코드

require 'MeCab'
tmp="" #文字出力に使う
count=0 #動詞出現回数
flag = 0 #サ変の前に文字が入るかどうか
lines = IO.readlines("all_hyouka.txt", chomp: true)

File.open("phrase_data_sahen.rb", "w") do |f|
  lines.each do | line |
    begin
      model = MeCab::Model.new(ARGV.join(" "))
      tagger = model.createTagger()
      n = tagger.parseToNode(line)
      while n do
        feature = n.feature.split(",")
        if tmp!="" && flag == 1
          if feature[0]=="名詞" and feature[1]=="サ変接続" and n.surface!=")" and n.surface!="," and n.surface!="～" and n.surface!="("
            tmp << n.surface
            flag = 0
          else
            tmp = ""
            flag = 0
          end
        elsif tmp!="" && count==0 
          if feature[0]=="動詞" and feature[1]=="自立"
            tmp << n.surface
            count=count+1
          else
            tmp = ""
          end
        elsif tmp!="" && count==1
          case feature[0]
          when "動詞"
            case feature[1]
            when "自立"
              f.print "\n",tmp
              tmp = ""
              count=0
            when "非自立", "接尾"
              tmp << n.surface
            end
          when "助詞","助動詞","フィラー"
            tmp << n.surface
          when "形容詞"
            case feature[1]
            when "自立"
              f.print "\n",tmp
              tmp = ""
              count=0
            when "非自立"
              tmp << n.surface
            end
          when "名詞"
            case feature[1]
            when "サ変接続"
              f.print "\n",tmp
              tmp = n.surface
              count=0
            when "接尾","特殊" #,"非自立"
              tmp << n.surface
            else
              f.print "\n",tmp
              tmp = ""
              count=0
            end
          when "感動詞","接続詞","接頭詞","副詞","記号","連体詞","BOS/EOS"
            f.print "\n",tmp
            tmp = ""
            count=0
          else
            f.print "\n",tmp
            tmp = ""
            count=0
          end
        else
          if (feature[0]=="接頭詞"and feature[1]=="接続名詞")#or(n.feature.split(",")[0]=="副詞" and n.feature.split(",")[1]=="一般")
            tmp = n.surface
            flag = 1
          elsif feature[0]=="名詞" and feature[1]=="サ変接続" and n.surface!=")" and n.surface!="," and n.surface!="～" and n.surface!="("
            tmp = n.surface
          end
        end
        n = n.next
      end
    rescue
      print "RuntimeError: ", $!, "\n";
    end
  end
end

###################以下、語の出現回数計算して大きい順に並べる処理

require "find"

words = Hash.new(0)

lines = IO.readlines("phrase_data_sahen.rb", chomp: true)
lines.each do | line | 
  words[line] += 1
end

File.open("output_sahen.rb","w") do |output|
  output.print "\357\273\277"
  words.sort_by{|word,count| [-count,word]}.each do |word,count|
    output.print word.chomp,"\n"
  end
end

단어의 예를 꺼내다