Problems with Language

Written to the song: FRENSHIP — Capsize

“Kindness is the language which the deaf can hear and the blind can see”
— Mark Twain

Can computers successfully climb mountains to beat the language problem? Belogradchik, Bulgaria.

Isn’t it the future? Why don’t we have robots that serve us breakfast in bed and computers that can cure our heartbreaks? One major problem, is the language problem. Language can be hard and tricky (this is especially apparent when learning a new language). The following factors may cause linguistic misinterpretations causing problems for computers trying to become more human.

1. Phonemes: a phoneme is the smallest discriminable unit with speech. For example, the words right vs. light, which have different initial phonemes /r/ and /l/. Each phoneme consists of a family of sound called allophones. Another example is /k/ in ‘kit’ and ‘skill’. The pronunciation of /k/ depends on the position in sentence and the speaker’s pronunciation, where some speakers add ‘aspiration’, adding a minor exhale when pronouncing the sound. Computers have difficulty differentiating these phonemes and the proper pronunciation.
2. Tonemes: In some languages, changing the tone when stating the same word, changes the meaning of the word. For example, ‘ma’ stated in different tones in Mandarin provides three different words. Lacking a knowledge of the different tones can cause misunderstanding or worse, the wrong word used in a bad context (example like the one time I said “I peed myself” instead of “I will write it down” by mispronouncing a word in Russian).
3. Chronemes: Phoneme used with different durations, varying by the accent. Take Australian for example, where ‘mate’ may be interpreted as ‘might’.
4. Sound-blend: I am guilty for sometimes speaking too quickly and slurring my words. Blending words to make a different sounding word can sometimes occur when speaking too quickly. The words “new display” may be understood as “nudist play”. Oops.

Communication has a high degree of ambiguity. Often when you are speaking to someone face-to-face it is easier to understand what they are saying through understanding their facial expressions, body language, and through comprehending someone’s typical speech patterns. Can artificial intelligence beat these language problems?