobIorIo$/^IS`"ʇIriW>JڕN哤S4T.L_/"6Gy>F<bh1/hy$F<b~1hy1/fn;FbfhY1/Bs[bfh1/fZ]bn.17hcK̔3%FL.1cK̤3)F6`3 6`aSz԰̷-1VZoty-`Tҩ+hCP\TA}MAot*N]SqT%ySq?TbKd Ŝ,ʖ-һ TGl*!W }2h7X`8b;.X`buI5 fZ癈f!vN0;V=ߍE|. BsiĮ Cl@a2Į VduvYimZuqM8&Nvy/N|t!N|d>n2i7qMf3boK4N|}eƩy_Li7`/̗qacmqaf[v!b]nOߧ'%;OI>iJOH>I>w|z5;N;̬SFZr8_S1b;U}W;Y<~gb#!q#6Tg b9Hk$Қs b =܎ ^v:!ylvAKľs^$,Cl%tW#V}[TSNpɛrs5p vKl+]Sj~l;{.uB+b]wLl]?[~7Dl)scNr3Bl\,b{ܭ31-qs[VEl{,۬W3Z}nah&mͤL,ͤIG3i34fhNVۉ6ݦǕՙF[l}jǵݎyݶZe\9ڭJk}k8h:cWZE|\n+*]5Qw⎅5Ok|w NĔ\Nv;nbyn3m[Sښ×T}(\|&m+MtJwڇa5F茰uTRLҖJ/'xߤQЁЁV@ jDD\;dCߡ(ՇgGՇ Շ\CC!3P}HTr9 ՇTB!CP}A WՇ셞w]FkZ=n@&ZH2ggZ[ZA+pKd;b7:BfAhӡ3z,8.s%-nٽ UMix(w E^Fм/BKChu >-Jυvi>Ϡ={B%h=|gZ_N񩴁=e:G}2ƨ%mYKƛJ}hg>m:@[kVP4EhoM)?[?4NګWイsѮ=;zm^v;yv݇FXQ,$̣LZ2yU:`)s"킱^ɺJ@WM0!|<UsA+D(FC)6  3 A2vpv+;~zh`!rvpv+;~>6 p*C@x]pU]HdS*$jU`dPcɵII4ih@+j3"#*ftq&8`Q;SUQ{voݾ2dZgXjFGC,k=]쬁Nje@<3ߎ,[M5T'4HW' ۾n脛އQ聭0CA~HH9m*nk|Gqyy( #L͸Wa؂Aieg.Vf!/?,5lDR&AN Kw.=ssp=ȼ{ʴLeSWWfłz' x/WPhlєeQr\m3\ӋyB |:?~A/>=օA<o /;Ÿ }>"U"Q>g9o >e3xMG 7q6W6R{Z-F}|7n3BQ6xF=A׏Oz ;y^>\wk7B<\}i1J)S]Up2(ǩmHgmp-eZَ5Hke\5Vbz%? ζ+TL*EY|%,} !*H)ou",OWB9lբ͔f2>]\xhsLC2~Yfa~ߖ뤳ߍ]ڌijI1MIeOu9uLZLe2w}8te !VH⅍xev1!G~2y [|{%;C)%jgj# ?w.ޣgY|{ug.úW{c3|{ͦI9͜Y'}|{e=KCS.^Ǚ7~K{E+2|&Cg}dvLO03=͔S6gs^L=(KF TVlYhʊ]EV:? +4AilL)X16!I33=)}jzZVr5=nH[#=tG25|$>>~#sH1u4ϩdH)~XʉkDe >Q2rHqJĩt%+@,ksyː~(M{BKJ;.GXf^X׎fAQɯw@iJ _ ^Ϙ9)yty(p1??ʉ8Tt7<DRO݈tk.R7{ (+ۅ+j<98BSO|^>-d<*?>c0g F)]{1y191i WL ||mjZbRuS~fDSaytz=$Kč=߿M50㰮 {됁/XB ,T7G1(?&Ƨ46>9v"V2G]|Oy;~~#oFq]A69Gz5x6 _|&zehQ 8jq#1~ᒛͬ>]3^=)㈡!ďa ?>kъ+}Huw`u{aIQc׵3j.'Iy~V9.6:ƏKJ1dzbJp[G&GxIp8GVG%GXElqQx4,Gﱦ%Gg}_,bZ,YnvO]6X2urAvbFޯl]D3apEE$wčE(v)>i55]6)캙us:S<]Ssz!55 Į`uŇ]?}G9:?1}D:VXk^^H}þF]KxJ.1u# ?뻤IDLHo .}R_MoG{$v{0u gyXO1W3-~tL~zTs9ȹ.Og`&e'z Dy̵ұn4"\3<]<>o?uOinc_DZB{>-dbʅ.LygO-nLcBćg)oq>H"^]{m~vHu܏z -Lz{{^Ǝ^ïO+V`Hz/0{9Lgjl^QY7317S{ep̮t7K׬)y?*ot~N8}/1aDymr?ȏ`x6 w4fFy:._r/cG Rjvզ|xΡm'wB:kc]G:iX30:vkcJשcTh62X >A\Tj_ҘWfhexxcw9lܘr\Uiz-S8CmcywǮ|.۵Wxw}PۥZjhҞ1 ^k&ZtgAp,_X.ZG'y{G=uuUU25~Uw(5'GWihZ>sy_)gF_i{h+:4S|S9Maz7j\-3j>-6=Z|.o(?6fy*6gY*l>BaZ!/F:SO<.ř,\ ka ^ϝ%,T^}׬G6~:y,x*ˣUV jħ=80tN[Pgʞ{{3>ܫ8\~bAO8uL50Osn휺1Mv=iGӿO.v9מ}^.g&x%?WV Y~9_ ,8b `Th.ú ZiurǼ50\kj b ޒcjIN!9?e4vN{1M~7]\rLWoe5Q8\ⓝ_ eI.J]Cdƪ(3+dҒBf4s+2\jkةf)3xpy^2}mG-tDPN3PNw4hb9eGY(O9?R[)|4"{>o,j'3%GLЗ'.Wzz-6z3pI߆s9, ޵}6e:zt E׌WAaW׻sÎ̞خ~ivemq_r5{bOukݸ=0]@VP⟰MQ@-tDPz_9b]KX!lQ-n׵Ijvobl "4N+WL "",[\zu{ }:OD4ŕx<ӷņ*4zC[zۋ_9Y]}^"*2s% 'r@~zDcta+kyM4M|wUW=ʟGvveY-K@3L& COCȶ,i 6 b<_v|B' h;;GsDgmYbj,>>O\'1lqUvg]]@X7ަp}/k=]/)ylb ?>'`(lFa ͢Qg*v,*sBwK]X$o)u+4qf>:&y]"ۼzf Jcj]Ͳy M(vBBz:vGtGt0=?8?@tuQkʩ=UA*4ED]ƇUq$F_pUchU>18q5==Zp(wK[Vl*,lm3?Ҍq6_1|`ha%K[W /n0ӻy>Alw5<>t=2K$yLAR[>Oh+'04@L` x    fA best-first alignment algorithm for automatic extraction of transfer mappings from bilingual corporao be Mike Carlsonaliikeike aclsub.dotn Mike Carlsonali38eMicrosoft Word 9.0e@4@l!@@NR{U՜.+,D՜.+,T hp  6 Microsoftu+h fA best-first alignment algorithm for automatic extraction of transfer mappings from bilingual corpora Title( 8@ _PID_HLINKSA 5 mailto:steveri@microsoft.com[omailto:arulm@microsoft.com  FMicrosoft Word Document MSWordDocWord.Document.89q) iT@TNormal$7$8$a$'CJOJQJ_HaJmHnHsH tH u\@\ Heading 1!$$ xx1$@&a$5CJmH sH tH u\@\ Heading 2&$$xx@&]^a$5CJmH sH uXX Heading 3$$ x@&a$5CJmH sH tH uRR Heading 4$$$7$8$@&a$6CJOJQJ]aJRR Heading 5$$$7$8$@&a$5CJOJQJ\aJ<A@< Default Paragraph Font` @` Footer,$ 0n8!x]a$<CJaJmH sH uX@X Header 0n#x6<CJ]aJmH sH u@&@ Footnote Reference CJEHaJ^"^ Footnote Text 0n1$CJaJmH sH uHO2H Bibliography '(`tH uNOBN Author#$$ (#1$a$ mH sH uV>@RV Title$$$ 0dx1$a$5CJmH sH tH uZObZ E-mail address$]a$CJOJQJ_HmH sH tH (U@q( Hyperlink>*B*DD Abstract1$Q1$]Q^tH u$OQ$ Title2CJrOrSection Heading$$ & F2^2`5CJ_HmHnHsH tH uROR Subsubsection Heading  & FxCJuHOHSubsection Heading  & FxCJ>> File Name1$CJOJQJtH uOa URL1$DD Abstract2$1$]^tH ufC@fBody Text Indent $hL7$8$^h`La$6CJOJQJ]aJ>V> FollowedHyperlink >*B* phX"XMTS main body text"d$7$8$CJOJQJaJsH Z2ZMTS table caption#$7$8$a$CJOJQJaJsH <TB< Block Text$]^uDR@RD Body Text Indent 2 %h`huZOqbZ References&x7$8$^`OJPJQJ^JtHu<Z@r< Plain Text'CJOJQJ^JaJhOqh References Heading (7$8$%5CJOJPJQJ\^JaJtHugR LJN_Zb[g1eŰ%QfgPQR[u V/is&B_VW2!!"#$?%@%%%@)))*,0--.].z../2!25;6O6^7v7799%===AAAAABB B%B/B0BKBPBQBoBwBxBBBBBBBBC CC%C&CXCdCE#GGGmHHHIPK[KKKLLLLM>MZM[MeMpM{MMMMMMMMMMMMMMMMN NN$N*N+NHNINJNUNNOPKQ#R$R%R2RTRUR`RgRyRzRRRRRRRTUU(WXY^Z_Z`ZtZ~ZZZZZZZZZZZZZZZZZZZ[[[[![$[%[`[a[b[~[1\\y]Q^a_l_#`q``ab^bcc?deeffggggggggggg00000000`0P0P0R 0R0R0R0R%0R`0R@0R 0R0R0R0R 0R0R 0R0R0R0R 0iR0R0R0R 0iR0R0R0R 0R 0R 0R 0R0R%0R 0R 0R 0R 0R 0R 0R0R%0R 0iR0R%0R%0R0R0R0R0R 0R0R0R 0-R0R%0R 0-R0R%0R 02R0R 0R 0^7R0R 0^7R0R%0R 0^7R00R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R 0^7R0R%0R%0R 0^7R0R%0R 0GR0R%0R 0GR0R 0GR0R`0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R`0R 0^7R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R 0^7R0R0R0R0R0R%0R`0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R0R`0R 0R00000(0&0&0&0&0&0&0&0&0&0&0&0&0&0&0@0@0@0@0@0@0 0000 &\2Xdk>BDV/(; FFKZQQQ RJR`V([^^^^$_b_kk?ACEFGHIJKLMNOPQRSTUWk@  8N ].v.x.64M4Q4gXX::8 @0(  B S  ?g _Hlt514733644 _Hlt514733645 _Ref511634362z.g@@.gpsb b6bCb/f3f4f;fggggZ .q3T4F!GHH IIMPWP__q``baa^bbcTcffgggg33333333333333 ))64R4Y]Z_Z_ZZZ%[%[0\0\\\x]x]P^P^k_l_#`q`b^bdd+f/fffggggggg Mike CarlsonAC:\BackedUp\Docs\NLP\Alignment paper\ACL-2001-Alignment-final.doc Mike CarlsonqC:\Documents and Settings\arulm\Application Data\Microsoft\Word\AutoRecovery save of ACL-2001-Alignment-final.asd Mike CarlsonqC:\Documents and Settings\arulm\Application Data\Microsoft\Word\AutoRecovery save of ACL-2001-Alignment-final.asd Mike CarlsonAC:\BackedUp\Docs\NLP\Alignment paper\ACL-2001-Alignment-final.doc Mike CarlsonqC:\Documents and Settings\arulm\Application Data\Microsoft\Word\AutoRecovery save of ACL-2001-Alignment-final.asd Mike CarlsonAC:\BackedUp\Docs\NLP\Alignment paper\ACL-2001-Alignment-final.doc Mike CarlsonAC:\BackedUp\Docs\NLP\Alignment paper\ACL-2001-Alignment-final.doc Mike CarlsonqC:\Documents and Settings\arulm\Application Data\Microsoft\Word\AutoRecovery save of ACL-2001-Alignment-final.asd Mike CarlsonAC:\BackedUp\Docs\NLP\Alignment paper\ACL-2001-Alignment-final.doc Mike CarlsonqC:\Documents and Settings\arulm\Application Data\Microsoft\Word\AutoRecovery save of ACL-2001-Alignment-final.asd Ӷa,B4DuL* ۔ 9MZM[MeMpM{MMMMMMMMMMMMMMMMN NN$N*N+N$R%R2RTRUR`RgRyRzRRRRR_Z`ZtZ~ZZZZZZZZZZZZZZZZZZZ[[[[![$[%[ggg@QQG QQg@UnknownGz Times New Roman5Symbol3& z Arial3Times?5 z Courier NewG5  hMS Mincho-3 fg;Wingdings"qHhJuU&"UU&{U+S*!20dhb=2Q&C:\BackedUp\Docs\NLP\xxxACL\aclsub.doteA best-first alignment algorithm for automatic extraction of transfer mappings from bilingual corpora Mike Carlson Mike CarlsonCompObj4j