ED 076 081 Beebe, Ralph D. TITLE The Determination Of The .

DOCUMENT RESUMEFL 004 106ED 076 081AUTHORTITLEIIIINSTITUTIONPUB DATENOTEEDRS PRICEDESCRIPTORSIIDENTIFIERS1I#44.4Beebe, Ralph D.The Determination of the Frequency of SyntacticalPatterns in Present-Day Written AustraliarEnglish.Monash Univ., Clayton, Victoria (Australia).7116p.; In Linguistic Communications, 3, 1971MF - C.65 HC- 3.29Computer Programs; Descriptive Linguistics; *English;Language Patterns; *Language Research; LinguisticTheory; *Research Methodology; *Sentence Structure;Statistical Analysis; Surface Structure; *Syntax;Verbs; Written LanguageAustraliaABSTRACTConfronted with the problem of determining thefrequency of syntactical patterns in present -day, written AustralianEnglish, the author employs a method of analysis which produces anoutput in the form of a two-dimensional line diagram showing all thesyntagms comprising the sentence under analysis. For the remainingproblem -If sorting the diagrams into divisions and sub-divisions ofsyntagms, the author advocates the use of a method of linearizationused for sorting structural diagrams of chemical compounds. Adescription of the methodology is provided along with an explanationof its adaptation to language analysis. OMT.A

FILMED FROM BEST AVAILABLE COPYIn: Linguistic Communications; 3, 1971.THE DETERMINATION OF THE FRECU:NCY OF SYNTACTICAL PATTERNSIN PRESENT-DAY WRITTEN ALETRALIAN ENGLISH.Report uated15th May, 1970.Ralph D. BeebeMonaphUniver: ityIn advisiy the writer en ;his prc2ect, ProfessorU. G. E. Hammarstrahau suggesteu that the frequency ofEnglish syntagms ccd Inc Jt.terni.uucorpus of Englishexamining a3.1tan--.:e.,. dividing then firsjntosentence types, thrn sub -civic ing thn sentence typesfurther, according to his system of syntactic terminology(Hammarstrbm 1967).A manual sorting of sentences inthat way would have been a procese or great magnituue.In searching fo :- a mcre elegant method, the writerfirst aimed at a crmputinrAhich would haveautomatically analysed pontincu,3 intr L!rijr Jyntagms.H. hoped to be aPlr-1he-e-lify the pr6gram1;the sentences and th:.iranalysis programFledL.o sortAlthuugh such andeveloped by 7ratley, Thorneand Dewar (1967), i. proven Lo he incapa'lle of being runon any computer in Aust:.al:.a due to ccmputer-languageincompatibilities.Pn eltr:nvp 'AP(Sager 1967) evolveu at Ww.k Hnive-z jty did notprovide an output in onevateproject.No other progt.amlprogramthr.: purposes of the'jr ntly available.were L2As a manual ana!ydi: sPenra.;.ei-:fore, inevitable,the writer turner, hjs attention to other large -scale manualanalysis work cone rreviou:ly.to ba in studies pc'A fruitful area appearedhe writing of children.Notableexamples were these of Ln(2.1)33), Strickland (1962),Loban (1963), and Hurt (n66).These studies showed a

2.growing tendency towarus a more formaluelineation ofsentence structure, but all in,icated thata more completestudy could not be made until somemore detailed system ofanalysis had been devised.The writer then turned his attentionto using a mathouof analysis which he h-3d himselfdeveloped primarily forteaching purposes.This methoo g,,,Je an output in the formof a two-dimensionalline uiagram showing all the syntagm3comprising the sentence analysea.It was essentially asurface-structure analysis using a form ofuependency grammar.The problem stilluiagrams into divisionsremained, however, of how to sort suchanu pub- divisions of syntagms.The writer had observed thata somewhat similar problemof sorting chemicalcompounds expressed in the fdriil ofmolecular-structure diagrams hau been solvedin variousways throughout the worlu.He selected one way devisedby the U.S. ArmyBiological Laboratories (Wiswesser 1954)anu currently popular with many U.S.urug companies.The selected method firstreouced the two-dimensionaluiagrams of molecularstructure to linear strings of symbols,and then sorted the strings byconventional computermethods.From the principlesemployed by Wiswesser, the writersucceeded in learning how to linearizehis own two-dimensionaldiagrams of sentencestructure, and the remainder of theproject can now be completed by writinga suitable computerprogramfor sorting the linear strings ofsymbols.

13.Further aid may be obtained in this phase of the projectby stuuies of the programsused in organic chemistry andof new languages for the computer such as PL/1 anu SNUBOLdevlseo especially for sorting strings of symbols.Compatibility with the Monash University computer complexwill be an overriding consiueration.A statistical analysis of the results will determinethe required syntagm frequencies, and the syntagms mightthen be allotteu hierarchical distinctions usingHammarstr8m's proposed terminology.By dxamining several different genres of present-uaywritten Australian English, the syntagm frequencies amongthe genres can be compared, thus reuucing the influence oferrors in the syntactical analysis.BR:EF DESCRIPTION OF THE WISWESSER SYSTEMThe methou of linearization used for sorting structuraldiagrams of chemical compounds in the Uniteu States,ueviseu by Wiswesser (1954), and reviseu by Smith (1965),first translates all conventional two-letter atomic symbolsinto single letters, and also proviu,ls single-letteridentification symbols for groups of atoms forming commonlyoccurring radicals.For example the halogens, bromine andchlorine, normally expressed by the symbols Br anu Cl,become E and G, so that the following list emerges:Ebromine atomFfluorin- atomGchlorine atomHhydrogen atom (although H is mostly unexpresseu)ioaine atomI

4.Added to the list are the following symbols forvarious groups:Qhy.roxyi group,Vcarbonyl connective,-OH.41111W-C-(carbon connected to throe other atoms)Unonlinear (branching) uioxo group as in-NO-50 -.Not uses, for linear (unbranched)2'structurefl such as CO2, 5i02, 1{02, 502.HMimino group,Zamino croup -NH2.Numerals are used to show the number ofcarbon atomsin unbranched alkyl chains or segments.Thus the following unbranchea compoundsare expressedin linear notation as shown:0(1)CH -C-CH(2)CH CH -0-CH CH331V13223202(3)HO-CH CH -OHQ2Q(4)0 N-CH -0-CH -NO22W11011 N12(5)222H N-CH CH CH -NH22222Z3ZFor brancheu compounds, a graphicformula is firstinterposed between the structural formulaanu the eventuallinearization, rules being laid down forlinearizingthegraphic formula.In the followingsimplifieu description,these rules are abbreviated to the pointof inadeouacy, butthey serve to demonstrate the basis for theeventual set ofrules devised by the writer for his sentencediagrams.

3.Thus observe the following linearizations:Structural FormulaGraphic F-,-,:ulaUHLi.-ear'-zationQ0 S'./, 0 10-CH3iCH2CH3CH CH -13-",322 B 22B2 &21.1CHCI:23The rules static that the linearization of a graphicformula is performed by citing the symbols along a mainchain until a branching point is reached, digressing alongthe branch, then returning, after ',hc! end of the branch isreached, to the main chain, inserting an extra symbol (8.)before resuming tne symbols .if -:.he noin chain.branch terminates inE-.If theslmliol ,hic'n ,-.,Jr.not be followeu inany case along that branch by nthcr symbols, then it isal-terminating symbol, anu there is no rsed to ir.sert theresumption symbol (&) when continuing along the main chain.In the first exampl,-; above,Li.s a 1Lrminating'symbolknown to be such by an organic chemist, so there is noneed to use the resumption symbol when cont;.nuing alongthe main chain after dealini 'with th,, branch chain.the second example, however,t'.i! 1-r,no',.In-.1)01,; rre not'terminating' symbols, as they can elch be followeo alongtheir branches by other symbols, jiformation which again isknown by the organic chemist who encor.;es the diagram.i

Thus the inherent technical knowledge of the encocierenables him to encode correctly.The Wiswesser system covers not only unbrancheoanu branchau chains, but also cyclic compounas, utilizingin all some 250 rules.In the encoding of sentenceJiagrams, however, only a few of the rules of theWiswesser system are needed.Thes selecteu rules havebeen araEtically simplified in toe brie. description givenabove.Their application to sentence-diagram encodingwill now be uescribed in detail.APPLICATION OF THE WISWES5ER SYSTEM TO SENTENCE DIAGRAMSThe appendix gives some examples of the encodingof sentence diagrams.The four basic types of Englishsentences, distinguisheu by their verb types, are encodedas follows:(1)John shudderedN D(2)John injured JimN D N(3)John was sickN B W(4)They elected JohncaptainR FN NThe D in the graphic formula of sentence (4) above hasbeen omitted from the linearization.This has been donebecause D is an essential element of a factitive preoicatorF and can therefore be assumed to 'le presentwithout beingspecifically mentioned.Its omission is similar to theomission of the hydrogen symbol from the alkylgroup in theWiswesser system.

7.A similar omission or the symbol for the prepositioncan be maue in every predositional phrase since every suchphrase must commence with a preposition.It is onlynecessary to insert the symbol H for the phrase anti gostraight on to consider the other elements apart from thepreposition.The normal element accompanying thepreposition in the phrase is the noun, but that elementcan be replaced by various substitutes such as the pronoun,or non-finite verb.If the noun is present, it can beomitted from the linearization;only the symbol for itssubstitute need be includeu when such a substitute ispresent.On the other hand,eny dependencies of the nounmust be shown, as in sentences (5) and (6).(5)John struck Jim in angerN DH N(6)John struck Jim in greatangerN DHQ NThere can be no ambiguity concerning the Q insentence (6) since an aJ'ective cer-lot be used todescribe a preposition.The Q must to a dependencyof the N in the phrase H.This is an example of the inherent technical knowledgeof the encoder enabling him to encode correctly, a paralleloperation to that of the organic chemist encoding chemicalcompounds §y the Wiswesser system.The advantages of the linearization system become moreevioent when more cofplicated sentences are considered.SeeAppendix,sentences (71 ano (8).It is clear that the sorting of the strings is, comparativelyspeaking, the least problematical part of the project.

6.BIBLIOGRAPHY1.Bratley, P., Dewar, H. end Thorne, J. P.1967Recognition of Syntactic Structure by Computer.Mature, Vol.216,December 9See also:Hamish Dewar, Paul Bratley, and JamesPeter Thorne, 1969A Program for the Syntactic Analysis ofEnglish Sentences, CACM,Vol.12,No.8,Aug.2.Hammarstr8m, G.1967On Linguistic Terminology, Actes du Xe CongrlsInternational des Linguistes, Bucarest, 28 rout -2septembre, Vol.I,po.321-3253.Hunt, Kellog W.1965Grammatical Structures Written at Three Grade Levels,NCTE Research FLport Mo. 3, National Council ofTeachers of English, 508 youth Sixth Street, Champaign,Illinois, 618224.LaBrant, Lou L.1933Study of Certain Language Developments of Childrenin Grades Four to Twnlve in,:lunive, Genetic PsychologyMonographs Vol.XIV,No.5,Nov.5, Loban, Walter1963The Language of Elementary School ChildrenNCTE Research Report No. 1 Nctional Council ofTeachers of English, 508 South Sixth Street, Champaign,Illinois6. Sager, N.1967Syntactic Analysis of Natural Language, AL:vane-minComputers, Vol.8

ea.7. Smith, ElbertG-1968The Wiswesser Line-Formula Chemical NotationMcGraw-Hill Book Company.8. Strickland, Ruth G. 1962The Language of Elementary School Chiluren:Its RL.lationship to the Language of ReadingTextbooks and the Quality of Heouing ofSelected Lhil.iren.Bulletin of the Schoolof Education, Indiana University, Vol.38.No.4,Julyg. iswesser, W. J.1954A Line-Formula Chemical NotationCompany:New YorkThomas Y. Cromwell

9.APPENDIX1.SYMBOL CODE FOR STRUCTURAL DIAGRAMAdjAdjective-AdvAdverbAG-Appositive GroupC-ConjunctionCG-Ccordinate e ExpressionF-FrequencyFV-Finite VerbFac PredFactitive PredicatorM-MannerN-NounNeg-NegationNFV-Non-finite itionS-SubjectSup-SupplementT-Time