often get answers that agree with what
the linguists believe is true from their non-
quantitative approaches. This then emboldens
us to apply our models to new datasets.
Our latest efforts in this area – and a big part
of the Mother Tongue project – are to develop
a statistical model of how words evolve at
the phonemic level. Phonemes are the basic
sounds of a language (such as vowel and
consonant sounds), and we are modelling how
those phonemes change through time. For
instance, the ancestral Latinate word
has become ‘father’in English and this change
has involved, among others, the replacement
of a ‘p’ sound with an ‘f’ sound. This p–>f
transition is also seen in words like
becoming ‘foot’.
Our macro-evolutionary studies – such as those
described above – attempt to infer and describe
howwords change among sets of languages, and
over centuries or longer. This is especially useful
for identifying general linguistic evolutionary laws.
Our micro-evolutionary studies, however, are
aimed at a different set of phenomena. How, for
example, do the various words for the same
meaning that exist within a single language
Statistically speaking
lthough our project is called ‘Mother Tongue’, our research is
less about attempting to reconstruct or infer what the original
language of humans was (our ‘mother tongue’) than to try to
discover laws of linguistic evolution, and to use those laws to trace the
history of how human languages have evolved. Still, the search for a
‘mother tongue’ is important because to do so one has to understand
the laws of linguistic evolution and then rewind them to get a glimpse of
our distant linguistic past. Understanding these laws of change provides
us with deep insights into how languages lodge in our mind and transmit
information across the generations.
Thus, one outcome of our research could be to find elements of language
that evolve slowly enough that it is plausible they could be retained long
enough to be left over from or echoes of our distant past. We have already
shown, for example, that some elements of language can be retained for
over 10,000 years. We want to understand why it is that some elements
of language can persist for so long, while others change more rapidly.
Opposing schools of thought
Although some linguists might be described as ‘particularists’ who see
each human language as a unique entity in and of itself, it would be fairer
to say that many comparative linguists have for decades recognised that
there are regularities in the ways that languages evolve. Still, many
of these linguists remain uncomfortable with the mathematical and
statistical rigour that we bring to our modelling within the Mother Tongue
project, remaining sceptical that general evolutionary principles can be
applied across languages that might have been separated by thousands
of years of evolution.
It is really a clash of inferential and theoretical styles. The great scientific
advances of the last four centuries or so have come from scientists who
can take a large number of observations and reconcile them in some
way with a few theoretical observations that can be expressed in simple
mathematical equations.
We do this when we use statistical modelling of the words that different
languages use to describe the same meaning (such as ‘water’ versus
‘eau’ versus ‘aqua’), and to infer a family tree of those languages (often
called a phylogeny). Our statistical model makes assumptions about the
way that words change over time to draw conclusions from observations
on living languages about what the past looked like and how it arrived at
its present state.
Some linguists oppose this style of modelling, not wishing to believe that
something as complicated as language could be reduced to a few
mathematical expressions. Yet proof nevertheless exists, because we
Pan European Networks: Science & Technology
Professor Mark Pagel
The project’s macro-
evolutionary studies
are designed to infer
and describe how
words change among
sets of languages, and
over centuries or longer
© karn z
1...,152,153,154,155,156,157,158,159,160,161 163,164,165,166,167,168,169,170,171,172,...300