Made the relationship between words cleaner and frequencies more cohesive. I would look at the naming of WordsCleaner now but I don't want to mess with the code too much and fail to demonstrate the core point - perhaps Tokeniser would be more descriptive though

word_count.rb

class Phrase
  def initialize(words)
    words_cleaner = WordsCleaner.new(words)
    @frequency_counter = Frequencies.new(words_cleaner)
  end

  def word_count
    @frequency_counter.frequencies
  end
end

class WordsCleaner
  def initialize(words)
    @words = words
  end

  def clean
    alpha_numeric_only(@words).downcase.split
  end

  private
  def alpha_numeric_only words
    words.gsub(/[^a-z0-9]/i, ' ')
  end
end

class Frequencies
  def initialize(words_cleaner)
    @words_cleaner = words_cleaner
  end

  def frequencies
    tokens = @words_cleaner.clean
    tokens.group_by {|e| e }.each_with_object({}) do |(word, occurs), result|
      result[word] = occurs.length
    end
  end
end

Yeah, I see the point now. Having what was in lines 8-9 in 33 keeps the data inside the Frequencies class.

I like that it drops the method args needed too.

Thanks!